An interesting new study produced in cooperation with eBay and sporting the weighty title “Consumer Heterogeneity and Paid Search Effectiveness: A Large Scale Field Experiment,” raises some interesting new questions about search engine marketing (SEM) effectiveness. The study involved eBay actually suspending various types of SEM on specific search engines, and then closely monitoring the results. The first big finding was that, at least for large companies, brand advertising via SEM (e.g. eBay buying the keyword “eBay”) was a waste of money. While it’s always good to see hard proof, at the same time, this finding strikes me as intuitively obvious. I see it all the time with big companies with strong brands: their paid ads appears directly above their organic listings. Drawn to the bold type and high positioning, consumers will often click on the paid click, enriching the search engines and yielding the advertiser a click they would have gotten anyway for free.

The second (and more controversial) finding is that SEM for non-branded keywords (i.e., products or services as opposed to company names) is also largely ineffective, because it tends to draw in far more existing customers than new customers, making SEM more of an alternate form of navigation than a true discovery mechanism. Indeed, the authors of the study go so far as to say, “Bluntly, search advertising only works if the consumer has no idea that the firm has the desired product.” In short, SEM only really works when you bring a user new information, such as the fact you sell a specific product, and the user didn’t know that information in advance.

The study is careful to limit its conclusions to large companies with big brands. And this notion of having to provide new information for SEM to prove effective would seem to be a perfect opportunity for smaller and less-known companies. Or would it?

While far less scientific, I have spoken with dozens of small B2B data publishers who have tried SEM and pronounced it a waste of money. While SEM reliably yields traffic (great if you are advertising-based), it doesn’t seem to dependably yield purchases (not so great if you are subscription-based). Yes, I know B2B is different, and site visitors often don’t purchase on their first visit. But my informal survey was of small publishers, who though they often lack sophisticated analytics, generally have a very good seat-of-the-pants sense of where their business is coming from. And for these publishers, they believe their business isn’t coming from paid search.

There’s no definitive answer here, but it does suggest we all take a much harder look at paid search efficacy, and consider the possibility we may be paying handsomely for clicks we would have gotten for free anyway.



If you sell online subscriptions, you’ll want to consider attending our Subscription Site Summit in New York City on May 8-9. You’ll fill notebooks with good, actionable ideas and techniques from some of the most successful players in the business. You’ll also benefit from the intimate format of this event, designed to promote frank and open discussions. Take a look at the program and register today – attendance is strictly limited.