My initial experiments with AdWords have revealed the importance, and capriciousness, of Google Adwords’ “Quality Score”. This number seems to have a great impact on the cost and success of one’s advertising campaign, but its provenance is obscure.
Quality Score
A keyword’s “Quality Score” is a number between 1 and 10 (10 is better) that attempts to quantify how happy someone clicking on an ad is likely to be when they arrive at the ad’s landing page. The lower the score, the more money you have to pay to get Google to display your ad.
Arbitrary?
The funny thing is, I’ve seen the Quality Scores assigned to a given keyword/page pair vary between 3 and 7 within a single day. (The number goes up as well as down; it doesn’t simply trend in one direction.)
Normally, this sort of thing would be only a minor annoyance, but the Quality Score is a big deal. It takes a $0.05 per-click bid to display an ad for a keyword with a quality score of 7, but a $1.00 per-click bid to display an ad for a keyword with a quality score of 3.
I’m not inclined to bid $1.00 per-click, so when Google drops my Quality Score, my campaign basically disappears. Now, I happen to think the low Quality Scores are bunk – my time-on-site for Google search traffic is over 2.5 minutes – but then I suppose I’m biased. In any event, I have to try to figure out how to raise the Quality Scores.
Useless
If you click on the “Details and Recommendations” link associated with a low quality score, Google’s only recommendation is to “Delete this keyword”. Google’s only explanations for the low quality score are:
Our system shows that users don’t find your ad or landing page relevant when searching on this keyword.
and
This keyword isn’t highly relevant.
Based on the keyword’s relevance to the associated ad text, CTR, historical keyword usage, and other performance factors.
Thanks, guys. That clears it up.
Solver
It’s fun to criticize, but I require a more practical course of action. One place to begin is with my solver; I suspect it may be better-regarded by Google’s algorithms than my puzzles, and it links to those puzzles. Therefore, I will create a new ad, with the troublesome keywords, linked to my solver. Yes, I am proposing to use a less relevant landing page in order to work around what I believe to be a flaw in Google Adwords’ quality ranking. If it works, it will be particularly amusing.
Design Tweaks
I also updated the markup on my main puzzle page today: I centralized the navigation elements underneath the playfield, eliminated the header, and used the resulting space to include a basic guide to the game. I also cleaned up some of the ancillary copy. This might – or might not – have an effect on the quality score.
Hypothesis
I do have a theory about (what I believe to be) the artificially low Quality Scores being assigned to my puzzle page. I do two slightly odd things, which might make the page look “spammy” to Google:
- To play a puzzle, you must provide a URL specifying the puzzle’s generation, size, and id. If the URL lacks any of these parameters, the play.php script will generate them for you, and redirect you to a complete URL. This system allows standard browser features, e.g. bookmarking, the back button, etc. to work properly. My ad’s landing URL, however, is incomplete. (The idea is that when someone clicks on the ad, they get a random puzzle.) I wonder if the immediate redirect is annoying Google.
- Every puzzle is displayed alongside links to other puzzles: one link for each size from 3×3 to 8×8. These links are complete (as discussed above) but randomly generated. This means that if you try to crawl the site, you’ll end up following a maze of links between different versions of play.php, each qualified with different GET parameters. I wonder if this, too, might not sit well with Google’s systems.
We’ll see how things go over the next several days; if my Quality Scores don’t improve, I plan to revisit these issues.
Today’s Stats
26 | Visitors |
34 | Visits |
227 | Pageviews |
6.68 | Pages/Visit |
5:43 | Avg. Time on Site |
Follow Along
You can subscribe to my RSS feed, if you’d like to follow along with this month’s project.