• http://www.pagezero.com Andrew Goodman

    Gak! It definitely is *not* intuitive to see 3.7 as the reported average ad position… when the bulk of the clicks in fact came in positions 1 & 2 !!

  • http://www.rimmkaufman.com George Michie

    Andrew, that was my reaction, too! I literally only had ~ 2 days worth of data when I wrote this and haven’t had a chance to study it more comprehensively yet. I also wonder whether our day=parting efforts are driving some of this. We bid more when the conversion rates are strongest and less at other times. I could see that making the average particularly unrepresentative of reality. Interesting nevertheless. We’ll do a more comprehensive study over at RKGBlog.

  • Alistair Dent

    George,

    Unless I’m mistaken this is the same data that has been available to users of Google Analytics for around a year now. This change makes that data available via a query string for other analytics packages.

    Ever since Google introduced the Top vs Side segmentation we’ve been able to see the vast distinctions in click-through rate that come from the banner positions. It’s fairly typical to see ads on the side get 0.2% to 1% CTR shoot up to 6%-15% CTR when appearing in the banner.

    These kinds of numbers mean that average position is becoming increasingly meaningless and difficult to analyse at all.

  • http://www.rimmkaufman.com George Michie

    Really interesting, Alistair, thanks for the insights. The top versus side placement may be the most interesting piece of the puzzle. I’ll be very interested to see if that distinction impacts conversion rates as well. I could believe that it is a different type of person who clicks the ads that are ‘between’ the user and the organic listings, from the person who consciously looks to the right to read ads. I think the key to changing the conversion dynamic (not the click through rate; the post-click behavior on the site) is whether the person reads the ad text fully or not, and I could believe the ads over the organic listings are read less carefully than those on the right.

    We shall see!

  • Matt Van Wagner

    This is great stuff, George, and glad to see Google is opening up more data for analysis.

    While Google is at it, here are some other data/changes that may be useful:

    - position reports that are numberic, not alpha-numeric. It would take less crunching to develop reports and analysis. That would, or course, totally disrupt avg position historical reporting, but as your analysis is showing, the current intuitions regarding avg position are not as useful or informative as we thought they were.

    - in conjuction with above, reports to show when 1s1 is only ad on page or 1t1 is only ad on page.

    – clicks and impressions by position.

  • Alistair Dent

    George,

    Anecdotally, I have seen in the first dozen accounts that I have checked (e-commerce, 40k+ visits from PPC per month) that conversion rate can be different between top and side.

    About 70% of ad groups have a significant (not statistically significant, just above 30%) difference in conversion rate between when those ads are in the banner versus the side positions. I haven’t checked if those average out to close to 0%, and I haven’t checked at keyword level or across multiple date ranges.

    This data is open to potential biases (e.g. best performing keywords already bid much higher so that’s a natural bias towards the top of the page having a better conversion rate when doing an anecdotal test like this) but there is enough evidence to make me believe there could realistically be a possibility.

  • http://www.rimmkaufman.com George Michie

    Good points, Matt. We’ll always support arguments for more data.

    Alistair, yes the analysis is very difficult because it really has to be done at the keyword level (same kw in different positions) and handling the sparse data challenge is a bear. We’ve developed a methodology for addressing this in the past, and having a cleaner view of the data should make the analysis more accurate as we had to rely on daily average position previously.

    Fun stuff!