Surprise! AdWords Ad-Serving Myth-Busting

SMS Text

I feel kinda stupid. Either I didn’t read the AdWords Help on Ad Serving settings closely enough (again), or AdWords made something super non-intuitive (AGAIN). More likely, a little of both. I’m sure someone in the comments will tell me they already knew this LOL 🙂

When you create an AdWords campaign, you can edit the Ad Serving setting to rotate or optimize the ad distribution.


I thought that meant I could control whether the ads were given equal impressions. Not true, it turns out.

The insight given to me by one of our dedicated AdWords reps today is this:

Even when ad serving is set to Rotate, Ad Rank determines number of impressions.

Ad Rank can vary from ad to ad, because

  • Different ads cause different CTR with the same keywords,
  • Quality Score (QS) is largely determined by CTR, and
  • Ad Rank is the product of Bid * QS…

…so every ad has a different Ad Rank.

The Factors That Determine The Outcome of Auctions


I had already watched Chief Nerd Hal Varian’s video (Just kidding, Hal! You are the coolest nerd at Google, and nerds are awesome) on the ad auction twice. I guess I should have watched it thrice. Now I have:

Introduction to the Google Ad Auction

Still, he leaves out the ad auction -> impression relationship in this video.

The Rotate Ad Serving Setting: Myth and Truth

To clarify, here’s a chart of the misconception some of you may have shared with me, compared to the actual, real, true truth.


All of this happens when you notice that your AdWords ads are getting different Served percentages. Click that link to read more from AdWords help about it- you want to read part I.

Ad Serving Setting and Impression Share

  • If you have it set to optimize, the better Ad Rank show will show more often, and you’ll win more auctions and get more impressions.
  • Even your campaign is set to rotate ads, the overall lower Ad Rank can give you fewer impressions. Using the rotate setting can lower impression share.
  • An ad with a lower quality score would get fewer impressions because it would show off the first page more often. Even when average position is fairly high (1-4), that’s still an average; sometimes the ad could be showing on the second page or lower.

Hmm… Let’s Read This More Closely


Ha now I see it says “more evenly”. That makes more sense now. I feel truthier. I hope you feel more truthier, too!

Brian Carter
Brian is author of The Like Economy: How Businesses Make Money With Facebook and Facebook Marketing: Leveraging Facebook's Features For Your Marketing Campaigns, How to... Read Full Bio
Brian Carter
Get the latest news from Search Engine Journal!
We value your privacy! See our policy here.
  • Tag44

    Hey Brian, thanks for the post and for sharing such useful information with us.

  • Plien

    Thx for the info! One question: Why set the ads on rotate then? I thought the point was setting the ads on rotate (for a while) was to find out wich ads performed better. Now it seems not to matter?

  • Mouli Cohen

    I’m hardly a veteran advertiser, but it’s hardly surprising that Google wants to show the ads that get the most clicks.

  • Brian Carter

    Plien, if you want to test new ads, and you should, to find the best CR and Avg Sale, you’ll want to put ad serving on “rotate”- you’ll get MORE impressions of the new ad that way than if you put it on “optimize”. Point is, rotate gives you equal auction chances and more impressions than optimize.

  • Aussiewebmaster

    This has to have been a newer way they do this… in the past I could equally spread ad delivery and it would be accurate within a percentage point or two

  • Brian Carter

    It might have always been Ad Rank based. All they’d have to do to get more clicks is weight the Ad Rank calculation more on quality score, or remove any bias toward helping newer ads.

  • brad aka eWhisper


    This is not new. Rotate has never given equal impressions. Usually the percentages are much closer than optimize, but it’s never been exactly the same impressions. If you look back through campaigns that are years old, you won’t see any ads with the same impressions.

    I’ve also heard from the high ups at Google that there is a bug in the system that contributes to this served percentage.

    However, later this year (hopefully, the screenshots are pretty cool) there will be a new way of setting impression percentage by ad instead of relying on the current settings.

  • Brian Carter

    Brad, I did not say it was new. Aussiewebmaster suggested it was.

  • brad aka eWhisper

    Actually, I misread something you said.

    Either way, it’s a good reminder that not everything is as it appears in AdWords and you often have to look deeper for the answer.

  • Brian Carter

    THAT i COMPLETELY agree with, LOL. Even people who are good readers can’t just read all the help screens and expect to get everything. Just checking the index for Andrew Goodman’s second edition book and it’s not clarified there either. Eventually, we might need some comprehensive training courses that are university level- there’s a lot out there but nothing that teaches you EVERYTHING, is there? Such as it is, it takes a lot of study and years of practice.

  • Marcos Nobre

    I’ve always hated the fact that CTR is kinda black box because Gevil has never disclosed real “kwXad” CTR. Though I’ve never realized they’ve being screwing impression share because of that!
    I’m quite sure there isn’t any mention about that in impression share help files. I’ll check that out.

  • Marcos Nobre

    Let’s see (urls blocked):
    What are ways to improve my Impression Share?
    Nothing about ad serving.

    How does Impression Share reflect on my campaign’s performance?

    How is my daily budget distributed?
    Nope but Oh there it is a helpful article:
    Which ad delivery option is best for my campaign?
    Oh not that helpful…that’s sad.

    Better check again ClickEquation Craig Danuloff’s amazing articles about impression share .

    Thanks a lot Brian!
    Most important PPC article this year so far!

  • Harrison Schmidt

    Good post.

    To improve your clickthrough rate, and therefore quality score, you should set your ad headline text to include the keyword you are bidding on.

  • David Rothwell |

    Last post is right on the money as a reminder – always include keywords somewhere in the ad, and test in headlines first.

    I could not disagree with the logic of this post, and it’s useful to know (and not unsurprising), but with rotation ad impressions are pretty closely shared with no large bias one way.

  • trebuchet

    Glad to see someone’s bringing light to this, it’s been going on for years.

  • Andrew Goodman

    Truthier, indeed. The ‘more equally’ verbiage has been in place for about two years now. I remember my rueful grin when I stumbled across that rhetorical hedge.

  • Steve S.

    Does dynamic keyword insertion affect ad quality score and thus impression percentage?

    I’m wondering because I’ve got two virtually identical ads running as A/B, with the same set of keywords (QS on those is almost all 10/10). The only difference is that I’m using DKI in one of the ads’ headline. For the non-DKI ad, I’ve got impression share of 95.45%; for the DKI ad I’ve got 4.55%. AND… the non-DKI ad has a 1.48% CTR, versus 3.03% CTR for the DKI ad. (I’m viewing stats for just the past 7 days here, and ad rotation is optimized.)

    So, double the CTR for the DKI ad, but drastically less impression share… which is exactly the reverse of what should be happening, according to Google.

    Only thing I can figure is that there’s some unseen ad quality score going on based maybe on the fact that the DKI ad has only been running for the past two weeks, versus a much longer period of time for the non-DKI ad. (Again, I’m narrowing down the scope of the stats to just the past 7 days, though.) Would the newness of the DKI ad version have an effect on the ad score, and thus somehow be skewing the ad impressions it gets?

    (And re: “ad quality score”… I keep seeing references to this term, but it seems the only real quality score that you can see in AdWords is _keyword_ quality score. There is no viewable _ad_ quality score, except for ads getting impressions on the content network. And since I just today turned on content network for this campaign, there is no existing history for content network yet.)

    Confusing enough, huh? FYI, I’m just coming in on this ad campaign (new job I just started a few weeks ago), so it’s a bit of a muddle, but this impression percentage thing is puzzling. As Brian said, “If you have it set to optimize, the better Ad Rank show will show more often, and you’ll win more auctions and get more impressions.” Which isn’t happening — it’s exactly the opposite. And, “An ad with a lower quality score would get fewer impressions because it would show off the first page more often.” Which makes me wonder how DKI would impact this… although average position for these ads is 2.8.

    Any ideas much appreciated…

    Steve S.

  • Miguel Gomez


    I had been wondering about this for a while. I was going back and forth in my mind about which way it was working. Thanks for clearing that up. Nice writeup.