Slice Aims to Deliver a Uniquely Powerful Shopping Data Set

Slice, a company focused on shopping data, has just released an API that delivers its unique data set to partners. Its mission is twofold: First, Slice offers users a way to organize their purchases. To do this, it combs their inboxes for receipts and extracts information and meaning from them to create bundled information. Slice then makes that information available anonymously to third parties who can provide exceptional user experiences based on the data set.



I spoke with CEO Harpinder Singh about what is both a simple idea — mining shopping data from a user's inbox — and a powerful proposition for everyone involved. It's uniquely interesting because it gathers information across merchants, categories of merchandise and devices because all of this appears as receipts in your email inbox. Slice reaches back to receipts from 2006 and recognizes information from 2 million merchants, scooping purchases from Amazon, Fandango, eBay, airlines and countless more.

And it's not just online shopping that makes up the data — anything that becomes a receipt in your inbox, such as an airline ticket that you may have booked over the phone or a receipt for an event ticket like a concert, becomes part of your shopping picture.

While Slice can't access what you buy in cash unless you get a receipt emailed to you, it's apparent that receipts are proliferating in our inboxes from all sorts of transactions, both on- and offline. The day is fast approaching when not having a digital receipt will be an oddity for many of us. Because we get receipts for everything, as the digital versions crowd out the paper ones, the data set mined by Slice becomes ever richer, ever more comprehensive.

Immediately it's obvious that this creates a more complete picture of a consumer's shopping interests than any one merchant, even Amazon, could amass.

Slice shares an interesting characteristic with other enterprises that are on to something valuable: They create something that is both brilliant and simple to understand, but they also grasp the limits of what they have built and see the need for partners with different strengths. Singh decided early to stay focused on mining the inbox data set and leave the development of things like recommendation engines based on the data to others. That strategy spurs the need for partners — and an API Platform.

This delivery of data has allowed inaugural partners such as TheFind to create deeply personalized recommendations. It can use the API to do some very useful things. If you bought shoes for $80 and it sees that they now cost $60, you get a push notification to redeem the $20 difference.

Ads are an obvious application. If you search for red shoes, TheFind might know based on your history presented by Slice that it should show shoes of a certain type (say, boots) and in your size, providing a relevant answer not possible without that data.

And the data makes possible a wide application of affinity markers: Because someone buys organic food and fancy things, yet drives an older car, it might understand that advertising a Prius — or a Tesla — makes more sense than showing a gas guzzler.

This is a giant leap forward from, say, using data off credit card statements, which show which merchants you bought from but not what you bought.

Another early API partner is IFTTT, which creates little "recipes" to do things based on certain conditions. Users can build these recipes, or use ones others have built. For example, it can notify you if an apartment appears on Craigslist within a certain price range. That gives you the chance to React quickly and saves you from plowing through reams of Craigslist listings. IFTTT is integrating data from Slice to let consumers follow packages they have ordered from the warehouse to their doorstep.

I kept focusing on what the implications of such fine-grained data are for advertising. Singh ever so gently kept steering the conversation in other directions, suggesting the possible use cases are much wider. Example: Using the service, you can quickly see everything you spent on Amazon for the last year. This provides budgeting controls you might not have had before, or could only have had with a lot of painstaking data entry. You don't just see dollar amounts and dates; you can figure out how many books you bought, movies rented and so on. How about reviewing your budget for eating out?

As a last inaugural partner example and one that isn't necessarily about advertising, Singh pointed to Trov, which catalogs all the stuff you own so you can manage it. That creates opportunities for everything from insuring things and getting an extra warranty to understanding what its value is if you want to sell something. It can suggest that the iPad you once bought for $400 is now worth $200 on eBay. The company says it's the cloud for what you own. But Trov has a big problem: What person in his right mind wants to catalog everything? By using Slice's information from your email receipts, Trov can auto-generate a list from what you have bought. That's a big head start.

Singh finds that a typical inbox generates 150 to 200 relevant items.

Which circles back to why Slice is releasing its API: The possible use cases aren't all obvious, and Singh wants to harness the inventiveness of others to see where this can go.

What's next for the use cases of this data? That's the question that many, including Singh, are asking. If you're a developer, it might be a question answered by ... you. Developers can sign up for the Slice API now.

Be sure to read the next Data-as-a-Service article: API Now Features Named Entity Extraction Capabilities