Has The Attribution Pendulum Swung Too Far Away From Search?
Columnist David Rodnitzky is all for giving credit where it's due, but argues that search should be recognized for more upper-funnel activity.
Though I am a search engine marketer (SEM) at heart, I’m willing to admit that last-click attribution gives paid search too much credit.
Consider, for example, a brand search like “Nikon” that drives a purchase of an expensive camera. Few would doubt that the consumer typing in this term had spent hours researching, searching, and being influenced by Nikon marketing prior to his purchase.
At the end of the day, SEMs have to set aside their pride and do what is best for their overall marketing team. In other words, if multi-click, multi-channel attribution techniques take away credit from SEM and better allocate that credit across different channels, every channel will spend more efficiently, which is a win for the company overall.
Where I have some trouble, however, is when metrics appear to be applied to some channels but not to others. In particular, I have noticed upstream channels getting credit for metrics that are not applied to search. Let me give you two examples of this phenomenon.
View-Throughs For Display
First, consider display advertising. It is commonly held that the value of display advertising isn’t just purchases made from people who click on ads, but also purchases that come from people who viewed — but did not click on – an ad. By doing a hold-out test, advertisers can determine the “incrementality” of conversions that came from view-through consumers who eventually converted.
This seems like a reasonable proposition to me, firstly because the statistical method is used to ensure that the view-throughs are truly incremental, and secondly because it just makes sense that a banner ad could have an influence on a purchaser even if the purchaser never clicked the ad.
But here’s the rub: view-through incrementality is never applied to search ads. Granted, search ads are not as visual as banner ads (though one could argue that Google Shopping ads are certainly getting there), but isn’t it fair to assume that a certain percentage of people who view a search ad end up converting anyway?
Heck, I almost never click on SEM ads and will instead directly type in the URL of the advertiser (though this may be because I feel some “code of honor” where I don’t want to cost advertisers money).
Google currently doesn’t offer any view-through tracking functionality for search ads (they should!), but even without hard metrics available, shouldn’t we at least assume some incremental value of the billions of impressions that are served for search ads?
Micro-Conversions On Mobile
For my second example, let’s look at mobile advertising. Unless you are a marketing manager at a game app, you are probably pretty frustrated by the ROI you are getting on your mobile marketing at the moment. Compared to desktop advertising, mobile typically doesn’t perform well.
There are myriad reasons for this performance gap: consumers often use mobile for research and convert on a tablet or desktop; most mobile user experiences are lacking and make conversions difficult; and mobile tracking is a lot less dependable than cookie-tracking on desktops.
A lot of marketers have looked at the poor conversion performance on mobile and concluded that the only way to justify mobile spend is to measure conversions that, well, aren’t exactly conversions. This often includes things like email newsletter sign-ups, video views, click on the “directions to store” page, looking at a minimum number of products, hitting the “click to call” button, and so on.
These are all valuable “micro conversions,” and it makes sense to measure them. It certainly seems better than the alternative, which is sticking your head in the sand and hoping that mobile just goes away.
But again, what troubles me is that marketers are very busy coming up with non-revenue conversion events to measure on their mobile campaigns and often neglect to apply these metrics to their search campaigns.
Few e-commerce advertisers would consider an SEM campaign successful if all it did was bring in tons of product views and clicks to the store locator page, and I doubt that many give any credit at all to these metrics.
Different Metrics For Different Channels?
The argument some have made here is that different channels need to be measured with different metrics. Avinash Kaushik,
Google’s analytics evangelist, argues that trying to measure an upstream (read: not SEM) campaign by conversion is ridiculous because “In those stages, the audience is simply not ready to hear your love song. They are not ready for your pimpy attention or aggressive raising of their interest, or for your desire to convince them to be interested in your product.”
I get this argument – if Super Bowl ads were measured by the number of incremental cars sold in the 30 days after the Super Bowl, there probably wouldn’t be Super Bowl ads at all. And yet, we know that a successful TV campaign can have a material and positive impact on a business.
My beef here is simply that we don’t take these cool non-conversion metrics and apply them to SEM too. Heck, we may find that some keywords are incredibly powerful from a view-through incrementality perspective but not from a click perspective, and that others drive tons of in-store visits but no online sales. Or we may not.
Whatever we do, however, let’s just be fair. That’s all I’m asking. And then, after we’ve truly completed an apples-to-apples comparison of SEM and every other channel, if the SEM budget needs to be slashed in half, I’ll support it 100%!