Sometimes you really wish some Web sites would forget who you are. We have all had this happen to us, I think. Several years ago I got my mother-in-law a novel by a mystery writer who set her fictional world among knitting hobbyists. It was one of those lame efforts to be relevant in my gift giving. She is a knitter, likes mysteries, and I happened to hear mention of a series that combined both worlds. If I recall, she liked it well enough but didn’t go back for more in the series. Hey, I write about tracking technologies like recommendation engines. That doesn’t make me much of a recommendation engine myself.
Of course, for years after this, Amazon thought I was a lover of knitting and this particular novelist, so all of my recommendations from the retailer usually had some of these sorts of items in the mix.
‘Tis the season for recommendation engines to become mightily confused about the true tastes of online orderers. “People are struggling to figure out how to adapt personalization strategies,” says Scott Brave, co-founder and CTO of Baynote, which works with Dell, J. Crew, Anthropologie, BlueFly and many others in retail as well as in media. In the last year the company has been focusing more on the e-commerce vertical where personalization can mean a number of different things at different times.
Many sites are trying to make you feel at home, acknowledging you by name at the front door, and contouring the experience based on your previous tastes. Retailers have to be more adroit than that, he argues, especially at this time of the year when the most lucrative traffic comes from people who don’t know the merchant and vice versa. “One of the things we emphasize is that it’s important to personalize to intent,” he says. The challenge for retailers is to understand quickly from just a few cues what this visitor is after now. “When you have those signals and know how to interpret them, you can personalize to what people want right now,” says Brave.
Working with adaptive landing pages and especially using search keywords are critical. Brave says Baynote pays close attention to the ways in which people are searching in a site as well as the search terms that got them there in the first place. The company’s client Altrec uses an intent-based approach because its catalog of outdoor and ski wear attracts many gift givers. If a visitor uses “ski parka” in a Web search to find a merchant, that user clicks into a page that adapts to the referring search terms and is able to show recommendations of items that appealed to others who used those search terms. Once on the site, the tracking and recommendations take into account what sizes and styles a user seems to be gravitating toward.
When tracking a gift shopper for cues, things like gravitating to brightly colored clothing can be a signal for serving similar recommendations. “It is a matter of watching those signals, the language you are using, the places you leave footprints and where you look and linger – to build a portrait of what you are looking for,” he says. Making the right recommendations for the gift giver and newbie to a site doesn’t take too much evidence from that visitor so much as “having seen those signals from hundreds of thousands of others. But not it is about you.”
Brave says that in addition to listening for the shopping cues from gift-getters, there is untapped opportunity just in asking the user. “No one ever asks, ‘who are you shopping for?’” He says there are interesting possibilities for givers who somehow can bring the needs and tastes of their recipient with them.
It’s always nice when the person you’re shopping for posts her wish list at a merchant’s site, but where is the surprise in that? In the future, gift givers may have ways of leveraging their friends’ profiles at retailers. Perhaps instead of posting a wish list at Amazon (which I have, if my family is reading) why not let friends and family tap into the generally excellent recommendations Amazon serves me generously no matter where I engage them? We wouldn’t be posting lists but profiles, so our gift givers can use the recommendation engine at our favorite site to find the perfect gift for us -- the one we didn’t even know yet that we wanted.