I’m back to writing my own controversial thoughts this month; although, if you didn’t check out the interview I did last month with Andrew Goodman and Matt van Wagner, you should, those are smart guys.
Since I was thinking about PPC and how incredibly mixed up and difficult it has become, it seemed like a good time to write about it.
My team and I are trying to figure out ways to look at PPC campaigns differently, finding a way to automate some of the grunt work while still delivering great campaigns. I know everyone and their brother has done this, but we’re convinced we can do it better. Just let me live in this dream world, ok?
In considering the best ways to evaluate campaigns, we’ve begun a very thorough and detailed investigation of Quality Score.
While you can do a search on Quality Score and find plenty of people who have quoted Google’s Quality Score Guidelines, I haven’t seen that many articles about how it really works, and no one is willing to share their “secrets” to playing the game.
Here is what we’ve found, and while I can’t share the details publicly, it’s all backed by solid data and testing.
Google’s Assault On Keyword Data
First, let’s begin with a quick review of (not set) keyword data. This is often linked with (not provided), but it’s actually very different. Not set results occur when something is lost between the AdWords click and the tracking on the analytics side.
It occurs most often with auto-tagging, or if there’s duplicate code on pages or duplicate analytics profiles connected to AdWords. Not provided is returned when a user performed a search with https://www.google.com and clicked on an organic result.
In a nutshell:
The End Of Display URLs
No, display URLs aren’t going away. Google’s made a big change this year though that impacts what you can put on the end of your display URL. Forcing a subdomain (like www) onto each display URL cuts down your space by 4 characters. Here’s the catch: you can fudge this with three different techniques:
- If your URL plus your keyword is exactly 35 characters, Google will display it as you wrote it. For example: domainislong.com/keywrd-is-alsolong (note how it doesn’t have the www on it?). If your requested display URL is shorter than 35 characters, they’ll shoehorn a www. in front of it. Anyone who’s done a lot of AdWords testing knows that presenting a URL without www increases clickthrough rate.
- If your URL is longer than 35 characters by just a couple, they’ll adjust just enough to fit it. For example, the display URL in the screenshot below is 37 characters:
- If you have a lot of really long keywords, consider using keyword insertion to get around the character limits for them. If a display URL is longer than 35 characters, Google will shorten it automatically. This can help you get some additional real estate in extreme cases; but you’ll want to watch it carefully, since Google will decide how to shorten it, and it might create a branding issue for you.
Since a major component of Quality Score is the clickthrough rate of your ads, you’ll want to maximize every available opportunity to increase clickthrough, including the techniques described above.
Random Quality Scores Of 10
Everyone’s thrilled beyond words when their campaigns suddenly start showing QS of 10. But, take a closer look. We’re seeing many keywords with 10s that have never had an impression or a click. Never. In the history of the account. As soon as they are shown, the QS drops.
My question is: is there benefit to having some “empty” 10s in an ad group even if you never plan for them to be clicked on? Will this raise your average QS for the ad group, and therefore, your perceived value in the auction? We’re testing it, but we’d love to hear your feedback as well.
Pre-loaded Quality Scores
Another thing we’ve seen consistently in our tests is that our keywords and ad groups are receiving quality scores before they’re launched, which is remarkably consistent with what they end up being post-launch. More on this as it develops, but it would seem that you can tweak your campaign architecture, bids, keywords and other facets to maximize your opportunity before you even launch.
I’ve heard from other PPC experts that this is simply based on competitor data for the same keywords. We have a client with a completely unique service that is new to the market who saw QS in her account before we launched. And they weren’t all 10s, suggesting that it was at least partly based on campaign architecture.
Keyword Match Type Targeting
The final thing that we’re noticing has a significant positive impact on campaigns is matching keyword types within ad groups. By creating multiple copies of each ad group for all of the match types we want to target, we can maximize opportunity for impressions.
Add that to the pre-testing that we’ve been able to do with the pre-loaded quality score data, and our campaigns are really humming.
Bid Management & Why We Don’t Use It – At Least At First
The final key that we found is that bid management doesn’t work for newly launched campaigns. Owners of bid management programs will tell you it’s best to launch in the program and let the program “learn” the campaign.
While it’s true that the programs learn incredibly well, they have to have a good foundation to start from, or they’re learning the wrong things. We’ve found a significantly higher return on the campaigns that we launched and optimized manually at first.
It’s a ton of time and expense, but the per-keyword return is much higher in the long run if the bid management program is introduced after the campaign has been live for a few days or weeks (depending on volume).
The Future Of Quality Score?
As Quality Score gets more obtuse and confusing, we’re seeing a lot of bouncing around. For example, we had one keyword that didn’t get any impressions yesterday at all, but bounced quality score from a 10 to a 3 and back again several times.
Couple that with the recent changes in impression reporting, match types, display URLs and more, and I have to wonder… could we be seeing the first steps of phasing out QS altogether?
The last time I saw this kind of volatility in a metric from Google, it was back in 2007 when Toolbar PageRank was getting discredited and phased out. If you think about it, Google doesn’t really need to show us QS. We’ve gotten spoiled. They could just as easily decide not to show it anymore, even though they use it internally. What do you think is the future of Quality Score?
Opinions expressed in the article are those of the guest author and not necessarily Search Engine Land.