Holy cow; these are some impressive numbers. Matt Marshall from Venture Beat reports that Aggregate Knowledge (AK) drove more than 20% of Overstock.com’s holiday sales. Matt estimates that AK might be responsible for $100M of Overstock’s $700-800M annual revenue. That’s 12.5% worth of sales on the low end. (Matt previously mentioned that Amazon generates 35% of its sales through auto-recommendations.)
Aggregate Knowledge is an ad platform, sort of. Just as Google displays your PPC ads to people who search for “web hosting” (or whatever keywords you’ve bid for), AK shows visitors who are already on your website that your hosting plans support blogs, CGI scripts, MySQL databases or whatever they’re looking for, as determined by their navigation behavior. Speaking of which, Collarity takes behavior targeting even farther by grouping visitors into affinity communities (new webmasters, developers, designers, for instance) based on their clickstreams.
If behavior targeting helps increase Amazon’s and Overstock.com’s sales, could it work for 1&1 as well? I remember having a conversation with CEO Andreas Gauger about diminishing returns on advertising: even an infinitely large marketing budget won’t generate an infinite number of new sign-ups, because there simply isn’t enough high quality ad inventory out there. Wouldn’t he like to squeeze 12.5%, 20% or even 35% more orders out of current website visitors without increasing incoming traffic? Wouldn’t you?
The problem is, most hosting providers’ websites (including 1&1′s, which I’ll use as an example) aren’t very conducive to behavior targeting. Visitors are usually presented with one or more service plans, which include loooong list of features. Reverse navigation generally isn’t possible. For example, I can’t start at the “MySQL” page and find out what my options are (1&1 Business includes 50 databases, and 1&1 Developer includes 100, etc). As a result, 1&1 loses the opportunity to find out what kind of hosting customer I am – *and* the ability to close the deal by letting me know that 87% of “people like me” chose its $9.99/month package.
In other words, it seems web hosting services are typically marketed with the assumption that customers differ only in their bandwidth/storage/monthly fee preferences. But based on my experience, website owners are a diverse crowd. Some are sophisticated developers, others don’t even know HTML. Some hope their websites will attract as much traffic as possible, others use their hosting accounts to store private documents.
Just think: if Amazon offered only a list of products with no behavior targeting, it’d collect 35% less revenue from the same website visitors. How much business could you be missing out on by not identifying and reaching out to your different audiences?
