It’s disconcerting when you consider just how much your bank or credit card company can see without even really trying: everywhere you shop, eat, and play–right down to how much you spend and when. (Suddenly, even Uber’s God View doesn’t seem so scary.)
Capital One is developing an app called Ideas–an optional stand-alone app from their main one–that mines customers’ spending histories to offer them relevant deals and events (for which Capital One takes no cut). Each day, it produces a short, personalized list of coupons (like save 10% at J. Crew) and things to do (like check out The Book of Mormon), all translated to a short, image-forward list you swipe through, kind of like Tinder. If a customer likes an event, she can save it to be reminded later. If a customer likes a deal, he can virtually clip the coupon. And if that coupon goes unspent, then shortly before it expires, Ideas will SMS the customer to warn him about it.
“You have all of this information about how to spend your time, how you shop,” explains Paul Sun, head of Capital One’s D3 innovation team, a small, mostly independent crew of developers within the company. “Everyone’s trying to market ‘how can we declutter your life, and only provide you the relevant info you want.'” With its Ideas, Capital One is attempting to solve the issue of information overload by providing the right information.
The truth is, of course, that generally you and I, the consumers, don’t really have our own shopping history prechewed for us in a meaningful format. But our banks do. Because Capital One can literally see how many times a customer shopped at a store like Banana Republic in the past year–assuming that person used her credit card–it’s relatively easy for the bank to make the connection that if there’s a good Banana Republic coupon, or even a maybe a coupon at a similar brand like Gap, that offer should be added to your short daily list.
Of course, they may be wrong about that. Maybe you just bought something for someone you hate there. For that case, Capital One has designed the system with machine learning at its core, so the software can learn from its mistakes. You can tap a heart when you like an idea, and tap a broken heart when you don’t. Over time, your suggestions should get better, and everyone else’s suggestions will improve from the macro trend data, too.
The app features a lightbulb button front and center in the interface, which a customer can tap for an explanation on the reasoning behind the offer–something along the lines of “you shopped here seven times last year”–though during the short demo of the beta app I saw, the reasonings were more general, like, “because you’re a movie buff.” Creepy? A bit! But also useful and powerful to know. This is where Capital One needs to deal with the uncanny valley of AI. As they dive into our data, they need to do so with just the right language, or we’ll just question why we gave them access to our personal life in the first place. (Though it’s worth it for all of us remember that we have all given our banks and credit card companies access to this information!)
“This was one of the learnings from our program. Sometimes companies try to give very refined recommendations, but consumers won’t always get it. They see something and say, ‘Why are you showing me this?” explains Ketan Babaria, head of consumer products for the D3 innovation team. “So if you tell them, this builds trust, and they think it’s more relevant in general.”
Capital One has no plans for directly monetizing the Ideas app. There is no referral fee or kickback on these events and coupons, and the data isn’t shared beyond the walls of Capital One itself. Instead, the company takes its standard retail cut of any resulting transactions, and hopefully builds some good will for suggesting the ideas in the first place.
Ideas is currently in a closed beta with select customers. The team plans to bring it live to the App Store later this year.