What You’ll Learn to Do: Optimize secondary metrics by pruning the “other” search teams from your SQR.

Where is the 70% Delta? Reporting Success.

What You Need to Know: When I defend my accounts, I don’t talk about CTR, Impression Share, or other secondary metrics. Instead, I’m hoping that our client is rational enough to accept statements like “We’re doing the best we can for your business, we’re the smartest SEMers you’re likely to find, and we’re crushing all prior baselines.” If that’s not enough for the client (along with their piles of newfound paid search money and incessant emails explaining what we’ve done that day), then we’ve gotta shift gears and give the client what he/she wants.

But let’s back up a bit before we address those secondary metrics.

If you read part one of this epic triology, you might remember I claimed that the Search Query Report was the ‘holy grail’ of adwords reports. To really have a great account, your mindset has to be in terms of “queries” and “keyword to query mappings,” not simply keywords themseves.

That being said, the Search Query Report is far from perfect. If you (or, more likely, your boss/CEO/check-signer) has an infatuation for secondary metrics, your work likely isn’t being presented in an appropriate light.

Why? Google’s query data is very incomplete. How much so? Well, I pulled some data from last month in one of my accounds and found that 60.2% of the time I had no search query information! That is, 60+% of my queries were “other search terms.”

What is an “other search term”? Google says:

These search terms resulted in impressions (and sometimes clicks) for your ads, but they aren’t listed individually in the “Search terms” table. Clicks and impressions for search terms are included in the “Other search terms” row if both of the following are true:

  • You didn’t receive any clicks on this search term within the last 30 days (from someone who wasn’t blocking their referrer URL).
  • The search term wasn’t entered by a significant number of people.

Also, clicks and impression data for search terms entered within the last 24 hours will always be in the “Other search terms row.” Check back later to see these in the search terms report.

Fair enough (I suppose), but  60% of queries fall in this bucket (ignoring the last condition, as all my data is at least 5 days old)? That’s a huge number! What’s more astounding is the performance differences between “other search terms” and “known search terms”:

CTR data

That’s right: if you “know” the search query, you can improve your CTR by 122,300%! So, how can an SEM be judged off of 100% of his/her portfolio when 60% of it is a mystery, and these “mystery” queries are awful, if you’re focusing on secondary metrics like CTR?

(An aside: in this account, we’ve certainly got more 1-token keywords than we probably need, and we have a lot of broad match roaming around. However, we also have very clean search query reports (i.e. almost all clicks are highly relevant) due to very aggressive broad and phrase match negatives to counterbalance our ‘loose’ positives. Anyway, back to the program.)

At this point, you might be saying, “Hey, Mike, not so fast – how can you give yourself credit for 1-click, 1-impression queries, but not include 0-click, 1-impression queries?” I agree, so let’s see what happens we look at queries with 500+ impressions:

impression share data

Still pretty massive deltas vs the rest of the “other” search terms! So what do we do with those troublesome buggers?

If you structure your PPC account in a way that parses out ‘Known Queries’ from “Unknown Queries” (download our whitepaper on Alpha/Beta structure if you’d like to learn the process), you can really starting putting secondary metrics to use. In that case, what appears to be good is actually good – and what appears to be bad is actually bad. You’ve reduced Google bundling, and your “unknown” is not affecting your “known.” So, at that point, bring it on – I’ll improve your CTR, your impression share, and any other “success metric” that Google might create. Oh, and by the way, you’ll probably make more money while doing it.

If you separate your “known” and your “unknown” queries, and the check-signer still insists that what matters is impression share, share of voice, and Click Through Rate, you shouldn’t be worried, but you’ve got to shift modes from “Do what’s best for the business” to “Buy only queries that are reportable!” This means “goodbye” to any and everything that’s producing these “other search terms.” This is when down is up, and up is down. How do you improve impression share in this scenario? You get rid of those strange, 1-token broad match keywords that bring ’em in. Now, they also bring in more conversions and ultimately more money, but you’ve made your stance on that already. How do you improve CTR? Kill low-CTR keywords. All of a sudden your PPC account is Alice in Wonderland!

This is the worst-case scenario – and this when Google’s sales pitch has worked (though, interestingly, trying to improve CTR and impression share here results in less money for them). I’m sure I could I cut out half of my “other search terms,” which would improve my IS and and CTR by roughly 30%. Doing this would result in just a few lost conversions, and I’m sure those could be made up elsewhere, so we’d still walk out net-positive according to check-signer/CEO/Boss’s criteria.

What You Should Get From All This:

If the success of your PPC account is being judged off of more than profit or some other equivalent goal, you need to structure your account in a way that you’re only responsible for what is “known.” If you’ve done that, and it’s still not acceptable, then get rid of the “unknown” as much as you can. You will get rid of many many many more impressions than you will conversions, so you’ll more than likely to live to tell the tale of how you survived  “optimizing secondary metrics”!

Mike Nelson, Senior SEM Manager

2 Comments

  1. Joey Muller March 7th, 2012

    Awesome post, Mike!

  2. Terry Whalen March 8th, 2012

    Jeeze – those are some tough clients you’re describing!

Leave a Comment