Engagement seems dodgy

Microsoft recently announced support for an engagement metric in their Atlas ad server. Engagement is based on the theory that it takes a viewer a while to ‘engage’ with your brand, so the first banner ad isn’t likely to work. Instead, you need to show your users multiple impressions, preferably multiple messages through multiple mediums, and understand that in the end, when the user types your product’s name into a search engine and clicks on a text link (paid or unpaid), all the previous advertisements played a role in the conversion process.

The cynic in me isn’t surprised that Microsoft’s got a long line of agencies waiting to test this stuff, since it justifies spending more money on ads with crap clickthrough and conversion rates even when the real purchases are coming from a keyword buy on a competitor’s search engine. But the basic argument – some people see an ad and then decide to act later – seems sound. I just don’t know what proportion of users do this, vs. the ones that just didn’t notice the earlier ads. After all, if you know a user’s browser loaded four of your banners without results, but then the user clicked on a fifth, you still don’t know much. The user might have become increasingly engaged with your ads – or he might just not have noticed the first four. Hell, maybe he meant to click the ‘audio off’ box on your annoying singing ringtone ad and his finger slipped. How many ‘oblivious until the nth impression’ users are there for every one that actually becomes more engaged with your brand over time? Given the state of banner blindness on the Internet, I’m guessing a lot.

Above all, I dislike the engagement metric’s ability to excuse mediocrity. If conversions see eleven banners on average before converting, don’t try to tell me that this is ‘engagement’ and my low performing campaigns are just part of the ‘conversion funnel’ – instead, get out there, optimize ads on the traditional metrics of clickthroughs and conversions, and see if you can further improve my bottom line. You don’t need ‘engagement’ to do that – you should be doing it anyway.

I’m going to keep reading up on engagement as a metric – it doesn’t pass my initial smell test, but I’m not knowledgeable enough to be able to authoritatively say there’s nothing to it. If you feel I’m missing something and have some good reference material for me, please guide me to it in the comments.