Probably the most frustrating thing, for me, in talking to SEM practitioners about SEM Analytics, are situations where I feel like everybody is doing the wrong thing. People are most willing to hear about and listen to real analytics when they know there’s a controversy and they aren’t sure if they’re doing the right thing. But try to tell them that something they’ve always thought of (and heard) as best-practice is wrong-headed and you’ve got a man’s work cut out for you!
What fundamentally wrong-headed practice might I be referring to? As you may have guessed from the title – it’s Google Ad Rotation. It seems like everyone I talk to thinks using Google Ad Rotation and trying lots of different ads is the state-of-the-art “best practice” for serious buyers. In most cases, unfortunately, that just isn’t true.
Current wisdom suggests that to be effective in Search Engine Marketing you need to track conversions not clicks. What matters, after all, is not how many visitors arrive at your web site but how many depart with a sales receipt! But in stressing the importance of optimizing bidding and keywords by conversion, current wisdom has only managed half-the-story. If you aren’t optimizing ad rotations by conversion – then you are doing your SEM program a major disservice.
Here is the simple, unadulterated truth – Google ad rotations are based on click-thru not on conversions. So if you are letting Google control your rotations then you are optimizing on the wrong variable with all of the concomitant problems that came with optimizing keywords on clicks not conversions. Indeed, the problem is much worse– because advertisements are far more sensitive to under-qualifying leads than most keywords. The wrong ad can be a far more effective (and expensive) driver of unqualified leads than the wrong words.
Every direct-marketer knows that there is only one sure way to find out what advertising really works – test, test and…test. That’s what makes the Google Ad Rotation feature seem so great. You can produce two, three or ten ads for a given set of keywords and let Google automatically rotate them for you. As Google monitors click-thru rates, they will adjust the rotations of each ad. Pretty soon, your one or two most effective ads are getting all the impressions. Fantastic!
Except they may not be good ads at all. They may be world-class crumby ads. And the good ad may be getting virtually zero impressions. Can’t happen? It happens all the time.
Here are some results from one of our clients who had twelve ad groups. In six of the twelve cases – fully half-the-time - Google was optimizing for the wrong ad.
Google Ad Optimization versus Conversion Optimization AdGroup Ad CTR Conversion Rate % of time shown by Google A 3 0.49% 2.80% 35.46% 4 0.51% 1.72% 55.49% B 2 0.31% 16.67% 1.01% 4 0.43% 0.42% 57.17% C 3 0.05% 1.39% 46.04% 4 0.05% 3.39% 46.31% D 3 0.30% 0.00% 32.41% 4 0.29% 5.56% 33.37% E 2 0.15% 25.00% 4.71% 4 0.60% 2.82% 51.92% F 3 0.37% 2.67% 14.03% 4 0.44% 1.46% 49.52% 5 0.47% 2.52% 17.82%
Google Ad Optimization versus Conversion Optimization
% of time shown by Google
In AdGroup A, Ad Number 3 has a much better conversion rate than Ad Number 4. The advertiser would clearly like to see Ad Number 3 appear more frequently, but Google, looking at a tiny difference in Click-Through-Rate optimizes on Ad Number 4. In two of these cases (B & E) – the number of conversions is too small to be statistically significant (true, also of some of the cases where Google may be picking the right ad). However, in cases like A, C, D and F, it’s clear that the Click-thru winner (or co-champion in case C) is by no means the conversion winner. Simple multiplication reveals that in each case, you are paying significantly more and getting fewer conversions than with an alternate ad. This mis-optimization was happening fully half the time – suggesting that if you are relying on Google’s built-in Ad Rotations, you have a coin-flips chance of real optimization.
This isn’t going to change. Google, after all, can’t optimize for conversions. Many websites don’t track or even have conversions. And conversion is apples and oranges to click-thru. Besides, the whole point of ad rotation is to optimize their real-estate – not your conversion. Marketers jumped on the idea – thinking it optimized their conversion as well – but it just isn’t so.
In a way, Google’s Ad Rotation feature is an example of SEM Analytics gone wrong. It provides a powerful, easy and sophisticated way to optimize for the wrong thing. One of my partners used to love the saying, “A fool with a tool is still a fool.” And it’s true. No amount of technology, tools or sophistication can help you if you’re looking at the wrong thing!