advertisement
advertisement

This Investment Platform Funds More Diverse Companies By Focusing On Data, Not Founders

By focusing just on numbers it predicts will drive success–culled from billions of data points in its system–CircleUp has come to support companies with a wider range of founders than traditional VCs.

This Investment Platform Funds More Diverse Companies By Focusing On Data, Not Founders
[Image: Dmitrii_Guzhanin/iStock]

When Madeline Haydon launched a startup called Nutpods–selling a dairy-free creamer made from coconuts and almonds–she wasn’t in a position to get funding from a traditional VC firm. The company was tiny, well below the threshold to get noticed. As a female founder, too, the odds weren’t in her favor; only 2% of venture capital goes to women-led companies. She is also a minority, and based near Seattle, not Silicon Valley. Haydon raised a small amount on Kickstarter, and then got funding from another source: a platform called CircleUp that uses machine learning and massive amounts of data to evaluate startups.

advertisement

CircleUp’s software, called Helio, tracks billions of data points about 1.3 million consumer and retail companies, automatically analyzing the same metrics that investors would typically consider manually, and predicting the likelihood that a startup will have breakout success. The company then recommends qualified startups on a marketplace for investors–and also directly invests or offers loans to some of those startups itself. Using data and algorithms, it turns out, dramatically changes the profile of companies that get funding: 35% of the companies on the platform are women-led, or roughly 17 times more than the industry average, based solely on the merit of the startups.

Ryan Caldbeck and Rory Eakin. [Photo: CircleUp]
At the time that the platform identified Nutpods, the startup was doing less than $50,000 in annual sales, but CircleUp’s algorithms gave it a high “brand score” based on product reviews and social media growth. The company raised $1.2 million on CircleUp’s marketplace in 2016, and the platform directly invested an undisclosed amount in 2017. Nutpods’s sales grew 500% last year, and the products are now available at major chains such as Kroger’s, Publix, and Whole Foods.

“Traditionally, private investors find and evaluate companies in a completely manual method,” says CircleUp CEO and cofounder Ryan Caldbeck. “They’re using a collection of heuristics that are developed over many years of doing their job, sometimes well, sometimes not well . . . those heuristics are inherently biased.”

One study of a tech startup competition found that male-led startups raised five times more than their female-led counterparts, in part because of the types of questions that VCs asked each gender. Investors also have a local bias; one long-term study found that half of investments happened within around 200 miles of the investor. Other studies have found that investors are more likely to evaluate startups favorably when the founders are most similar to themselves.

Even investors that use CircleUp’s platform, Caldbeck says, sometimes second-guess the platform’s recommendations. “Investors love funding disruption, but they believe they will never get disrupted,” he says. “It is always funny talking to them–they will talk ad nauseam about how an entrepreneur in any other industry on earth is naive if they don’t think they’re going to be disrupted specifically by data and machine learning. I’ve heard that so many times from investors. But they will always tell you it can’t happen in private investing. There’s just not enough data, or there’s other reasons why it can’t happen, or you need 10 years of experience to do it successfully, etc.”

In one case, an early version of the platform recommended Halo Top, a high-protein, low-sugar ice-cream brand launched in 2012 by a former lawyer. The startup didn’t generate much investor interest on CircleUp’s marketplace. But by 2017, Halo Top was selling more pints of ice cream than Ben & Jerry’s or Haagen-Dazs or any other American ice-cream brand, based largely on the fact that it has dramatically fewer calories (and consumers feel like they can eat the entire pint in one sitting). CircleUp realized that if the investors on its platform–who they speculate may have been relying on a gut reaction or a particular metric that had little empirical evidence as an indicator of success–weren’t always fully supporting promising brands, it should start a fund to also begin investing itself.

advertisement

Along with its algorithms, the platform relies on collecting far more data than other investors, both from private and public sources. “It’s painful, it’s hard, it’s messy, and it’s taken us years to pull together,” says Caldbeck. “But that also helps to make it proprietary. The technical complications of normalizing data and performing entity resolution across literally hundreds of sources is extraordinarily difficult.” Other investors typically rely on retail level sales data about consumer brands, which is only available for the largest brands. Roughly a year ago, when Halo Top’s sales exploded, Caldbeck says that 19 of the top 20 private equity firms reached out to him at the same time, because they were all looking at the same single source of retail sales data.

The platform only works with consumer brands, because that is where the data is strongest–consumer brands have similar business models, regardless of what they’re selling, and comparable data is available for everything from changing price points and SKUs to consumer reactions. The platform automatically categorizes products, so a nail polish is compared to nail polish, and kombucha to kombucha, and then uses its billions of data points to evaluate a company’s relative strength of brand, product quality, distribution, team, and financial performance.

The same can’t easily be done for tech companies. “In the technology space it is common to fund entrepreneurs that have little more than a 10- or 15-page PowerPoint deck,” he says. “Even at the Series A level, they often have little revenue. There are some other metrics, but they’re metrics that are kind of cherry-picked to tell a good story. There’s not consistent data to look at across hundreds, thousands, tens of thousands of companies.”

Because CircleUp also evaluates early-stage startups, it discovers companies that larger funds might automatically pass by, including those in parts of the country with less access to local venture capital. If a startup has less than $10 million in annual revenue, it might only raise $1 or $2 million; investors might not find it financially worth their time to fly across the country to visit and write what to them is a small check.

Along with identifying geographically diverse startups, and more women-led companies, the platform has also surfaced several startups with social impact missions. REBBL (roots, extracts, berries, barks, and leaves) makes healthy drinks and provides funding to fight human trafficking. Barnana makes snacks from bananas that would otherwise become food waste. Sustain makes natural, organic, fair-trade condoms, lube, and tampons.

CircleUp doesn’t specifically look for companies that have B Corp certification or female founders. But it thinks that its methods may help some of those companies get support earlier. “One of the things we believe to be true, we hypothesized early on, is if we are data-driven, that can lead toward less human bias, but also hopefully a greater percentage of underfunded entrepreneurs getting the capital and resources that they need,” says Caldbeck. “With the belief that if there is a bias against them in traditional venture capital and private equity, that we should help to eliminate that bias, or certainly help to limit it meaningfully. And we think that that’s been proven out.”

About the author

Adele Peters is a staff writer at Fast Company who focuses on solutions to some of the world's largest problems, from climate change to homelessness. Previously, she worked with GOOD, BioLite, and the Sustainable Products and Solutions program at UC Berkeley.

More