How Celect CEO John Andrews grew Celect to $5.8M revenue and 16 customers in 2017.

hello everyone my guest today is John Andrews he spent the last two decades helping retailers distributors and brands optimize their omni-channel strategies and operations before his current company's select John was VP of product marketing strategy for Oracle commerce coming to Oracle via the indica acquisition where John was VP of marketing and product management he started his career with Deloitte and consulting strategy operations practice and he holds a BA in economics and computer science from Boston College and received his master's degree from the Harvard Business School John are you ready to take it to the top that sounds great all right so tell us about select what are you doing what's your business model how do you make money right so we are at the highest level where predictive analytics technology company we are focused primarily on retail the company was actually founded out of MIT by a couple of MIT professors we'd been collaborating on a lot of research for the better part of the last decade specifically around this idea of understanding customer choice which I'm happy to give a little bit of detail on but the the main area that we're focused on with retailers is around inventory optimization so helping retailers optimize inventory optimize turns reduce stock outs reduce markdowns etc if you think about inventory right it's been the largest number on a retailers balance sheet it's also at the end of the day the the most important thing a retailer has to figure out right what products to bring into inventory how much of them in what assortment and then how to allocate them to all their customer touch points in terms of stores fulfillment centers to get them shipped out to customers if you can optimize that so that you've got the right products at the right place at the right time you're gonna make a lot more money really really in in in retail right now yeah that's a that's a that's an important thing yeah I mean look I have always wondered how these retail entrepreneurs figure out not only what style of things to carry but then multiply by another factor of complication because you have sizes right different sizes different colors I mean you have literally infinite choices that is exactly the kind of thing I don't want to be doing because I like simplicity so tell us how you mean can you give us examples or you're working with and how you help them double down on what's working through the data you provided yes yeah absolutely so the the process that we help our customers with right is kind of through the the merchandising and planning process and into the supply chain process right so through kind of a you can think of the inventory optimization cycle as kind of a plan buy allocate and fulfill process right on the planning side we're helping our customers and our customers include folks like Urban Outfitters Anthropologie Free People aldo of montreal up in montreal the shoe manufacturer designer and and and retailer Saks Fifth Avenue etc and if you think about that if you think about that process on the planning side right what we're helping retailers figure out is very specific as they're making their strategic planning decisions if they're figuring out how much inventory dollars to put towards you know following the trends and understand where it understanding where the demand is right understanding should I be going bigger into women's clothes versus men's clothes at the macro level and then figuring out you know something just as specific as attributes of specific items for shoes should I be going with riding boots versus ballet slippers and what colors are trending etc there's you know as you mentioned a minute ago the combinatorial explosion of all of the different decisions that need to be made and the attributes associated with that in terms of you know trends changing the amounts that need they need to spend it's incredibly complex the way that these decisions are made today is generally based on gut instinct and Excel spreadsheets right and you know our tagline is bringing science to the art of retail it very much still as an art form especially with fashion retails which is the area that we've had the most success it's on how are you making the brain smarter there right so your your system your engine what you are paying for is only as good as how well you've trained it and maybe measured by number of lines of data you've put through it right how are you training this thing so it's a good question there's it there's a few elements here right when people think about training one of the things they think about is is machine learning right in terms of you know that the data that you're bringing through and how are you how are you optimizing that model one of the things that we're doing is understanding it sure you know you can you can look at a product that you're looking to sell and train the model to look at images of other products you sold or other data right so kind of come up with a model that's important but more specifically it's getting through a deeper level of detail in terms of understanding customer choice and customer preference so let me give you a specific example right a customer walks into a store and they buy they buy a product right great every retailer is going to use that data and that information that transaction level data to optimize their operations in some way right but what if you knew in addition to what that customer bought what if you also knew it was available to them when they made that selection right what if you know what their options were when they made that choice said another day what what if you knew what they didn't buy in addition to what they bought right I now have a sense of the customer preference I have a sense of what their intent was when they made that when they made that selection and when you look at every single interaction with a customer when you look at the the intent when they choose one product over another and you can do that by understanding you know what was in inventory in a particular store you can do that by looking at browse information if you go to a retailer's website and you look at five product detail pages but then you only put two of them in your shopping cart that's context we're now able to build up a model a pretty robust model we call our our choice engine right is a choice model that allows us to then answer the question not just what did somebody buy but what would they have preferred to buy if given the choice across an assortment of products all right because what's happened in the past isn't necessarily wasn't optimized right maybe you didn't have the the the best assortment in front of them right I bought a purple button-down shirt but maybe the blue one wasn't available and I would have bought that if it if it was part of the overall assortment so being able to normalize against that and then build out that model now you're now you have to as you start to bring in new products into the mix something that you've never you're designing you've never sold before that's where then the machine learning comes into play to say okay help me build a model of this product that I can then bounce against my choice model to I diamond to identify what that demand is going to look like when you do that right we've seen customers with anywhere from you know that points 5 to 7 percent increase in revenue to you know afterwards of you know 13 to 14 percent increase in gross margin yeah I mean look many I mean you sang all this I can't help but think and I'm sure people listening or thinking yeah this is why Amazon is so big they have the best data collection engine anywhere and they can make it smarter than anybody I mean you even mentioned you gave the example you like product pages because you can see what was viewed you can't necessarily see that in the store you know was there and you know what I bought and didn't buy but you don't know if I pick something up and put it back down or I mean I don't think you do unless you have trackers and things in every single store ya know now the technology is getting better where there are in-store sensors there's RFID where you can see what goes into the dressing room what you know what comes out and then what goes up and people buy the reality is nathan is that you don't actually need that level of granularity to get the signal out of understanding customer preference right part of the part of the technology you need to identify what the selection set is that people are low likely looking at just by understanding what was in inventory and then what the customer bought gives us significantly more signal than just the transaction level information of what a customer bought now you you brought up Amazon right and everybody in the retail perspectives looking at Amazon in terms of what they're doing one of the benefits that Amazon has is just an enormous amount of data yeah the challenge that other retailers have even though they feel as though they have a lot of data the issue is is that they actually has very sparse data about an individual customer and individual products and specifically what those customers interacting with those products right so the reality is they need to be able to pull signal out of what is actually very sparse data and that's one of the things that makes this extremely hard to kind of pull that signal out where that idea of understanding choice becomes becomes critical I get it yeah what's the business model how do you make money so we're a SAS based subscription model what we do is we take a customer's data we run it through our engine and then we expose that information via a web-based interface that customers can interact with on real time okay so I'm mentioning earlier from you know the different solutions that I talked about on our own plan optimization by optimization the the interaction on the plan side is it's it's...