Data Engineering Podcast

Share on social media:

Support Data Engineering Podcast

[00:00:13] Unknown:

Hello, and welcome to the Data Engineering Podcast, the show about modern data management.   When you're ready to build your next pipeline or want to test out the projects you hear about on the show, you'll need somewhere to deploy it. So check out our friends at Linode.   With their new managed database service, you can launch a production ready MySQL,   Postgres, or MongoDB cluster in minutes with automated backups, 40 gigabit connections from your application hosts, and high throughput SSDs.   Go to dataengineeringpodcast.com/linode   today and get a $100 credit to launch a database, create a Kubernetes cluster, or take advantage of all of their other services. And don't forget to thank them for their continued support of this show.  

Data teams are increasingly under pressure to deliver. According to a recent survey by Ascend. Io, 95%   reported being at or overcapacity,   With 72%   of data experts reporting demands on their team going up faster than they can hire, it's no surprise they are increasingly turning to automation.   In fact, while only 3.5%   report having current investments in automation,   85% of data teams plan on investing in automation in the next 12 months. That's where our friends at Ascend dot io come in. The Ascend Data Automation Cloud provides a unified platform for data ingestion, transformation, orchestration, and observability.   Ascend users love its declarative pipelines, powerful SDK, elegant UI, and extensible plug in architecture,   as well as its support for Python, SQL, Scala, and Java. Ascend automates workloads on Snowflake, Databricks, BigQuery, and open source Spark and can be deployed in AWS, Azure, or GCP.  

Go to dataengineeringpodcast.com/ascend   and sign up for a free trial. If you're a data engineering podcast listener, you get credits worth $5, 000 when you become a customer.   Your host is Tobias Macy. And today, I'm interviewing Tommy Ioannolis about using data to improve efficiencies in multi location service businesses with Ops Analytica. So, Tommy, can you start by introducing yourself?  

[00:02:04] Unknown:

Thank you, Tobias, and thank you for having me on today. My name is Tommy Anolis. I am 1 of the cofounders   of Ops Analytica, which stands for Operations Analytics.   And like many of the founders that you have on your show, I have a hotel restaurant and stand up comedy background.   Actually,   I got into tech as part of an operations job in the restaurant industry where   my bosses came to me and were like, hey, man. We need to start reporting on all these audits we're doing out in the field. And an audit, for those of you who might not know, is when, you know, either a third party or a representative of a company   will go out and just basically inspect   a location to make sure that it's clean and it's operating the way it's supposed to, kind of a brand protection thing. And they do those usually quarterly, sometimes monthly   if a place of a location's having some issues.  

This is back in 2008. Right? Their idea was is that me and my assistant would just hand enter, like, you know,   5, 000 audits a month into an Excel so we could report on it. And, like, I was at Quiznos at the time. And for those of you who don't know, Quiznos was, like, the largest 1 of the largest sub chains in the country, like, in 2008, 2009.   It's also got the infamous reputation of being 1 of the largest   franchise system implosions   ever, meaning that basically all those stores went out of business. And they went from over 5, 000 locations to about 350.   At the time, we had about 5, 000. And we were auditing constantly because we were closing stores because people weren't making money. And so they wanted to get the field guides in the stores   as often as possible to see that they were clean and operating well, but also to kinda gauge   their business health   so they could kinda predict who might be closing next, if that makes sense. And so that's kinda how I got into tech is I tried to buy a solution, couldn't get 1, went to IT, and IT did what they do best. No offense to bias. They said, no.  

You can't have this. We don't have the resources.   And I'm like, what else if I build up myself? And they were like, no. Not secure.   IT is like I think when you get an IT degree, you have a whole class on different ways to say no. Here's the 50 reasons why you can say no to this. And so, anyways, finally, my boss is like, I don't care what they say. Like, what is this gonna cost me? And it was really cheap because I ended up using a platform called Form Router. Which I don't know if they're still around, but it was kind of an early   it was kind of an early SaaS business where you could subscribe.  

You know, at that time frame, think about 2008, 2009,   build your own form, and then download the data   from their servers onto your own database.   And so when he found out it was gonna be, like, $25100,   he's like, I'll just put it on my corporate card. Just build this thing because we're getting yelled at by, like, the CEO and the COO. So, like, who cares what the IT says? And that's how I kinda got my foray into tech. And now I have a company called Apps Analytica, which   is basically we help multi location businesses   and large high volume   businesses   manage their teams   and measure what their teams are doing   from sort of an operations   perspective. So   making sure that we're controlling what we can control, making sure we're safe, we're clean, we're ready for guests,   you know, our customers or patients or whatever. Right?  



[00:05:33] Unknown:

In terms of the types of information that you're working with at Ops Analytica,   I'm interested in understanding kind of the sources of data that you're dealing with, the form that that data takes. Like, are you dealing with,   you know, form entries where people at those different locations are prompted with, you know, some input and they just have to fill out a form? Is it sensor based   IoT data? Are you dealing with, you know, imagery? Just curious kind of the variety and scale of information that you're working with.  

[00:06:02] Unknown:

We have sort of 3   sources of data right now, and we hope to expand that into the future. Right? First off, what we're doing because we're managing the team. Right?   So what I care about like, so you think about a restaurant or a retail or doctor's office, whatever it might be, like, they've got a system. Like, oh, let's just do a restaurant to keep it simple. Everybody eats at restaurants.   We're all familiar with them, so   it's an easy use case. At a restaurant, they've got a POS system, and the POS system is how they manage the transactional   sale thing. And it's also how they track sales mix of products and whatnot like that. And it provides a workflow   for capturing sales information,   transacting the transaction, whether it be credit card or money, and then telling the people on the back, hey. Go make this food. Right? So that kind of stuff's handled. I'm more focused   on, you know, was the restaurant clean?  

Did we taste the food? Did we stock the line correctly?   Did we make sure the food was safe? I focus more on those   human activities   that have typically been in this gigantic black hole where nobody could see them because you couldn't track what humans were doing. Right? And so a lot of the history of these big multilocation chains   has been backing into what's happening operationally   by looking at some other indicators like sales and profits and costs and all that. So the primary   I'm getting to your actual answer. The primary answer is, I would say, 90% of our data today comes in from forms   filled out on a schedule checklists   on a schedule by employees. Right?  

And we put them on these, like, schedule windows so that they get done when they're operationally significant for the business. Within those forms, we can capture   truth bulls. I mean, you have a more technical audience. So we were getting bulls. We're getting decimals.   We can get free text, but free text is tough, obviously, because,   like, for instance, we have free text where you can just write in an answer, or we have, like, what we call multiple choice text where they have to select a button,   but then the button has a value, which is text, but at least it's, like, formatted.   Like, it's always the same   options, if you will, which makes that a lot easier for reporting. Right? Because reporting on free text is garbage. Just impossible   to do it well, at least. And then we can capture dates, times,   and then we also capture a lot of photos too. But we don't do any analysis on the photos today. We really are just capturing the photo for human analysis and validation of what's happening.  

We also have we can monitor source on   decimals specifically.   So we do integrate with digital thermometers.   So the idea being that you can have a Bluetooth thermometer, and as you walk up to temp something, you simply tap into the the number box.   Your digital thermometer's connected. You see a live temperature feed coming off the thermometer,   and then you insert the thermometer into the food, let it register, and then you push a button, and that button then inserts the temperature into our checklist. So we do track that source data,   where it's coming from. And then 1 other thing is you mentioned IoT sensors. So we are   you'll be the first that I'll be mentioning on the show. We should be releasing that IoT next week. We have   partnered with Cooper Atkins, which is a large thermometer and IoT company,   and we will be displaying their IoT sensor data in our platform.  

And there's so much I wanna do with sensors. It's not even funny.   And we can get into that because we implemented   the sensor things API.   So we were starting with the SkuBracket integration, but we implemented the sensor things API standard   so that we can, in the future, take on more sensor data and have a way of getting   structured data from other entities into the system so we can, like, analyze it and display it for our customers.   So   most of it's coming in through the forms. We do control the source of the forms   on certain questions, like the thermometer questions, and then we will have sensor data coming in as well.  



[00:10:19] Unknown:

And so in terms of the types of questions and insights that business owners are looking to be able to get answers to and understanding of for their operations,   you know, what are the   different pieces of information that they're trying to extract from this raw data that you're collecting, whether that's at the   CEO level of a large franchise   organization or the individual   location managers who need to understand how their employees are performing against the corporate standards and the types of questions and answers that you're looking to provide?  

[00:10:52] Unknown:

You know, that's a great question. And it's interesting because this software, this type of software,   I'm not gonna say the technology.   I wouldn't say   this is just all new for these businesses. Right? Because   what what happened is if you think about it, like, if you go back to that Ray Kroc movie, that McDonald's movie, you know, McDonald's was really   the launch of the big, large, multi location chain. I mean, you had Howard Johnson's, you had hot shops out of Marriott,   but McDonald's was truly the first, like, super systems integrated,   like, restaurant chain. Right? But they all started in the fifties sixties. And so,   you know, the technology   just didn't exist   to do what we do today when it comes to capturing this data. Right?  

So, historically, what happened was the restaurant industry and and a lot of these big multilocation industries that kind of ran on their heels, national retailers and everything else, hotel   chains, they never had access to, like, getting real time operations data. And so they morphed their businesses   to be a more training focused,   systems focused, but paper systems.   And they just focused on training because they knew they could control that, and they focused on paper systems.   But the problem obviously with paper systems is there's no visibility. There's no accountability.   All the data just gets lost on the paper. No 1 does them. I mean, the big joke in the restaurant industry is that, you know, all the checklists are pencil whipped when they're on paper, that type of thing. That's been prevalent   all the way to today. Like, my biggest competitor is still paper.  

So when we talk about the checklist that we're creating,   this is still the 1st generation of people that are digitizing   checklists and procedures   and moving them off of paper   into digital. Right? Whereas, like, if you think about the POS system, as an example, the register, that started in the late seventies, early eighties. That's been through multiple iterations, and so now you're getting more like, Toast is a big,   Boston company. Right? Like, you're getting more and more,   like, you're on the 3rd or 4th or 5th generation of this POS software, and it's just getting better and smarter and more intuitive.   We're still just taking a lot of times these guys' paper checklists   and moving them to digital. And for them, that's a gigantic   leap forward.  

Right? Because now they have visibility and accountability   that they never had before,   but they're not at a point where they're designing checklist for a digital world.   They're still sort of just, hey. Let's take these what we call red book processes.   The red book for your listeners was literally a red book that had all your checklist   built for every month. So you would buy them a quarter in advance, and they would mail you these 3 red binders. And it would be like, here are your made checklists. And there'd be 1 for every day. You know? So we call them red book processes. We're taking those red book processes and moving them into our platform.   So that's just a little bit of a history there. The kind of questions we're getting are from a food safety perspective. We're temping a lot of food. Right? So temping sauce, temping ranch dressing, temping meat, temping coolers.  

In restaurants,   poor temperature control is the number 1 cause of foodborne illness. Right?   Keeping cold things cold, hot things hot. Basically, you don't want anything between 41 and, like, 135,   140.   That's considered the temperature danger zone. So if you ever get served food in a restaurant and it's lukewarm,   that's bad   Because that means that that food's like a warm moist petri dish that if there was foodborne illness, germs, bacteria in there, they could be growing   at a really fast rate in that temperature danger zone. You kinda want it above 135 is where you start killing things. And then depending on the certain bacterias, it goes up to 165   is where you need to get the food temperature. So lukewarm food is just generally bad   across the board. Right? We're getting a lot of operational questions too. You know, just as important as food safety is cleanliness,   appearance of the restaurant.  

Those are also big things. Appearance of bathrooms, cleanliness of bathrooms. Those are huge. So we're tracking a lot of that. We're also tracking a lot of what I would call prep or stock   in the respect of not inventory,   but, like, you know, if you're thinking about going through a Starbucks drive through or a McDonald's drive through or something for breakfast or Dunkin' because of the part of the country you guys are in,   like, you know, when you see that drive through line, like, how many times have you been driving? I could just use an Egg McMuffin and a hash brown and a coffee. And as you're driving, you come up on your McDonald's, and then you look at the drive through line, and it's, like,   longer than normal. Like, maybe there's 4 or 5 extra cars in there. And mentally, as you're driving, you calculate, I don't have time for that. Like, if it had been in a normal spot for that line, then I would go in. But because the line's way back, I just gotta keep going, and I go with that type of thing.  

That's generally tied to speed of service, and that's generally tied to either staffing or stocking. Right? If you're in the middle of cooking egg McMuffins and there's no eggs on the line and you have to run to the walk in to go get eggs, that slows everybody down. You know what I mean? A couple of minutes.   So it's those kinds of things, making sure that they are properly prepped for the shift.   So those are the kind of operational things, like, on a daily basis   that we're capturing at these stores.   Now then you expand out to audits. When you do an audit, once again, that's an inspection of the whole store. So they're looking at menu boards. They're looking at the the shape of the seats and the boost, the cleanliness of the restaurant, the floors.  

They're doing the entire scope of the building.   So that would be an audit. But then, you know, what's interesting about our platform is because we're 1 of the most advanced platforms in our space in in this area,   we get people doing a lot of other stuff. It just truly becomes an operations management tool. So, yeah, we're doing those Red Book processes, those daily things. We're also doing their,   work, putting in processes in place for their cash outs and their deposits   and labor laws and what we call LTOs or limited time offers, marketing events, where you're gonna   see a lot of ads running, like, in the world. They need to check that people have, like, learned how to make that food, what their signage is. You know, you wanna see table tents and window claims and menu boards are correct and register programming's correct. So we kinda run the gamut of everything. Also,   a lot of compliance stuff too. Like, you have 14 year olds working. Do they have work permits?  

You know? In certain markets,   schedules have to be posted 2 weeks in advance. And if you change them, you have to pay a fine and pay the employee and, like, Seattle and some of the more liberal cities have done these very   strict scheduling rules.   And so they have to make sure the companies have to make that they're not violating those or they'll get fined. You know what I mean? So we kinda just run the gamut of daily operations.  

[00:17:59] Unknown:

To that point of what you're saying about you being 1 of the more advanced options in the space, and you also mentioned platforms such as Toast that are the point of sale systems that a lot of these franchises have   oriented around, and I'm sure that that provides some information to the site managers and the business at an aggregate level. Wondering if you can just   give a bit of a sense of what the kind of competitive landscape looks like in this area of working with service businesses,   particularly if you're dealing with multiple locations even if it's just 2, and just some of the types of differentiating features that the   companies in this space are looking to provide?  



[00:18:37] Unknown:

I think we're all   startups for the most part. Right? Like, there isn't like because this thing this really only began   late 2000.   There was 1 company called RISE Point. They were originally called Steven, but they they got bought by private equity. Now they're called RISE Point.   And they started off doing audits. Right?   So this is actually this will be kinda interesting. So everyone started off as an audit platform because that was sort of the main thing that peep restaurants and these chains wanted to do was make sure that their field guys could go in and audit their stores   to make sure, once again, from a brand protection standpoint,   that the stores were operating link clean and that they   weren't selling random stuff like you weren't like a Subway selling boba tea and, you know, weird things like that. They just make sure everything was cool, copacetic,   you were following the rules, and that you were safe.  

And audits   are really great at identifying   large system wide issues.   Like, hey. Everybody's struggling with, you know, question x.   That's what audits are really good at doing. And then you go, okay. Well, now we know everyone's struggling with question x. It's probably not that they're actually struggling with it. Maybe our procedure's bad. Our equipment's bad. Our training's bad. Let's try to uncover why they're all struggling with this, and then we can fix it and move on. And that's what audits are really good for. But this is a quote from 1 of the head of, like, StereTech, which is a big third party auditing firm. She's like, we're not gonna audit anybody to greatness. Like, all we can do is just come in   and just, you know, tell you what's kinda happening. And their approach is to try to come in and coach. Right? Hey. This is what you're doing wrong, and this is why it's not safe. Let us help you fix it type of thing. And so that started, like, late 2000   into the 20 tens. And, obviously, the iPhone was a game changer. Right? Because it took us off those palm   Windows smartphones,   which my little first platform was made to run on just basically a web form. And it gave us, like, this more dynamic environment to build   web apps and mobile apps for. Like, then, obviously, the iPad came out whatever after 2010 or right after that, sometime in that area. And that's really what opened the door   technology wise for platforms like ours. And it's funny because you think back like, I'm 50 now. But, like, you think back, and you can't even remember a time where we didn't have smartphones or tablets.  

But, literally, it was, like, 15 years ago we didn't. You know what I mean? Like, it's not even that long.   But those enabled us to start doing more stuff. And so when we started our company,   I was like, yeah. Audits are fine, but you need to be able to, like you need to be able to change behavior at the location level. So I always came in focused on daily checklists.   And the big difference between dailies and audits   is that audits are sort of a singular   hard coded. You get 1 audit. Right? And everybody knows, especially your listeners know, doing 1 of anything is easier than doing 2. Right?   And so everything's cleaner on an auditing platform because you're gonna run 1 audit, and you're gonna have all your reportings as singular   reporting based off of this 1 audit. So the big auditing platforms, they could do a different audit for each client, but they were only doing 1   audit   per client. Does that make sense? So then all the reporting was singular in nature. How is this audit doing over time? And blah blah blah blah blah. And when we came in on day 1, we focused on daily checklists.  

And we knew we were gonna go dailies because that's a, you can make more money off dailies, but b, you're gonna have a better impact on the business. So we coded our platform always to be able to handle   more than 1 checklist or audit or whatever. Right? But that also adds clicks and a little bit more complexity because,   you know, you're not just dealing with 1. You're dealing with 2 or more   across the board.   So the big competitive landscape in our space is you have the audit platforms, and then you have the daily checklist platforms. And a lot of the audit platforms added the capability to do dailies. Right? But what I what I have found with those guys is that they struggle   because so much of the stuff was hard coded in because they started with that sort of singular audit. They've had to, like, really kind of retrofit   and and claw into their code base to try to get something   to work with multiple. Right?  

Multiple versions of a checklist, multiple checklists across the board.   So   in our space, we're competing against audit platforms and daily checklist platforms.   And then, you know, I mean, in reality,   if you were to look at the competitive landscape, I mean, you can do this for free. You can do this with Google Sheets   or a Google Form or even a free SurveyMonkey if you wanted to,   all the way up. So that'd be 1 end of the spectrum.   You know, doing it like an online Excel if you wanted to.   You can totally do it that way, but it's really brutal. Right? Because, you know, if you've ever tried to use a spreadsheet on a phone, you gotta click into the cell, and then, you know, everything changes, and it's hard to manipulate, and it's hard to move around, and you can't capture structured data. So that would be like the easiest,   like, would be the cheapest, but the hardest to use. Then you can kinda graduate into sort of like   a Google forms or a SurveyMonkey   world. Those are good   because you can get structured data out of them. You can have data types. You can even have some follow-up questions, a little bit of logic in there. But they don't do a great job   in the space because   they're all on demand.  

Right?   And if you're running a busy business,   you're gonna forget to do things at the right times. And so just having this sort of on demand world is okay for some stuff. But when you really need to make sure your food is safe before lunch is served, you need to get it on a schedule.   And that's where those guys kinda fall down.   And then I would say you have the bulk of our competitors are there in the middle, which they've got a platform   that has scheduling,   that has accountability, that's measuring not only the answers to the questions, but is also measuring how well the teams are doing at getting the questions done.   And that would be the bulk of our competitors are in that space.  

And then we're kind of on the far end of the spectrum because we have all of that stuff, but we also have this rule builder.   And we created our own scripting language within our platform   that gives us this rule builder as our sort of secret sauce. And it allows us to create   features   that we can market and sell off of and more importantly, win deals off of   that are all based on just creating interesting   logical, what we call rules in the system that allow us to make dynamic checklists and   all kinds of things like that. So that's kind of the spectrum of competitors   in there. And what we find is is that we're not always the first guy that gets into   a client,   but we are   a lot of advanced clients' second platform   that they graduate to when they realize, hey. The first platform just didn't give me enough control   to do what I needed to do. Right? I would almost equate it to, like, think of an Excel   spreadsheet. I would say the bulk of our competitors are like an Excel spreadsheet that doesn't have functions.  

They have all the buttons at the top, and they can do a lot of stuff, but they can't just write custom functions to do whatever they want.   And we would be more like an Excel with functions.   So we can actually write those custom logical statements that allow you to basically conform   the platform to your business, which is easier  

[00:26:22] Unknown:

than conforming your business and all your employees to do what the platform needs you to do. An interesting thing that's coming out as you're describing all of these challenges and the way that your platform works and the type of work your customers are doing is that you're very obviously very experienced in these problem domains. And I'm wondering what you see as the   kind of requirement and the competitive advantage of having domain expertise   in the types of businesses that your clients are running, so food service, hospitality,   and just the level of importance of understanding those businesses at a deep level   in order to be able to build a product that actually solves problems for them.  



[00:27:02] Unknown:

Our platform is vertical agnostic, meaning it is more like an Excel in the respect of I can go to oil and gas, and I could go to anywhere I want. And the stuff would function properly because there's nothing that's hard coded about it. It's all custom.   Not like custom development. It's just a platform like an Excel where, you know, an Excel can be a shopping list, or it can be the spreadsheet that you use to run   a multinational   corporation and do financial analysis on. Like, it has that power to do that. But I would say the domain   expertise comes in more on the marketing and sales side of things.  

Like, we're a bootstrap startup. We haven't raised any money. We have no investors.   But, you know, like, we came in, and the thing was we were a bootstrap startup, so we had to choose a vertical to start in because you can't be all things to all people. And restaurants was the natural 1 for us to start in because of my experience at Quiznos running that auditing program.   And then my other business partner, Eric,   he was a restaurant he was a waiter and a bartender. Even our head developer ran the McDonald's drive through in his town in Minnesota,   you know, back in the day. So, like, in high school before he was a developer. So we all spoke restaurant.  

And that   is so important when you're talking to clients and you're creating content, blog content and   podcast content, that you can actually   speak the vernacular   of the customer   and understand what problems they're trying to solve that are unique to their business. You know what I mean?   Because the system is more like an Excel,   you know, you can put any data in there and and do any equations and that type of stuff, and it it doesn't matter. But definitely from a marketing and sales perspective,   this is massively important   to to having real conversations and being able to come up with useful   real world use cases that you can help people solve stuff on.  



[00:28:58] Unknown:

Another interesting angle is that you're talking about the role of periodic audits in terms of making sure that the business can   identify and address systematic problems   that are kind of the macro scale   and that the work that you're doing and some of your competitors can help drive some of the more microscale   efficiencies   of making sure that the food is the right temperature, making sure that the line has been prepped properly so that you don't have these interruptions in terms of the food service production.   I'm wondering what are some of the other types of   inefficiency that these businesses are contending with and some of the ways that those inefficiencies   can compound across multiple locations as you scale the business   and the ways that they're using the information and the insights that you're able to provide to be able to feedback into their   on location   processes and policies   to kind of optimize those cycles?  



[00:29:55] Unknown:

Great example of this. So, like, the cool thing about our platform is is that,   you know, the day you implement our platform, the first day,   we start   collecting data that's usable. Right? And our platform helps the business   at the location level, at what we would call the district or above store level, the VP level, all the way up to corporate.   So   because we're now   collecting data and shedding visible light or visibility onto how the business is actually operating. So at the store level, we are focused on   controlling what you can control   and doing what you're supposed to do every day to make sure you're safe, you're clean, and you're ready for guests, right? So at the store level, it's all about taking the guesswork out of running the business. And when you look at the current labor market, the amount of turnover,   the great, especially in food service, the great restaurant exodus that happened during COVID, where so many of the professionals in our industry,   and I mean professionals, I mean managers and servers and chefs in every level, just said, I'm done. I can't deal with this anymore. And they went to other businesses. We've lost probably over a 1000000 experienced people, if not more, in the hospitality industry. Just people bolted because they couldn't deal anymore.  

You know? So at the restaurant level, it's all about daily operations and taking the guesswork out of the business and making sure we can control what we can control. But once you move up 1 level so, like, in a Starbucks or a typical   corporate owned chain, you're gonna have 1 district manager for something around 10 ish, give or take a couple on either side of that locations.   And that guy's job is to make sure that those 10 stores are operating the best that they can operate   and that they're supporting those general managers and their teams   to deliver great service and do good things. Right? And that's in a corporate   franchise. And a like, let's say,   in our corporate owned business where they own the stores. And in Dunkin', that district manager is gonna be working with about 70 to a 100 restaurants, and they're gonna be primarily working with the franchisees that make them better at operating their stores. Right?  

But at that level, you now give somebody the ability   to see what's happening in all their locations operationally in real time.   And so they can now address problems at 1 location while they're sitting in another, and they don't have to physically be there to coach and   direct. So it extends their effectiveness   at that district manager level to be able to manage multiple locations   simultaneously better.   And then that data just keeps cascading   up the chain   to corporate where now they have a viewpoint   of the entire system.   And when we started a big national burrito chain, and they have, like, 750 locations,   we interviewed them, like, a couple of months after they started. And we interviewed our guy, the guy that was managing the program, and it's usually an operation services role, which is the technology   wing of operations that   purchases and implements products for the ops teams in the field.  

He gave us this great quote. He said, you help us identify problems we didn't even know we had, and you keep us from making multimillion dollar mistakes, which I mean, as a marketing quote, is a pretty damn good marketing quote.   But there was a specific instance of this. Right? Kind of elaborate on it. But, basically, they were getting reporting   that their restaurants were dirty.   And they were like, oh, no. And with that level of information,   right, hey. Your restaurants are dirty.   They were, like, thinking to themselves, well, what we'll do is we'll add a couple more labor hours   per lunch and per dinner during the rushes   to offset.  

So we have someone out in the front of the house just cleaning the restaurant. Make sure the restaurant's clean. Clean is 1 of the top 3 pillars   of success of any multi location businesses. It's actually the top 1 being clean. So they were like, with the amount of information they had, reports coming in and the restaurants were dirty,   that was their initial thought process. And to add those 2 to 4 hours a day in those locations,   you know, if you just did the simple math at $15   an hour, you're basically over $5, 000, 000 a year in new labor costs just for their 350   restaurants,   plus not even including all the training and COBRA   and employment, blah blah blah cost. It could be maybe even,   you know, 20 or 30% higher than that. So that was their initial thought. We're gonna do that. But then they utilized the platform   to crowdsource   information from the restaurants, to get their people in the restaurants, and really start identifying   what was actually going on. So they started asking their teams, because that's a big part of what we do is we require comments and photos when things are wrong. So we can crowdsource   data from the line employee   and get that data up to corporate   in an easy, effective way where they can just look at photos from across the chain or look at all the comments that are coming in, and they can see what's actually happening. And they can obviously change their questions or add new questions   to dig deeper   when something's been identified. And that's exactly what they did.  

They identified that this dirty restaurant assumption was just a high level   assumption. And when they got into the nitty gritty, they figured out their trash cans were too small,   which was a $20 fix   at certain locations versus, you know, hiring another guy for 4 hours a week type of thing. And so that's where that quote came from. You kept us from making a multimillion dollar mistake because they were literally thinking,   okay. We've gotta pull the trigger on these extra employees.   And they were like because that was the level of understanding they had of their issue. And at that level of understanding, that was a good decision to make. We got dirty restaurants. Let's get some people in here to clean them. It's, like, logical. You know? But when they were able to dig deeper and crowdsource data and get questions answered and drill in and start looking at photos, they realized very quickly, this is a trash can issue. That is a much simpler fix. And they were able to implement those bigger trash cans in certain locations, and they solved their problem,   But the platform   had facilitated that,   right, in a quick and easy way.  

I think that answers the question, but I think that's how it's being used at all levels of the business.   You know, the employee   now has a say to corporate on what's really happening.   Right? And they can make an impact in the business.   And I'm gonna keep going here, but this is a kind of a theory of mine. But I feel like operations   technology, like our platform, or operations management, like our platform, is going to be 1 of the next big   technology   revolutions   in multilocation   businesses. These aren't just restaurants. These are retail.   This is automotive. This is dental, medical, you name it. Like, there's all these giant multilocation chains   and that have never had access   to really control operationally what's happening.  

But more importantly than the control part   is they've never had   real time access to operations data where they can identify   issues.   And then having a platform like ours where you can identify an issue,   crowdsource a response,   like, hey, what is really happening? Is there an easy way to fix this? And then implement the change, meaning they just change the checklist or the procedure   right in the platform, a complete feedback loop, if you will, within the platform to identify an issue, to crowdsource a a solution, and then have a place to implement a change   all within 1 system.  

I do believe that those businesses   that are adopting this type of technology   in their day to day operations,   it's going to be the next big competitive   lift.   And what we're finding is is that it's generally not 1 issue.   Like, when you have a bad experience at any of these businesses, it's not because   generally some 1 giant horrible thing happened to you. It's actually the opposite.   It's this sort of death by a 1, 000 cut type of deal where   a bunch of little things sucked and you were like, ah,   this blows. Like, you go in and the line is long, the floor is sticky, your table's sticky, there weren't any straws.   You know, it's like the bathroom was kinda dirty.  

Any 1 of those things would not be enough to throw your experience off the deep end, but it was the it was the sum of the parts,   all these little things that just kind of sucked because they didn't do a good job of managing their business   that added up for you to go have a blog experience. Right?   And so right now, you have these companies that are like our clients and our competitors' clients   who are getting this data and they're identifying their problems and they're fixing them faster   than their competitors who don't have solutions like ours.   And as they get better at doing that, they're going to start incrementally   growing.  

And I think what's gonna end up happening is is there's gonna be a bunch of people that were neck and neck competing against each other today. And in 2 or 3 years, they're gonna look at their competitor and be like, how is this guy doing so much better than us? And you're not gonna be able to attribute it once again to 1 thing. Oh, they cut lettuce prices.   It's because they're addressing these death by a 1000 cut things that they can control better,   faster than their competitor. And then once you start making customers happier, repeat sales,   new sales,   more marketing dollars,   better money more money to go get better positions in the strip malls, get into the best   strip malls.  

Your success is just like that snowball rolling down a mountain,   and it legitimately,   like, just gets bigger and bigger and bigger.   And, you know, that Avis to your Hertz is just gonna be sitting there going, what the heck's happening? You know what I mean? How are these guys blowing us out of the water? How did they open a 100 stores last year or a 1000 stores? You know?   So I think this is gonna be a whole revolution   for these businesses   because it's a huge gap   that, like, you know, we've got visibility into everything else, manufacturing,   supply chain,   sales.  

We just don't have visibility in operations. But once we start getting that,   the guys that get it are gonna start crushing it.  

[00:40:17] Unknown:

RudderStack helps you build a customer data platform on your warehouse or data lake. Instead of trapping data in a black box, they enable you to easily collect customer data from the entire stack and build an identity graph on your warehouse, giving you full visibility and control.   Their SDKs make event streaming from any app or website easy, and their state of the art reverse ETL pipelines enable you to send enriched data to any cloud tool.   Sign up for free or just get the free t shirt for being a listener of the data engineering podcast at dataengineeringpodcast.com/rudder.   Digging now into   your platform itself, can you give a bit of insight into the technical implementation   and the way that you think about the   design and requirements and the overall evolution   of the kind of platform design and the goals that you're trying to achieve with it? So, you know, it's interesting. When you build software from scratch  

[00:41:13] Unknown:

you know what I mean?   Like, it would be great if this all operated in a vacuum, if you will. And you could just sit down on day 1 and sketch it all out and just build it until it was done and then release it into the world. Right? But that's not the reality of the world we live in. I often think about software development as being a lot like capitalism   as a system in whole, in that   you're constantly   you're being pulled by the demands of the marketplace, meaning your customers are telling you you don't have a feature or some functionality that they think is important. You're being pushed by what your competitors are doing. Right? Like, they are adding features and functionality that other people like. And so you're like, ah, crap. I gotta go do that too. And so you end up building it and kind of like   fits and starts. So you go, okay. Well, we have our core product,   and we can do checklist now. Okay. Good. And then you go, well, we don't have enough reports. So then you're like, okay. Well, let's go build some reports real quick. So then you build version 1 of your reports, and then you get those out there. So now you can say, hey. We've got reports. And then you go, oh, well, we need tasks now, which are different than checklists. A checklist is basically a list of individual tasks, but they're easier to kinda complete. Right? So now we need tasks. So then you go build version 1 of tasks. So it's like you constantly are building out all these MVPs of your features,   you know, across the board, better administration,   better this, better messaging, better notifications.  

And you end up, as you're creating this product, you have, like, 20 different, like, feature points that are on MVP level. You literally have to stop developing what you're doing because you realize you're starting to lose deals because you don't have this thing,   and that's the capitalist part of it all. Right? And so you go,   we keep losing these deals because we don't have this. So now we have to stop whatever we were working on and go add a version of this so we can at least say we have enough   to, you know, not get blown out of the water   in the preprocess,   and we can at least get to a pilot or further into the deals flow. And so we did that for, like,   5 years. Right? We just kept adding feature sets,   and everything worked. Right? It was always important that it worked because if it doesn't work, then you have other issues, which is massive customer support tickets, and you have a lot of churn. So we always built things that worked and got them out.  

And thank God, like, my developer,   my head of development,   he's actually a partner in the company.   And I think he was already a great developer. Just anybody who's, like, looking to start a company   and you're looking to start it from scratch, and you're looking to do it   bootstrapped, where you're not just gonna be able to go out and just throw down big salaries on people to get the best people in. You're actually,   like, sweat equity bootstrapping this thing.   You should make your head of development   a partner.   They don't have to have a 1000000 percent of the company, and they don't you know, whatever. But you should make him a partner because our developer from day 1   wasn't an employee, he was a partner.  

And so when, you know, you hit the million decisions that a developer has to make every single day when they're coding,   he   always leaned not like,   an employee goes,   okay. This is what the spec says,   and   can I hard code this, or can I make it extensible and make it so that it can work in multiple different things? And then they have a boss telling them, hey. We gotta get this code done, And they don't have any say in the company, and they might not even be there in a year. And so they choose the easiest path for them to get the code out as quick as possible. But generally, that's more of a hard coded,   less extensible,   less flexible   thing unless that was completely specced in. And developers make that decision a 100 times a day.  

And I think from our perspective,   our developer always thought like a partner, always thought about the future,   always pushed back us, and said, hey. We're gonna do it right the first time, and we're not gonna revisit it. So, yes, it's gonna take a little bit longer, which is miserable for me because I'm not a patient person. And I'm dealing in the real world trying to get code out so I can win some deals so I could get a paycheck. You know? And I was literally fighting to get a paycheck. You know?   But he did it that way, and it's been the best for us forever.  

So, you know, just as a side note for those of you who are at MIT or who are listening to podcasts that are gonna go bootstrap this thing,   highly think about giving your guys some equity. And they can earn into it. And I'm happy to tell you how we structured ours because I got it off a buddy of mine, and and it worked out really good for us. But, anyways, him being a partner,   that led to better development   overall. But it didn't stop us from having all these feature sets that were kind of MVP. You know? They worked. They were good. They got the job done. We could market and sell off of them, and they were they added value to the the platform, but they're all at, I would say, different phases of doneness, if you will. Right? Like, some of them are 80% done. Some are 30%   done. And   so probably for the 1st 5 or so years, it was just getting all of the different   modules that you kinda needed to be able to compete in the world and say you could do stuff. And then about 2 years ago, once we kinda hit that   thing, that level of doneness,   we made an agreement that we would go through before we started adding a bunch of new stuff, we would go through and try to get all those MVP things to, like, a level of 100%   done. You know, I'm not saying that we will never add anything to them in the future, but they were still, like, you know, like, in our rule builder, there's 2 parts to our rule builder, and our rule builder probably sit it's still better than everybody else's, but it's probably at 40% of what we envision it being done as. You know what I mean? And there's another 60% to do. And we've been working on reports and dashboards   for   probably, like,   18 of the last 24 months. And and it's not been 24 of the last 24 because we sold a couple of deals where we had add some functionality, and so we had to pump that in to that 18 months   in there. But you know? So we are just simply going through and looking at all the modules and going, how can we get this to a point   where not only is it, like, a thing that works, but it's, like, the best version of itself so that the whole platform   is, like,   just amaze balls across the board. Right?  

So that's kinda how we've been focusing on it   from a development perspective. And, you know, it's a constant battle. Like I said, it's a capitalism. Like, you know, if somebody starts making a grumble   over here, there's a need or a want, and it's going to affect sales or customer satisfaction,   then, you know, sometimes we have to stop what we're doing and prioritize this other thing and fill this gap.   You know, that's what we have to do. And so if you have to be flexible   in this world, but then it's also frustrating because, like, every time we do, like, that quick fix for a customer,   I know there's something that I really want that's getting kicked back a couple of months. You know what I mean? And it's tough because there's some things in there that I know we're gonna have in the future that I'm like, I cannot wait because they're gonna just propel our competitive advantage   so much further, you know, and what our core strengths are. So it's just like it's it's really tough. And if you're not a patient person, which I am not, it is even tougher.  



[00:48:45] Unknown:

Another interesting element of the data question of what you're working with is the challenge   of kind of data quality, data validation. You already mentioned a little bit about kind of the lineage aspect of making sure that you can track, you know, who entered the data, how it was entered, when it was entered, where it went.   But   because of the fact that you are relying on, you know, site employees who are gonna have different motivations, different kind of levels of commitment to it.   You know, they're saying, oh, I just have to fill out this checklist so I can get to the next thing, or some of them are, you know, maybe the site managers, and so they're very invested in making sure that everything is correct. I'm just curious how you approach some of that challenge   of data quality, data validation,   given the fact that there is a human involved and just trying to account for that aspect?  



[00:49:30] Unknown:

Absolutely. That's great. And, like, there's 2 parts to that. 1 is metadata. Right? We haven't even talked about metadata. Like, I should have probably mentioned it earlier when we talked about the actual data types of data we're collecting, but we're also collecting metadata   on every response.   And the metadata is   where it was done geolocation   wise. We're capturing a geocode   or coordinates on every question being answered. We're also capturing   the date, time stamp of when they answered it. Right? So that's 2 factors of metadata that we capture on every question.   So you're right.  

Here's the issue, and I have tried. And if anybody on this podcast knows a way to do this that doesn't involve, like, some sort of electroshock through tablet or something, you can call me up. I'll you know, and tell me how you can do it. The 1 thing I can't do   is I can't make people be honest or do their jobs correctly. And, man, I have tried to brainstorm the hell out of that, but I cannot figure it out. There's nothing I can do to make you go and diligently walk   around and do what you're supposed to do. There's always a way to cheat the system. Right? And so 1 of the things that we've invented, and this is what I was mentioning to you offline earlier that I'm curious if you've ever heard of this anywhere else.   We are in this blessed   position because we are an analytics platform, but we control the means of data entry.  

You know, if you're a locker or, you know, whatever,   Microsoft BI,   if you're working on 1 of those platforms,   you can only connect to data that's already in databases that's coming in.   And you now have the problem of scrubbing the data. Right?   So you're making decisions off of good data, not garbage data.   But because we control the means of data entry, means it's coming through our system or it'd be coming in through our sensors, I suppose,   which are pretty locked down   closed loop systems, though I guess you could move a sensor to get up all the sensors in 1 fridge and everything else could be 90 degrees, but that would be dumb. I don't know why you would do that. But, like, we have this ability to do this thing, which I have named as a feature set as called data accuracy scoring.   So I can look at the checklist that's being completed in real time.  

I can determine if you are pencil whipping it or being dishonest   or are not on-site   or are not using the proper equipment,   and I can tag   we call it a question response set. That's basically a completed checklist. I can tag the response set with   accurate or not accurate. And if it's not accurate, I can tag it with why I'm telling you it's not accurate.   And what that does is at a checklist level and it's most prevalent in our dash boards,   you can go into the dashboard, and you can see all the response set tags. So you can literally do 1 click   on a checklist   and click the accurate tab, and you've now   just excluded from your response set   all the non accurate data.  

And by the way, I preach to my customers   all the data is good   because what we do is we tell you what's really happening in your business. So the non accurate data   is as valuable as the accurate data, but for different things.   So if I need to make a decision on a question   that's being answered by people,   like for instance, 1 of our companies,   they changed how they stored the they moved the position of sour cream on a make table, and they put an ice sleeve around it. And they actually wanted to see if this ice sleeve   was actually doing anything   to keep the sour cream colder because sour cream stick. It could be 1 of those things that a health inspector would go to pretty quickly because they go, I bet I can catch them wrong here. You know what I mean?  

So, like, they wanted to look at just those sour cream temps.   Well, because of our system, they could click on the accurate   tag. They could then filter out all the inaccurate b s data and just look at accurate readings. They could get a better sense on that question level.   Okay. Did this ice sleeve actually help the sour cream temps?   The not accurate data is not good for looking at actually what's happening. The not accurate data is really good to look at from a coaching managerial   perspective   of who's not doing this correctly,   and how can I coach them on the why behind this   so that I can get them to do what they're supposed to do? Right?  

That's the difference. So all the data's good. It's all coming in from these human beings. So like you said, they have different motivations. Their most of their motivation is to get this done as quickly as possible to get on to other stuff that I feel is more important.   And   so I just wanna give them a complete picture of this is what's actually happening in your business, and this is how the accurate and non accurate data   helps   your you manage your business better.   And it's a really interesting phenomenon   in restaurants. There's a culture   in restaurants specifically, but I suspect it's a culture across   the world, which was when you had paper checklists,   right,   no 1 ever looked at   them. No 1 ever used the data.  

You were very rarely getting caught   in real time or within, you know, a day, a shift   of not doing those procedures.   So the only people, which I we find is about 20%,   use the Pareto principle here. 20% of the people, because they are systems driven people, because they value systems   and they understand the value of systems,   taking away the guesswork,   making it easier, reducing stress, those kinds of things.   They embrace systems, whether people are following up with them or not. 20% of the people are gonna do the systems because that's their job or they value   the benefits of the system. Right? 80% don't care.  

And especially because in the restaurant industry, they always did them on paper. And like I said, no 1 ever followed up on them. No 1 ever looked at them. You know? You got in trouble 4 months later when someone came in to audit the restaurant. Hey. I've been doing your checklist. Yeah. Okay. Cool. And, like, on a lot of audits, like, the question about doing your checklist or not is, like, you know, if you miss it, it maybe got you down from a a 198. It did it had no effect on your life. There was no consequence for not doing them. You know, people just didn't do them. And so it's this giant   joke in the restaurant industry   that every checklist you look at is pencil whipped.  

And we even catch a lot of pencil whipping in our platform   all the time because unless you're managing to the pencil whipping you're not going to change the behavior.   And what's interesting about it too just as a side note is because I have this rule builder, I can lock these things down, man. I can make it so you couldn't do jack if you were doing it wrong. But then what happens is is when you do it that way, people just stop using the platform. They say it's impossible to do their job, and it is.   So what we decided to do was take a more passive approach to this,   allow people to do bad things,   but just track the bad things so that management could address it. Because it technically is a management issue. It's not a platform issue. Right?  

Like, if you're not doing your job, it has nothing to do with the platform. It has to do with you're not doing your job and managers need to come and talk to you. Your bosses need to talk to you about doing your job better.   So that's kind of how we approached that aspect of it. But that's that data accuracy scoring thing. Like, I don't know any analytics platform   that has the capability   of tagging stuff and making it 1 click to get accurate or not accurate data.  

[00:57:18] Unknown:

Absolutely.   And another interesting element of this is   the   way that you structure the inputs and the types of questions that you're asking in order to   kind of encourage compliance, encourage cooperation,   where if you just say, okay. I need you to check off these 30 things on this checklist just to show that you are doing the different things, people aren't gonna care. They're gonna say this is onerous. This makes no sense. It's not contributing any value.   I'm just figuring out what is that balancing act of collecting the information that you need, but doing it in a way that is kind of   empathetic to the employee who has to do that work and making sure that you're not wasting their time just because you are curious about what's happening in your business.  



[00:58:00] Unknown:

Absolutely. And that's where I feel like we're still in the infancy of this because we are just taking these paper checklists and converting them over, and we can't add dynamic elements to them. I'll get that in a second. But, like, the reality is is that, like, those checklists were the lowest common denominator   checklist. They had to be because they had to fit on 1 to 2 pages of paper, and they had to to sort of incorporate   the entire world. Right? Every different restaurant   just had to be kind of in there. You know? There are a lot of things that we can do personally that our competitors can't, once again, because we have this rule builder, we have this function capability   that we can do to make this better. Now let me give you some examples. 1 thing is,   Taco Bell, their temp log, they have about 4 or 6 ingredients that they wanna temp, but they usually run 2 to 3 different lines. So if you think about a Taco Bell, you have 2 exact or maybe even 3 versions of the exact same line,   1 for drive through and 2 for in die in store dining. Right? Well, they only care about, like, taco meat, the cheese, the chicken, the, you know, the steak. Like, there's 4, like, rice and beets. They only care about those ingredients.  

But, like, on their paper checklist,   they would go, okay. Line 1, write down what you count and then what the temperature was. And you would do that for 2 items on each line.   So,   like, that ended up being, like, a 55 or 60 questions   that if you did this on paper, you would have to go fill in. So because we have this rule builder, I was like, well, this doesn't make any sense. Like, first of all, if you leave it up to the the person, they're just gonna keep keping cheese sauce because they know that was hot. Right? And then you're not gonna get, like, a spread   or the safety or the coverage you want. So what I ended up doing is I built in a randomization model   where depending   it was actually it was random. So depending on when you when you got the checklist, you got   version a or version b of this line check. And what I did was I strategically did different ingredients on each line across the system. But I also told the people what they had attempt. So it might be ground beef and beans on line 1, but then it was chicken and cheese sauce on line 2, And then line 3 might have been rice and steak. Right? So you've got full coverage of food safety across 3 lines,   but you also didn't have to think about it. Does that make sense? So that's, like, 1 way we were able to design a more efficient checklist   for those guys.  

And there was a franchisee, by the way, not Taco Bell corporate. We were able to design a more efficient checklist to do that. The other thing is because we have this rule builder, business logic engine, we know who's completing the checklist.   So we can take that good like, we shouldn't treat all employees equal. If you're a good employee, you never pencil whipped in, and it's, like, always in there taking comments and doing it correctly.   You can get an easier version of the checklist. And if you're a bad employee, I can give you a harder version of the checklist. And when I say harder,   I just mean I would probably just add more photos   because photos require you to walk over to a place and take a picture, but that   action   drags you from being able to pencil with this thing in a booth in the front.  

So, like, our big analogy is always, like, I don't want people to come in, grab a cigarette, grab a cup of coffee, and bang out their food safety checklist for the next 15 hours at 7 in the morning. Right? I want them to go at 4 before dinner starts and make sure the food's safe then so we don't get a bunch of people sick at dinner.   So by adding more photos   and adding more, like, temperatures that you have to do with a thermometer, I can actually make the checklist   force the person   to at least walk to that general area,   and then hopefully, they'll just do the work because it's the less onerous thing to do. Right?  

So   I can do those kinds of things as well.   And what I'm finding is is, like, you know,   a lot of these people become like dictators. The checklist right here is not us, but like the clients. They become like these dictators because they've never had control, and they build these   hour and a half long checklists that, like, nobody could ever do because they're just too busy running the restaurant to do. And it is a battle for us to try to fight these guys and go, hey, man. You need to chill out.   And, like, you're making it too hard on your team. You know what I mean? And, like, I try to get them to a point, like,   ideally,   this should only be a 10 or 15 minute process that hits those key items. And that's where randomization can come in too because you can get randomization across a day. Because some of these guys are doing these checklists every 4 hours. We couldn't randomize things so we get coverage across an entire   table   or food group   by doing the 2 questions every 4 hours versus trying to do 20 questions every 4 hours. Do you know what I mean? Like, at some point, we have to make some assumptions that, like,   if the table's cold, then the food in the table's cold, and we should be fine. We don't temp every ingredient.  

So there is, like, a lot of play there.   And 1 of the cool things we can do too, just as a side note, we can do what we call a dynamic checklist, which means, like, you know, if you take a chain of restaurants, 90% of them are the same, but 10% is different. They might have different equipment. They might be in a mall versus a drive through versus, you know, all these different little attributes, testing of different food items and stuff like that, different menus. We can build 1 checklist that can dynamically   conform to each location as it loads   so that it's always what we call sheet to shelf. That's another benefit   of using a system like ours   is that now you can require every question to be answered and you take the guesswork out of the guy. You don't you don't need some high school kid determining whether this cooler temperature is important or not. You know, you just need them to go temp it and tell you what it is. So we can do stuff like that too. And those are all things that we try to do within the checklists   to make people   do them correctly. But I do think there's gotta be a revolution   in checklist design   to really figure out   what's important and what's not. Because right now, we're working off this antiquated 19 eighties paper version,   you know, that's been just updated with menu changes over the years. And it's not   really,   like, you know, like, we're in the infancy. We need to get to that next generation where we're looking where people are designing checklists based off of what makes the most operational sense to check to make sure we accomplish our goal as quickly as possible.  



[01:04:27] Unknown:

You wake up to a Slack message from your CEO who's upset because the company's revenue dashboard is broken. You're told to fix it before this morning's board meeting, which is just minutes away. Enter Metaplane,   the industry's only self serve data observability tool. In just a few clicks, you identify the issue's root cause, conduct an impact analysis, and save the day. Data leaders at Imperfect Foods, Drift, and Vendor love Metaplane because it helps them catch, investigate, and fix data quality issues before their stakeholders ever notice they exist. Setup takes 30 minutes. You can literally get up and running with Metaplane by the end of this podcast.   Sign up for a free forever plan at dataengineeringpodcastdot   com/metaplane   or try out their most advanced features with a 14 day free trial. And if you mentioned the podcast, you get a free in data we trust world tour t shirt.  

In terms of your experience of building this platform, working with your customers,   understanding,   you know, how that is being applied to these different businesses, and particularly as you start to expand across verticals,   what are some of the most interesting or innovative or unexpected ways that you've seen the platform used?  

[01:05:33] Unknown:

That's my favorite part, dude. Like, honestly,   like, when I see people do something in our platform that I had no   like, I never even thought of before, I'm like, oh, this is the best. And what it's kinda cool too just as a side note. As my job has evolved and as we've grown over the years, I'm doing less and less, like, daily operational stuff, and that frees me up   to go. And kind of 1 of my main roles now is to meet people in these other verticals and just hang out with them   and see what their challenges are and try to figure out, can we   help them   do their job better, right, and get those first use cases   for that vertical, but also just understand what their business is really doing. Right? So here's a dumb 1, but it's, like, totally cool.  

The Pizza Huts on there, they use their register as also their time clocks. Right? And so it can produce a ratio, which I don't understand the ratio at all. I don't know how they calculate it. All I know is that anything less than a negative 2 is good, and it's a labor to sales ratio.   And once again, for my job, it's really not important that I understand how it goes into it. I just know   less than a negative 2 is good. So negative 4, good. Right?   And so   they have a checklist of these, like, 75 Pizza Huts where every 2 hours, they make the guy go in, pull that number, and he puts his puts it in our system. And then we can obviously evaluate the number. And if the number's, like, a 0 or a 1 or a 2,   then when they submit that, we notify   the management team   of   the company that says, hey, dude.  

This labor ratio is too high of sales right now, so somebody can follow-up   that day.   Most people get that   labor to sales number   for Tuesday on Wednesday   when they cannot impact a change that affects profitability. Right? But if you can get that in real time every 2 hours,   if the labor's running high, then somebody can pick up a phone and call the restaurant and be like, hey. Why is labor running so high? And there might be a reasonable excuse. We're training new people. We have a giant catering order at 7 o'clock. I needed 5 guys in here making pizzas.   Totally understand it. No problems.  

That's just a normal aspect of business. But being able to catch that, like, in real time   and get it to the right people,   that's, like, a really interesting use case. Another interesting use case that I'm working on right now,   I have a guy who has, like, on the cutting edge of recycling paint. And he recycles paint in New York and in Colorado and Texas. You know, the big paint drives, and the state pays them to take the paint and mess with it. They've got a report to the state about how many pallets, how many buckets, how many   gallons of paint came in.  

And then what happened to the cans, and what happened to the plastic buckets, and how much went out the door. You know what I mean? Like, they've gotta be able to, like, justify to the state that they're actually doing an environmental   good. And so right now, they do that on slips of paper. You know what I mean? They just jot them down on paper, and then this guy has a data, enter it all in at the end. So we're working on an app. We're working on a checklist where these things come in. They can be just data entering that stuff into the system as it comes in, taking photos of it so that when they have to go to report to the state, it's like, here. Here. Just here's the report. I just emailed it to you versus,   you know, having to run all this stuff in an Excel and double check every number and whatnot.  

So that's another interesting use case.   We are getting some interesting use cases.   I probably had some of the first digital COVID screens.   By March 20th 20,   I had   national chains using my COVID health screens   where it was a series of questions that you went through to term that's like, you know, the paper you get everywhere you go. Blah blah blah. I always said this. We were doing that digitally, but and so I have millions and millions of rows of health screens, but I had built 1 that people could do at home   before they drove into the town because it was actually for a hotel in San Francisco   because we were like, well, why the heck do we wanna get this person who has to sit in San Francisco traffic, and this could be true of any major metropolitan area, have to drive downtown,   find parking, come to the back door, and then we find out that they have a fever and they can't work,   you know, or they took the train,   or now you've got COVID germs on your back door or in your, like, little, like, you know, entry to your building. So we literally built 1 that people could do at home. They could take a picture of their thermometer. They could answer the questions, and then it would alert their bosses,   and it would tell them right in the checklist, like, you cannot come into work today.  

We're alerting your bosses. Stay home.   Like, we built that kind of, like, interesting stuff too. You know? Because it's like an Excel, man. It's just like we can just build cool stuff that help people actually run their business. And that's like the most exciting thing ever.   Because,   you know, when we first started, I knew every customer, and I knew every checklist because I was a part of every 1 of them. And now that we're bigger, it's like people just keep using the system and doing cool stuff. And I don't even know of it sometimes. You know? And then I find out about it, and it's like, wow.   This is like I built something. You know?  

It's cool. That's, like, the coolest part of being a software founder is that when people actually utilize your platform and do things that you   never envisioned,   Or they just keep staying on and keep paying you. And they don't leave. And you're just like,   wow. Like, this is real. You know what I mean? Like, we did something that's, like, didn't exist before, and   now people   see value in it. You know? It's so it's that that's the best part.  

[01:11:15] Unknown:

And in terms of your work on building this platform, bootstrapping it, understanding   the problems that your customers are trying to address, and how to   provide a kind of technological solution that is able to be repeatable and scalable? What are some of the most interesting or unexpected or challenging lessons that you've learned in the process?  

[01:11:35] Unknown:

Definitely, I would say and you asked this question earlier about product knowledge or, like, vertical knowledge.   You don't wanna start in a vertical that you have no knowledge in. You know? You're just adding more complexity and pain to your world than you need. Right? Like, we spoke restaurant, but we spoke it well. And we actually have, like, a, you know, a background in restaurants that we could trade off of and gave us some credibility.   So I think that   first part is huge.   The second part is when you're bootstrapping,   everything takes 4 times longer than you think. I mean, first of all, I I developed too. For about 5 years, I developed workflows. In our previous company, we were in the Symantec workflow Symantec partner channel before Symantec got bought by Broadcom. And we did we did a product called Symantec workflow where we did integrations to large   enterprise, Symantec customers,   usually integrating, like, security software into help desks or IT management software, that kind of thing. I spent I spent a lot of time at TJ Maxx outside of Boston working on stuff for those guys.  

And that was a horror I mean, not TJ Maxx in general, but that was a very hard, horrible job because we were building custom software brand new all the time. That was brutal.   Because they were just consulting gigs, and they were expensive, and they were hard. It was so it's a refresher to come over here and just do 1 thing.   But,   you know, developers cannot tell you how long anything's gonna take. And it's always, like, 3 to 4 x longer than they think it's gonna take. And I equate development to, like, every HGTV   show where they're rehabbing an old house, and then they have to do something in the basement. You know? Like, love it or leave it up in Canada or whatever where you know? Then they're all mad at her, the lady, because she's trying to rehab their basement, and they've got, like, a puddle and, like, piranhas in there and, like, wolves living in the den. Like, well, the wolves are biting the construction workers, and we're gonna go over budget. And now I can't have a pool table. You know? And you're like, whatever, dude. Like   like, that's what development's like. It always takes longer. The human brain can't fathom   all of the little challenges that are gonna pop up and little things that are gonna take time to solve. So if you're bootstrapping,   you've got to understand   that that's everything's gonna take 10 x longer. You know?  

Like, I didn't tell you this, but when we first built our platform,   we had a version on semantic workflow that I had built in 2013 and 2014,   and had beta tested and got into a chain of restaurants on the East Coast, 18 unit chain. And that was our initial product. And 1 of the reasons our developer came on, and this is something for you guys that are gonna be founders,   developers get asked all the time to join companies. If they know how to code, they get asked constantly to join companies,   but their companies never have a business plan. It's always like, hey. I got an idea for an app. You wanna come develop it for free, and then we'll make a $1, 000, 000? And developers are tired of that. They're like, screw you. I'm not gonna do this. Like, because they know, like, they're never gonna turn in anything. And everything looks good and simple at a 30, 000 foot level. It's when you get on the ground and start actually implementing it that you realize there's 4, 000 tech challenges that you didn't expect, and that it's gonna be harder and take longer.   And then all these ideas fizzle.  

1 of the reasons our developer came on with us, because we were another guys that came to him and said we had an idea,   is because, a, we had a beta platform that we had built on workflow, which most of you aren't gonna be able to do, but unless you know how to code something. Right? But we had a beta platform, and we were fronting all the cash. So that's why our developer actually signed out with us. But our first version of the platform   was, I guess, the workflow version, which was just a stop gap to start marketing and selling   on day 1, but was never gonna be the platform forever. We were getting off of it. We had something on day 1. We could at least go out and sell and demo and have conversations and learn how to do all that because we had never done that before anyways.  

But our first version was a very 2015 version where we had a hand coded   our own app that we had coded, a front end mobile app that you could use on an iPad or a phone or on Android   and do the checklist,   but we actually built our back end   on a workflow platform.   So it was actually a platform called decisions,   which was   basically the people who had built semantic workflow. So we knew them. When they had sold out to Symantec, they went and built decisions right after it. So we were getting on to a platform. It wasn't exactly the same, but it was similar.   And we used the workflow platform for, like, stuff like user management, location management, reporting,   rules,   all those kinds of things. So we had an app, then the data was was just a data collection point that then shipped all the data   into   our API that the workflow   platform then did. So that that was a good hybrid model. That's an antiquated model in that most people will not buy an app or a platform that's got an app and a back end that aren't the same. Like, that's just that's antiquated. Right? It's that's so 2015. You know what I mean? That's not 2022.  

So but that's how we did it. But that got us to market faster. So I would say the next part of this thing is if you're gonna do this,   you gotta get selling even before you gotta start having conversations.   You gotta have a website,   a URL,   a blog,   and maybe a podcast   when you're in development. And you have to start having conversations   immediately knowing that your software is coming   and you have to be disciplined to get an MVP out and get it into people's hands as quick as possible and start getting customer feedback on what it's missing versus just trying to guess.   Because you oftentimes guess wrong about what's actually important and what's not important. Just know that your development's gonna take longer and your sales are gonna take longer too by 2 or 3 or 4 times.  

So you have to start having conversations   like the day you start the company even though you have nothing to show for it. You have to start learning   and exploring and walking the floor if it's in manufacturing and just understanding what's actually happening.   And then you just gotta keep going.   I I can't stress it enough. You know? Like, you just have to keep going. We didn't make money for years.   You know? It took us like we weren't a uniform. We just didn't have a $1, 000, 000, 000 and a 100 developers on day 1. We had like 3 guys.   And we just had to keep battling   every year. And the only reason   we didn't stop, and we should've stopped. Stopped. There are many times where we probably should've been like, let's just throw in the towel.  

But it was because the numbers, even though it was painfully slow, were creeping up. They were never going down for very long. You   know? Even though   we weren't a unicorn, we weren't a rocket ship, we weren't we didn't go from 0 to a1000000 in 6 months.   We went from, like, 0 to, like, 500 in 6 months. You know?   The numbers kept creeping up. That's the 1 reason why I think we both kept going   and kept battling it out. It was because we just were getting   better. And we weren't losing customers.   You know? That was a huge part of this. But, you know, unicorns are unicorns.   We all hear about them, and we think about how many 1, 000, 000 of dollars their founders make. And it's, like, kind of our motivation, but 99.9%   of us aren't unicorns.  

And you just have to keep battling. And if your numbers   are going up, even if it's slightly, I think you gotta keep going.   You gotta keep battling. Because   it wasn't until   April of 2017   after we got off that workflow platform and recoded our entire back end and added schedules   that we really started to get   good traction in demos.   So 2 and a half years,   almost 2 and a half years in, is when we finally had a platform that we could show people that was like,   oh, that's an interesting feature that, like, nobody else has. This might be something I wanna use. It took that long. You know?   Because we only had 1 main developer, and I had actually coded a lot of the workflow back end just to make things work   in year 1.  

So,   I mean, just keep battling. And then you know what?   It really does start to happen. You know, it's kinda like when they talk about when you meet your spouse and you're like, you just know   type of thing. And, like, until you just know, you don't know and you don't understand that concept. But when you meet the person you're supposed to be with, you go, oh, I just knew. Like, this was the right person. This doesn't feel weird.   We really started to get momentum in, like, 2019.   We started winning big enterprise head to head battles, and we were able to win.   That was the kind of turning point year from us where we went from, like, a tiny company to,   like, kind of a a small company. Yeah. People are teeny tiny to small. You know? But, like, that was a big year. And then really since 2019 on,   it's just gotten easier. And it's like, we'll get deals today that in 2017,   I would have, like, cried and, like, hugged people. We'll get those, like, on a daily basis now. And I'm like, that's cool. But, like, I sometimes I have to remind myself, if this deal had come in in 2016,   you would have literally, like,   had a heart attack because would not have believed that you were fortunate enough to get this deal because we are just growing, and we're just getting bigger, and we're getting faster, and we're getting better. And it does get better.  

But don't believe it's gonna be easy or it's gonna be fast   because it's not. And it's gonna require a level of patience   that you can't even imagine.   Like   like like Herculean patience.  

[01:21:12] Unknown:

For people who are interested in being able to gain better insights into their business or understand more about kind of the unit economics of the their different locations? What are the cases where Ops Analytica is the wrong choice?  

[01:21:26] Unknown:

I would say we don't do well with single unit operators.   They just don't   tend to stay on. And there's no mandate that they have to be on digital checklist system, which is a crime in and of itself, that we don't mandate   temperature logs in any, like, digital fashion so that when someone screws up, we can actually, like, investigate it properly,   which is just a failing of the federal government and the FDA.   But we're not great for single unit operators.   We are probably not the best choice   for you if you just want to audit   because there are some platforms out there that are strictly audits, and the reporting's,   just a little bit easier because you don't have to account for those kinds of things.  

Also, I would say,   I mean, we can build a simple checklist just as good as anybody else. But, like, you know   here's the thing. We're kinda, like, in the middle on we're, like, on the low end on pricing. So, like, I would like you to it'd be different if I was, like, like, I'm on the technical   edge, but I'm also the highest price. And I could say, well, if you don't really want all these advanced features,   then, you know, you shouldn't come to us because we're the Salesforce. Right? You can go somewhere else and get something cheaper. But we're actually   on the high end of technology, but we're on the low end of cost.  

So it's kinda tough.   So, you know, do you so, you know, we could still get you on a great price, and you don't have to use all of our advanced features. You know what I mean? So that's where we're kind of a unique thing. Like, we're purposely   small, nimble, and cheap so we can get as much market share as possible. Yeah. I would say that. Those kind of things, single unit operators, probably not.   And I would just say if you're,   you know   yeah. I guess that would be it really. Just I would say small single unit operator businesses or, you know, here's the thing too, by the way, and this is a problem that founders have and I have especially have,   is I have a way that in my mind, because I am like I have a hotel restaurant degree. I ran restaurants. I worked at this auditing thing.  

I have very set ideas on how the platform is supposed to be used.   And what I've learned over the last year   is I gotta shut up and not tell people how to use the platform. Let them use it however they want. And as long as they're seeing value   and they're happy and they're gonna stay on, then however they're using it is cool by me.   But that has been a huge lesson that I've had to learn because I've sat there and told people, you're doing it wrong, and they don't like to hear that. And rightfully so. Yeah, it's like the old adage,   people who can't do teach, right? It's sort of the same thing. Like, well, if you could do restaurants, you'd be running restaurants. So why the heck are you in software, you know? And so I've had to, like, shut up and learn that as long as the person's seeing value and they're happy, then however they're using it is fine. You know what I mean? Like, I'm sure the guys that built Excel,   when they found out, like, they're like, someone was putting a Christmas card list in there or putting in, like, a a grocery list in there, they were like, it's not for groceries. It's for finance and science. And they had to, like, let it go and be like, who cares? They're using it. Like, let's just be happy that they're using it, and they see value. And I feel like I'm in that growth as a founder to just shut up. So when you ask that question, it's kinda like, I was, like, in my mind going, well, you really need to be using it for all those advanced features. But in the back of my mind, I'm like, I've got some giant clients, couple 100 units, that they just utilize it for, like, action plans when they visit the restaurants. And they're happy as clams.  

So why would I care? But that's all they wanna use it for. They like it. They see value out of it.   What do I care? You know? That's been tough for me because I'm very opinionated.  

[01:25:09] Unknown:

And are there any other aspects of the work that you're doing at Ops Analytica or the overall data challenges of multilocation   businesses that we didn't discuss yet that you'd like to cover before we close out the show?  

[01:25:21] Unknown:

I I just wanna reiterate because I don't want this to be like I love our platform, and I hope that if someone's interested in learning more about it or looking for a way to collect data from the field, whatever that looks like, and being able to do some advanced calculation stuff, that they would reach out and check us out. But what I I wanna reiterate is this. I truly believe   that   operations management software   and utilizing   technology   to uncover the black hole of what your employees are actually doing or   what's happening out the world, right, in your locations,   is I truly believe that that is the next technological revolution.  

And it's not only going to exist in being able to capture data,   identify problems quicker, and then crowdsource solutions, and then fix those problems because you have the means   to do that all within this sort of operations management world that I do believe it's truly gonna increment sales up and that businesses that embrace that,   though and you see it with the Walmarts and, like, the Amazons and their distribution centers and stuff, the efficiencies, and how they get so good at stuff. That's what we're talking about, but we're just talking about it from human activity, whether it's the receptionist in a dental office to the restaurant employed or whatever.   But, also, we're rapidly approaching a world   where we're going to have employees working next to sensors, working next to robots,   gathering data, data coming in from POS systems and from the Internet. And these are all gonna be siloed systems.  

You know?   These operations management systems, I think, are going to especially where we're trying to go, is to play the role to coordinate all of the siloed systems and to take that data in and make it actionable.   So, like, this is the next big technological   battleground. We've got delivery. We've got online ordering. We got apps. We got all the things   that facilitate   sales.   But what we need to get better at is making sure that when people go into our businesses,   they have consistently good experiences   in clean environments,   and they get exactly what they paid for as fast as possible.  

And that can be your X-ray   or your Egg McMuffin, but they're it's the same thing. And, you know, just   as a whole, I want people out there to understand that this is next. Like, this has to go in place before you can have wide scale automation   across your businesses.   Because you gotta have a way to know what's actually happening and then tell these robots what to do. You know what I mean? Like, they're not just gonna figure it out. So we've gotta have this layer of integration   and this layer of data.   So I mean, that's important. I think too, just data driven decisions are so much easier and so much less stressful   to make than just gut feel experience based decisions. So you can implement a system like this and actually   make your managers and your leaders better   because they can make better decisions. So I guess it'd be the 2 points I'd wanna get out there.  



[01:28:22] Unknown:

Alright. Well, for anybody who wants to get in touch with you and follow along with the work that you're doing, I'll have you add your preferred contact information to the show notes. And as the final question, I'd like to get your perspective on what you see as being the biggest gap in the tooling or technology that's available for data management today.  

[01:28:38] Unknown:

The biggest thing I see is, right now, we have to utilize people to hand enter the data.   How can we get to a point   where   we can get the human data without having to stop somebody   and having them enter it manually.   Right? Which I guess is sensors, but, you know, like, are we gonna do those, like, light studies that that guy did in the early 1900 where he put lights on people's hands and did slow release film to watch them assemble a product to try to find where they crisscrossed and stuff? Like, how do we get to a point where   we can either use like, because sensors are stationary. So sensors are in the most part, sensors are usually stationary. IoT sensors are at least. They're put on a device or in a box or they measure a singular   aspect.  

So that's fine,   but that's not the answer. I think the answer is,   is there a way where we could be collecting data off of a human being? Maybe because they were wearing gloves or smart glasses or something where we could be checking thing. We could be collecting data as they move, or can we get more mobile robots   who can be doing some of these human activities   so that a human doesn't have to physically sit there and enter data? That to me is the big   question mark on where do we go next? And I will tell you, I'm partnering with a robot guy right now.   We're trying to figure out   how do we get the data.  

And then the thing is too, it's like, you know, it's easy to put a robot on a stationary thing.   But, like, you know, think about a restaurant. It's not a stationary thing. There's 50 different things you gotta check on all over the place. So it's gotta be mobile. You know what I mean? How can we get mobile data collection   that comes in either off of a human or off of a robot   that we or cameras or sensors that we got to have 100 of them to do it. Right? You know? How do we get that gap so someone does not just physically go, yes, no, yes, no, 33, 33, 33?   And fill that gap where we're getting the benefit of the check without having to make a human being do it. And it's gonna be a combination of things. Right? It's gonna be sensors,   robots, drones, cameras, people.  

I wanna get as much of the people part out of it and get as much of it automated as possible. That's the big gap. Absolutely.  

[01:30:53] Unknown:

Well, thank you very much for taking the time today to join me and share the work that you're doing at Ops Analytica and your insights into the   data and reporting needs of multilocation   businesses,   regardless of industry. It's definitely a very interesting and important area of effort. So I appreciate all the time and energy that you and your team are putting into that, and I hope you enjoy the rest of your day. Oh, thank you so much, Tobias. Take care.  

[01:31:22] Unknown:

Thank you for listening. Don't forget to check out our other shows, podcast.init,   which covers the Python language, its community, and the innovative ways it is being used, and the Machine Learning podcast,   which helps you go from idea to production with machine learning.   Visit the site at dataengineeringpodcast.com.   Subscribe to the show, sign up for the mailing list, and read the show notes. And if you've learned something or tried out a product from the show, then tell us about it. Email hosts at data engineering podcast.com   with your story. And to help other people find the show, please leave a review on Apple Podcasts and tell your friends and coworkers.