2023-Management-Covariance

MIKE FLEDER: Hey, great to be here today. I'm Mike Fleder, founder CEO of Covariance. I spent a decade here at MIT in LIDS and CSAIL. And the roots of our research started here as part of my dissertation. We do one thing. We tell you about the world you can't see today. We're the operating system for companies on the external world.

Our team consists of a lot of folks here from MIT, and machine learning, and quantitative finance. And we are focused on what is happening outside of the world that you can see today. So when companies want to get insight into company into what's happening outside of their walls, they buy things like research reports, which can conflict and not agree with the internal numbers. If your competitors-- if you're interested in competition, your competitors, maybe some of them are public. And you'll get some numbers a few times a year in their quarterly reports depending on the country.

But those public disclosures, information that you can glean from. Reports are delayed course aggregated. They don't have the KPIs or customer insights that you want. And are a very limited segmentation. This is really hard to do. So to get any extra resolution. So maybe you have a customer, maybe you could see revenue on a customer-- or sorry, on a competitor. And best case scenario. But how does that break down into segments? And how do those break down into regions? Or what are individual customers-- how are how are they shopping at your competition or interacting with your competition?

And so when you look at the problem of, hey, what's happening outside my business? Are customers turning? Are my competitors winning? What's happening regionally. It's mostly question marks when we're talking about the world outside of your business. We fill in those gaps.

So the structure of this problem is very different from the type of machine learning that big data analytics where there's been a lot of success in recent years. In the last decades. So in a lot of problems, you have tons and tons of data. Tons and tons of signals. For the problems we're talking about, there's barely any ground truth. Maybe you have-- we operate today trillions of rows of data. Petabytes of data. And those are all signals. But in terms of ground truth information like what are your competitors? What are their actual financials? Or why didn't someone walk into my store today? It's unbelievably limited. And so the scale imbalance is huge. Worse it's all time series, which makes the problem increasingly difficult.

And the reason and the structure of this problem is such that new theory was needed. And that's why we started in academia developing a lot of the new theory that's needed to handle this type of data imbalance. And now it turns out it works really well in scaling that up and have been successfully commercializing it. To do this, it's a very complex scientific process. We're today a managed platform where we take the data assets that you have, we can combine it with additional external third party commercial data sets. Can operate securely in your environment as needed. We run that through a tightly controlled scientific forecasting process where this is the same type of process that you would see at a tier one quantitative hedge fund.

And the reason I make that comparison is Wall Street is the place where most of this type of work has historically been done. So the people who have spent the most effort historically tracking companies have been hedge funds in terms of tracking companies from external data. So we're taking those types of tools and skill sets. We've extended them. We're now at the bleeding edge and making that available to a broader market. And what comes out is really simple. Tell me what's happening across the street. Or tell me what why are my customers turning. And where are they turning to.

A big theme of our work is zoom in and enhance. So probably not that interesting. Maybe we do work for folks who are interested in everything from course aggregate financials like what's going to come out in my competitors quarterly report. How are they doing today. But zooming in enhanced for me. So where is my competitor's revenue coming from? Is it coming from in-store revenue or online? If I remodel a store, what's the effect on my customers? Or what's the effect on my competition? Zoom in and enhance again. Who is going into my store? Or who's buying my product? Why or why not? And how is that having an effect on my competition?

And so we've done this type of work for Fortune 100 firms across a variety of industries. And the general theme is what's happening outside of your walls. Most of your customers time is spent not interacting with your company. And that's what we're providing a view on. The alternative today is really to buy a bunch of conflicting reports, which are often can be directionally wrong. How do when they're wrong. And these reports can come from surveys, panel data. There's tons. There's thousands of data sets, which is fantastic. There are tons of external data sets you can buy today.

But data is not enough. And it's actually very dangerous to use raw material and treat it like the internal data that you have. Because the external data that you buy has bias and tons of tons of problems, and errors, and gaps. And renders it unusable unless it's processed properly. And so in our hands, we're processing combining all these types of data into a single view, whether it's on your customers, your competitors customers, or the market.

And what comes out is not just a number at a time, but give me a full view-- a full depth view of everything from individual customer interactions all the way up to how does that relate to financials. Changes in my financials or my competitors financials. We do this work to track things like in retail we've done work tracking share in various cuts of that. TAM. It could be things like for a retail grocer, where if people aren't shopping with me, are they going to restaurants? And so the general theme is take the data that I have, the first party data that I have as a company, and how do I combine it with these third party data assets to get a single view on both the external consistent view with the external world and the internal world that I know?

We're part of stacks 25. And we've really enjoyed all of our interactions with MIT and ILP. We're working with a variety of industries. Lending financial services, CPG, retail. We have had an enormous success with pilots, which is where we love to start and prove value really quickly. We have a highly successful track record and in going from pilots to adding long term value. We really enjoyed our interactions here with everyone at the conference and ILP. And look forward to your questions.