BBVA NDB New Digital Businesses

NDB News

How Upturn uses data science to improve its product


Alec Macrae
Alec Macrae, Lead Data Scientist at Upturn Credit

As Upturn’s user base continues to grow, there’s an increasing need to better leverage the data it’s generating and use it to the startup’s advantage.

With that in mind the BBVA New Digital Businesses portfolio company decided to take the leap to hire a full-time data scientist, Alec Macrae, in February 2019. Since Macrae joined the startup, he’s been charged with interpreting data across a variety of metrics, from user acquisition and KYC to changes in users’ credit scores over time — particularly those who have used Upturn’s dispute function in an effort to improve their scores.

As part of these efforts, he’s helped build out a sophisticated system of dashboards. “We have more than 50 different dashboards now — there was only one set up when I got here,” he said. Data is being collected on more than 1 million (and growing!) credit-reporting accounts every month, and the team reviews the insights generated across those dashboards at least once monthly.

Employees at every level of the company have access to a dashboard featuring real-time data (scrubbed of personally identifiable information, of course). Offering democratic access to the numbers to anyone who’s curious or needs it allows Macrae and top brass to focus on bigger-picture initiatives, he said.

Mobile-first insights

Upturn’s decision to create a mobile app experience was influenced by data, particularly the discovery that a large percentage of the service’s users were accessing it from mobile devices.

“We were already planning to build an app, but that data has taken our thinking to the next level where we’re considering being an app-first company — and maybe an app-only company down the road,” Macrae said.

Data designed to help consumers

Another of Upturn’s big data initiatives involves building an engine using machine learning that aggregates credit-report data and other information to offer personalized recommendations that users may consider to improve their credit scores, from debt consolidation to lines of credit.

“A lot of people aren’t super knowledgeable about what affects their credit scores. We found that some users were actually disputing accounts that were helping their score, which would hurt them in the long run,” Macrae said.

In another data-driven decision, Upturn introduced warnings for users who were disputing accounts in good standing after discovering in the data that this was an issue that needed to be dealt with.

Macrae notes the sensitive nature of the data Upturn is collecting; it’s not lost on him that the tools the startup offers can have a profound effect on its users’ livelihoods. “We want to use data to help users help themselves,” he said, noting that Upturn is now collecting and digesting the right data to do just that.

Tools of the trade

The majority of Upturn’s data —found in its users’ credit reports — is sourced directly from the API it uses to connect to TransUnion. That data is pulled every 15 to 90 days, and the history remains associated with users for as long as they’re active with Upturn. “The fact that we have a user’s full credit history from the day they sign up for our service is another key differentiator for us,” he said.

Macrae also leverages data from Google Analytics and tracks prospective users’ progress through the funnel using a tool called Mixpanel. A separate tool called Segment aggregates data from each source and stitches it into digestible dashboards and reports.

Advice for other fintechs

While it’s easy to get lost in the tools and tactics of data science, the most important factor is whether data is being used strategically — and, in Upturn’s case, in service of its mission to help Americans understand, improve, and protect their credit.

It’s easy to fixate on a big data-based idea that might require a multi-year project to set in motion. “You could spend months creating something only to realize it’s not the right thing to do in the long run,” Macrae said.

Macrae’s advice to other fintechs — in fact, to businesses in all industries everywhere — when approaching data science: Start small and iterate. “Question everything constantly. Spend more time than you want just digging in before building out something,” he said. “You don’t know what you don’t know.

Visit Upturn at


More news

Upturn reaches 200,000 users: this is how they do it
Covault and the role of banks in the future of digital identification
Social physics: understanding human interactions through big data