Kelly over at Hunch has done some investigative digging into one of their affiliate relationships that was mysteriously not converting. Domainr and Hunch share an affiliate-based business model, where both are search sites trying to help people find what they're looking for:
"We signed up to the Best Buy affiliate program through Commission Junction, and tested Best Buy links in several topics for which we already had a history of strong post-click conversions. While CJ’s reporting platform showed nearly the same ad impressions and affiliate clicks for BestBuy as we tracked internally, CJ reported not just a lower post-click conversion rate than we had historically seen, but actually zero. Several more days went by, clicks were still flowing to Best Buy, but they reported zero resulting sales."
"We decided to test this by buying some merchandise ourselves. Two of our employees, on two different days, followed Hunch affiliate links to Best Buy and then bought something. CJ still showed no sales activity. Houston, we have a problem."
Long story short, Kelly uncovered a huge snafu with Best Buy's affiliate service, and Hunch promptly discontinued their relationship with them. While we've yet to do this digging through Domainr's logs, we're curious what we'd uncover. And more broadly, Kelly highlights the inherent need for trust in the system:
"Bottom line: for cpa-driven affiliate relationships to work, there has to be trust and reliability in the system for accurate post-click reporting and payments. So it’s disheartening when even a big brand like Best Buy can have such a complete failure in their affiliate model, as we experienced above."
"One implication of all this: I would gladly pay a 3rd party to conduct “real transaction” audits on the sites with whom I have an affiliate relationship. I’m not talking about click matching, but actually buying a low-priced physical good from time to time (and perhaps then returning it later for a refund). This approach couldn’t be efficiently scaled to take representative frequent samples for a given affiliate, but still, at low volume it could potentially identify the most egregious offenders. Because as we found out with Best Buy, where there was smoke, there was fire."
We agree wholeheartedly. Thanks for posting about your research, Kelly!
