Meet the startup helping sites like Fab and Etsy court their customers

Imagine that from a first date, you could predict how frequently a new beau might call you, the kinds of places he’d want to go for dinner and what the warning signs of waning interest might be.
That doesn’t exist yet for courtship (to my knowledge at least), but New York-based Custora is trying to do something like it for commerce. With its predictive analytics software, the Y Combinator-backed startup, which launched last year, helps popular commerce sites, such as Fab, Etsy and Birchbox, get an early read on their customer relationships to figure out which ones will be the most valuable over time – and how they can keep those high-value customers happy.
“In general, online marketing is very visitor- and session-centric – they convert or they don’t,” said Corey Pierson, co-founder of Custora. “We’re really trying to expose that longer-term way of thinking about each and every customer.”
Pierson, who met his co-founder Jon Pospischil while getting an MBA at Wharton a few years ago, said the company grew out of a business school class on the lifetime value of customers and using behavioral data to forecast it.
On its own, Pierson said, thinking about long-term relationships with customers is “beautiful” because it isn’t so transactional. But he realized that the modeling technique used to calculate customer lifetime value (CLV) came with another positive side effect: it was simple enough to inform algorithms that could power a software-as-a-service product.
“If you had a hundred data scientists looking at every single variable in a business, you very likely could create a solution that is totally optimized for every single nook and cranny of that company,” said Pierson. “But with this probabilistic notion of marketing we can get really close to that ultimate solution and leaps and bounds above where they are today.”
Searching for patterns among purchasers
That means for reasonable monthly fee (about $200 to more than $1,000 a month, depending on the size of a company’s customer base), growing commerce startups like Fab and Birchbox (and a few dozens more) get to have data insights close to those generated by full teams of Ph.D. data scientists at bigger online retail sites like Amazon, Pierson said.
Based on so-called Bayesian probabilistic theories, Custora’s algorithms analyze the purchasing behavior of a retailer’s customer database to look for patterns among shoppers. When a new shopper enters the site and makes a purchase, the software analyzes information like where she came from, what she bought, how much she sent and more, and then matches her up against similar customers to generate insights.
For example, after just one purchase, Custora can tell that a given customer of a commerce site is on track to make six purchases, totaling $275, over the upcoming year, placing her among the top 5 percent ot his customer base. The program could also predict that half of her purchases will be art-related, 25 percent food-related, with the rest falling into other categories, Pierson said. As for warning signs, Custora can learn that if the customer doesn’t order an item for 3 consecutive months, they can assume that she has only a 10 percent chance of returning.
Even though the program analyzes each consumer’s individual behavior (anonymously), it gives clients an intuitive way to understand it in aggregate. Each dashboard screen answers a specific question with a design-sensitive, easy-to-understand infographic. An opening screen, for example, might show a client that there’s a $250,000 opportunity in targeting a certain percentage of customers who have only purchased once and then give recommendations for how to achieve it. Another screen might display how much a company should spend to acquire a customer.
No more ‘batch and blast’ emails
It also lets clients segment customers based on whether they are one-time purchasers, repeat purchasers, at-risk or lost. It then provides a list of recommended strategies for communicating with them, but doesn’t take what Pierson called a “batch and blast approach.” For example, instead of sending all at-risk users the same “haven’t seen you in a while” email message, he said the software looks at each person’s unique shopping habits and tailors messages specific to their circumstance.
For now, the program integrates with a client’s email provider so that it can directly monitor and analyze their email communications’ impact on consumer behavior, but Pierson said it could eventually integrate with other touch points, like social media and mobile applications.
While other companies, such as startups GoodData and SumAll or Google Analytics and Adobe’s Omniture, might compete against Custora in different parts of its analytics business, Pierson said no other company operates across as much of the consumer experience spectrum or provides the same level of CLV insights.
In a New York Times article last month, Custora got a shout out for helping Fab forecast that over the next two years iPad users would generate more than 25 percent of the company’s revenue. And as subscription commerce service Birchbox tries to get more subscribers to purchase full-price items from its online store, it said Custora helped uncover strategies for increasing profit per customer by more than 70 percent.
Earlier this year, the nearly 10-person company announced that it had raised $850,000 in funding from SV Angel, Founder Collective, Highline Venture Partners, Valhalla Partners, the Start Fund and other angel investors.