'We are data guys, pure data guys.' Eric Gillespie, founder and CEO of Govini. (Jennifer Milbrett / GovMedia)
In 2009, President Obama signed into law the 2009 American Recovery and Reinvestment Act to jump-start the broken economy and vowed the government would track “every dime” of the law’s $840 billion stimulus to weed out fraud, waste and abuse. That gave Eric Gillespie a great idea: Use big data technologies to collect, synthesize, analyze and present government spending data in a digestible way.
“At the time I believed that was largely impossible, so I built a website, the private-sector corollary website to Recovery.gov and I called it Recovery.org,” he said in a recent interview with Federal Times Editor Steve Watkins. “And we did track every dime of stimulus spending a good nine to twelve months before the government was able to stand up a website and begin tracking data. What I saw when we built that was the largest users of that website were actually government agencies because we were able to track and do more with the data than they could even do on their own internally, which was an indicator that there was a data problem.”
Quickly, federal vendors also came to realize how vital that data was from a business intelligence perspective. “And that was the very early genesis of this business,” said Gillespie, the founder, CEO and director of Govini, the newest government market analytics firm, based in San Francisco. The company continuously collects information from 91,000 sources to maintain a deep database of government market intelligence. His company’s secret is the technology he uses to synthesize it all into an accessible and searchable resource. “It is data-as-a-service in the way software-as-a-service emerged,” he says.
Gillespie, who held senior positions at IBM, Scient and CSC, had launched several startups prior to Govini. “You learn the most from the ones that fail,” he said.
But Govini is different from the others, he adds. “The market opportunity for this is significantly larger than anything I have seen. Again, and it comes back to the volume of data that the government produces these days and the things you can do with that data. The challenge is it is so fragmented and so unstructured you cannot make sense of it as a business or a voter or a taxpayer. And because we have solved that, we are now able to spin up products very quickly and provide answers to big problems that businesses have in ways you could not three, four or five years ago.”
One of the compelling aspects of the Govini story is that it is a tech startup and represents kind of a marriage of West Coast Silicon Valley and East Coast federal government cultures. Can you discuss how Govini got started?
The thesis for the business revolved around three points. One, we saw this massive amount of government spending that was not getting any smaller. The capital that was exchanged between government agencies and contractors was at least going to stay the same for the foreseeable future, if not increase.
But with that, business was being done in the same way it had been done for 50 years. It really had not changed. The business processes around revenue capture were identical to what they had been 50 years ago despite in the private sector there being great advances in those business processes and in how technology was leveraged to power businesses. The interaction between government contractors and government agencies had not kept pace with the private sector. That kind of is the second dynamic.
And then the third dynamic was the amount of data that was being sprayed around as a result of those interactions between government agencies and government contractors was growing geometrically. You had all this content being put out that allowed you to look at those interactions between government agencies and government contractors as a closed system, and you could analyze it empirically. You could look at it from a data science standpoint but nobody had.
So the thesis for the business was taking some of these modern big data technologies, viewing this problem as a classic big data problem and saying that is something I know we can solve. And we set out to do that.
How did you go about doing that? What were some of these new streams of content coming out and how did you essentially tap into those and essentially manipulate those into more workable, manageable data sets that could actually be telling?
The first part of starting anything like this, the founder’s perseverance has to be a key ingredient in the formula, but it takes some combination of entrepreneurial drive and group motivation. So for me as I started this business, I believe that nothing of consequence could be built without the right group of people to help build it, so we had to go out and hire really great people from the data science community to help stand the business up.
Once you have got those people you have got to convince them that shoving a spine in a jellyfish is a really interesting thing to do and you have to hire the right people to be able to do that. As an aside, I was actually bowling with a colleague who is a world class entrepreneur, built one of the biggest brands on the planet. And he said the key to motivating people is to hire motivated people. You have just got to put the right DNA in it to make it work. And for me starting this, it really started with that, with the business thesis and then from there we sat down and said all right, how do we look at the data?
Once we started looking at the data and thinking about a go-to-market strategy and how to build a business around it, like most startups, you have to kind of suspend disbelief. You have to believe that the laws of business physics do not apply to you and you had to get up every morning and kind of try to defy gravity. And this was especially challenging because one, we were on the West Coast with all of these great big data technologies that had emerged. Most of the private sector that was facing these challenges was on the East Coast.
So we had to figure out a way to marry those two things together and cross that chasm and it was pretty apparent both from a data standpoint, a technology standpoint, a constructing a team standpoint, understanding the deep needs of the private sector, we weren’t playing checkers, we were playing chess at that point. And we started looking at the spray of data that was coming out of government agencies and began breaking it apart into as many pieces as we could and then figuring out where to take the friction out.
So we completely kind of disaggregated the value chain of government data, agencies that produce data, and the needs on the private sector side of those companies that are trying to understand the market landscape and figure out what mattered to them. It was not just a technical problem, but a big business strategy problem to unwind that.
When you were securing the venture capital at the beginning, did you already have a proof of concept in hand?
Yeah, so just from a sequence standpoint, like many startups, you have a concept and you seed it with a little bit of capital up front. And you prove out this proof of technology baseline and you say, ‘Hey, I have got something really interesting that I think can be applied to a problem that people are having trouble solving.’ Most entrepreneurs divide the number of business cases into the cause and based on the result of that they go oh, that is an interesting thing to go figure out if the outcome of that is right.
So we have raised to date two rounds of traditional institutional capital from two of the best investors on the planet. And they believe as we believe that as the data from government continues to increase geometrically, the companies who are able to unwind that and solve problems with the data using technology are the ones that will win. Those are the ones that will succeed in the marketplace.
In the proof of concept that you presented to your investors, what did you demonstrate to them in terms of how your concept was going to differentiate yourself from the rest of the market?
So the market side, everybody cares about – this is 40 percent of GDP. So first and foremost, you have to look at it and say how big is the problem, how much capital flows into it, who cares? And lots of big enterprise businesses care, but in terms of differentiation we are typically put in the category of Deltek and Bloomberg. We do something slightly different though and as we talk to our investors about how we would be differentiated, we talked about it in a couple of different ways.
The traditional approach to this market was, we felt, fairly antiquated. The Deltek business is a back office cost accounting business at its core. Along the way they have picked up some data businesses. They have bought Input and Federal Sources and Centurion, but it is really a bolt-on to what their core business is. And in their own right they have built a fantastic business. It is a billion-dollar-plus business that the private equity industry took an interest in a few years ago. But at its core it is still an accounting software business.
On the Bloomberg side, the other company we are typically compared to, also an incredible business, but largely content, right? Editorial content, great stories around policy, regulatory issues, legislative issues. And in their own right, they do that as well or better than anyone. What we do is just fundamentally different. We look at it purely through a data lens and believe that you can learn more from the data than you can from the other ways of looking at the market. When you have some malady and need a dentist or a doctor but you do not go to the dentist when your hip goes out, right? We believe that’s the same with this market and looking at these transactions in the market, you do not go to an ERP provider to look at the data from a data science standpoint. You have to look at it through a data lens and we went out and hired data scientists. So at its core we differentiated by focusing just on the data, the analytics, the benchmarks, the metrics as a leading indicator of what will be happening in the market to be prognostic and say here’s what the future holds.
What we do today you could not do five years ago or 10 years ago, 15 years ago. The technology just simply was not there. The traditional businesses in this space have solved problems through muscle. They’ve put labor against it and come out the other side with answers just by simply putting people against the problem. We’ve proven through our clients today that we can get you 70 or 80 percent down that path without putting labor against it. Now at some point you may need to put labor against the problem, but the data science and the data technology allows you to get a lot closer to an answer without the huge expense of either internal staff or professional services staff or anyone else you might throw at the problem from a human capital standpoint.
Can you provide an example of that?
There are many professional services firms that just help companies with revenue capture. All they do all day long is go in and manage processes around revenue capture. They, by hand, pull data and run metrics and shove stuff into Excel and in databases and come out with answers.
In our tool set, we provide those same answers with three clicks. So we’ve taken a lot of the heavy lifting out of getting to those answers that big enterprise businesses in the space have traditionally wrestled with and struggled with.
And using the same databases, same sources?
Similar sources, yeah. I think at scale what you find from a technology standpoint is the platform that we built allows us to go out and in the market tap thousands of different sources instead of tens of sources. And we can do it rapidly in a way that you could not again several years ago because the technology did not exist to do it several years ago. Doing that at scale and not only doing that at the federal level, which has the smallest number of government agencies, but doing it at the state level, the county level and the municipal level of government where the number of agencies begins to explode is highly differentiated.
Doing it across every sector of government gives you a way to see through a single lens where your market is, where it is going, where it has been, who the buyers are, what the average price is, who is consuming your products and services in a way that you otherwise could not.
You just alluded to the thousands of different sources that you are tapping into at all levels of government. But some of these sources are data that is coming out maybe in a flash, it may come out on one day and then disappear. Give us a sense of what we are talking about.
There are transparency laws that require government agencies to provide visibility into where your tax dollars go. The government agencies have to provide transparency around things they spend money on so that you as a voter can have influence as a voter into where those funds are going election cycle to election cycle.
There are not any permanency laws, though, so even though government agencies are required to put up these contracts and this information, there’s no requirement that says they have to do it for any length of time. So oftentimes what you see is that the letter of the law will be followed, it will be made transparent for some period of time, but the spirit of the law isn’t followed because that content and data is pulled down. And when that disappears, when that is no longer on a website or easily accessible on the web somewhere, the only way to go back and recreate that longitudinal history is through a FOIA request, which is cumbersome and oftentimes in your business impossible to do.
So from a platform standpoint, we had to build something that was constantly out there pulsing the market, grabbing things and pulling them back in so that we did not run the risk of them disappearing. And the audit trail that we would otherwise have, we could not follow the breadcrumbs back. And that is again this business challenge or this technical challenge of things disappearing and not being able to recreate the history, and we have solved that.
Can you provide an example of the kind of thing that might just appear very quickly that your tools can capture?
Sure. Most all of this data is highly unstructured so there is no standardization even across the federal government for how this data is produced, much less at the state and county or city level. Sometimes you will see that documents related to a contract are eliminated from the contract history because they take up space. And that is probably a decision made by an IT guy who is conserving space or looking to move things out of the way to put more current information on a server someplace and it is not nefarious, but it disappears. And if you do not have the source documents that go along with the two lines of text, the context that you have for that specific project or that event, the market goes away and you cannot recreate it.
Anyone who works with the federal government and its data is quite familiar with how unstructured all these data sets are. And when you are talking about doing this on a scale of thousands of sources, how do you sync all that data up?
It is the secret sauce. That is really what it is. It has been done on a smaller scale in the private sector in other places. The life sciences industry has done this really well over time.
Gene sequencing is one example of where big data has come into play and you have learned really interesting things that were not obvious over time.
We spent two and a half years just building that, just solving that one problem of taking highly unstructured data that comes out at a very high velocity in a massive volume every day all the time and processing it into data that is meaningful, data on which you can build analytics and do interesting searches and find ways to see things that you could not otherwise see.
Where that becomes really interesting is taking what I would call orthogonal data sets that are not obvious and ladling them onto that same core data, the contract data for example that comes in from thousands and thousands of agencies and looking for correlations and trends that the market never see. And that is cool when you find those things. It is extremely rewarding.
What is an example of that?
The real world example, one that we have been talking about for the last couple of days with a couple of our clients, is taking the patent database from the USPTO[Patent and Trademark Office] and mashing that up with the contract database. So if you are interested in unmanned aerial vehicles, if you are looking at the drone market place, and you want to know not only who won the contracts but who has the best intellectual property behind remotely piloted aircraft, that intersection of who got the contract, how many dollars were spent and who actually put all the R&D money and research into building IP around it is a really interesting view. The market does not think about it that way.
So that is one example of what I imagine is a differentiating type of product that you can provide from your own technology, the secret sauce as you say, that perhaps you could not get from other vendors in the field. Can you discuss some other types of products or revelations that your data can provide that you might not get elsewhere?
The one obvious one, which I mentioned is the ability to look at the same company or product across a wide swath of buyers. So if you are selling copy paper, if you are selling airplane parts, if you are a systems integrator and you are looking at IT services or products, being able to not see the tiny granular pieces that the individual contracts but being able to see them in aggregate where hot spots of activity are happening and where there is a significant growth rate is a novel concept for this space.
And I think we do that uniquely. I have not seen anyone else doing that in the market where you can look across the entire market place on a single screen and see everything that has happened, everything that is happening today and everything that we expect to happen in the future based on all of that back testing and that history.
I can see how contractors in the government space could get a lot out of this. What about the government itself? Do you see any benefits or insights there for federal managers?
The government should absolutely be looking at it through this lens. We do not sell to the government. I am not saying we would not but our focus is almost exclusively entirely on private sector. If you think about the value that the government could get from it, we can tell you the average price paid for any commodity in the market across every government agency in the United States. We can tell you the standard deviation of that price, we can tell you, “Did you pay too much, did you get the best deal on the market?” Next time you go back and negotiate the price, you have got 50 data points to say, “Why did I pay 27 percent more for this than these other guys?”
How are your customers using your service?
We have lots of customers in the federal space today. One, a large defense contractor — a top-five defense contractor — who uses it purely for competitive intelligence. They look at it and think about how they approach their probability of win on specific deals relative to their competition.
Systems integrators use it to figure out what the technology install bases look like. IT product companies use it to understand where there might be professional services components to contracts around the core products that they sell. Law firms use it to fortify a defense strategy or a bid protest strategy, as a few examples.
Do you see Govini putting on events and conferences?
We are data guys, pure data guys. We happen to have solved this – the biggest of big data problems for government contractors and we know a ton about how they play. We sit at the intersection of government data and contract knowledge, which makes us pretty unique. If we can – I would rather us be the intel inside of the events and the conferences and the things that are happening out in the market place rather than be the host of those things. And I think we can provide a ton of value as the provider of analytics and data and forecast to those players in the space.