Last week, someone asked me”what do you see as the future of E-Commerce?”
A little taken aback at such a far reaching question, I gave a decent, if long-winded, answer that touched on web optimization, analytics, behavioral modeling, suggestive selling, and (I’m not kidding) Kanye West. It made sense, at least in my mind, but if I had it to do over again, I’d say something far more succinct: the future is in creating (and using) more data.
What do I mean by this?
Well, as an example, take this shoe:
Adidas Y-3 Yohji Yamamoto
According to a prominent apparel E-Commerce site, it has the following attributes:
- Adidas
- Y-3 white leather ‘Y-3 Boxing’ sneakers
- RETAIL VALUE: $230.00
- Color: White / Ivory / Silver
- Supple grain leather upper
- Round toe with silver leather detail
- Lace-up vamp with logo embroidered patent detail at tongue
- Grosgrain strips at sides with printed logo
- Leather lining with padded insole
- Rubber sole
- Leather / Rubber; Imported; style#303564601
And that’s it, which, luckily, is usually good enough. The basics, combined with the photo, are informative enough for me to make a buy decision, especially if I already know that I want this particular shoe.
But what if I didn’t? What else would need to be in the product database in order for this website to help me find this shoe in a search, or better yet, recommend this shoe to me? Is this a performance shoe? Is it a hip shoe? Does it work with a suit? Is it made with organic materials? Is it part of a Japanese specialty line? (As it happens, it is.) Is it rare? (Fairly.)
Without shoe-geeking too much, suffice it to say that there are lots of data points that might be relevant to certain customers which this listing doesn’t include.
And that, in my opinion, is the future of E-Commerce.
For a peek at what one might do in that data rich future, I submit that you need look no further than Pandora, the internet radio service. As I previously reported, I recently had the opportunity to attend a talk given by Pandora’s founder Tim Westergren regarding the service and its underlying engine (more on that here). I was amazed to learn that its ability to make recommendations rested on a database of four hundred “genes” or elements for each song; things like tempo, instrumentation, and key changes.
Four hundred! How many data points are stored for the product above? Twenty or thirty, at best?
While Pandora is not a pure E-Commerce play (they do receive commissions when listeners click a buy link and purchase from either iTunes or Amazon), their approach is on the vanguard of the data intensive predictive selling that I believe will replace today’s “search, find, buy” systems. 90% of Pandora’s catalog of 750,000 tracks are played each month by at least one of its millions of listeners, which makes their recommendation engine perhaps one of the best on the web. All that additional meaningful data is what allows Pandora’s suggestion engine to succeed at introducing its consumers to related products much better than today’s “listeners that liked that also liked this” engine ever could.
Coming soon: A new generation of powerful E-Commerce platforms that will harness richer product data to draw deep inferences and make recommendations that astound customers with their relevance.
(How we get that data will be the subject of a future post…)
