This is a content network campaign with 300 or so ad groups, with 4 identical ads in all ad groups. (I know about using different ads for different ad groups, but that's not the point of my question.)
The optimized serving is enabled in the campaign.
The new ad has a much lower CTR than the other ads. Yet, it started getting more and more impressions (and the % served).
Today, it looks like it's going to get served 90% of the time, while the three old ads won't get served much.
In the past, the ads with a higher CTR usually got served more frequently.
But in this case, it's different.
It looks like Google's prediction algo is wrong in this instance. (I think they use an algo to predict which creative would get a higher ctr and serve it more often until they get some historical data to overwrite the prediction.)
Has anyone run into a similar situation? Would calling G help or will they simply tell me to wait for a week/month/forever?
Is it a glitch or a normal behavior?
If you have it on Rotate, just change it to optimize and back again. I do remember years ago there was a glitch and that was the suggested fix. Seemed to work.
In your campaign settings, make sure you choose Rotate and not Optimize.
That's the thing. I want them optimized.
I used to use "rotate" but found that when I add some poorly performing ads to the ad groups, the overall number of clicks drops. My guess is because of even rotation, the poor ads simply waste the slop in the ad auction.
So I want to keep them optimized, not rotated.
In the past, optimized did exactly what it's supposed to do. Better ads got more exposure.
But this time, it's the opposite.
Otherwise, Google will choose the ad it thinks is the best more often and it's not always correct in my opinion.
Hold long do you think it would take for Google to figure out that the new ad shouldn't been getting such large share?
Of course, if one ad gets lower CTR, your group as a whole will be lower. The solution is to pause that ad and try something else. That's the whole point of testing, you want to compare apples to apples, compare the same number of impressions.
You are doing yourself a disservice by leaving that poorer CTR ad run more often. Obviously the optimizer is somehow confused in your case. Stop running it or do a better test by rotating the ads. If it still gets the lowest rate, stop running it. Don't wait until it "figures it out". It may never.
The ad's QS.
What if the new ad has a much higher QS (in relation to publisher pages, keywords, my landing page, whatever else), which makes it appear on more sites. And those sites happen to have lower overall CTR (for all advertisers).
If set to standard rotation, it would look like the ad simply has low CTR and should be deleted.
Yet, it could be the wrong decision in the long run.
I read it a few times that on content network, ad's QS is pretty much just CTR. But I'm having a hard time believing it. There might be other factors affecting it.
> If set to standard rotation, it would look like the ad simply has low CTR and should be deleted.
That's the purpose. Chances are, it wouldn't be the wrong decision. It might be for some sites but that's what those reports are there for. Run them, analyze them and optimize your campaign.
By the way, don't delete, pause. You may want to come back to that ad later and test under different conditions.
There are three factors in QS: keyword-ad-page relevancy, other landing page factors (load time is the major one) and CTR. CTR has by far the biggest weight. Assuming a perfect score (or close to perfect) on the first two, QS is in effect your CTR.
Then, it would be easy to see if one ad in an ad group gets low ctr because it happens to be shown on an unrelated site while other ads in the ad group aren't.
But as of right now, I don't see a way of doing it.
