• Neptune Moon

    Great post! Thanks for sharing actual figures too.

    I so badly want this report to be useful, but I have not yet found a way for it to really be truly useful. I know that there are always discrepancies between AdWords and Google Analytics, but I feel like adding Google Web Master Tools data into the mix only makes it that much harder to get accurate (or at least solidly reliable) data.

    Seems like you could cherry-pick the data source that makes you look best, which I think is seriously problematic. With the blackout of organic search terms in Analytics, I suppose we will have little choice but to at least consider this data going forward. That does not comfort me very much.

  • HarrisNeifield

    Hi Lucky,

    I agree the blackout in GA is not helpful regardless of the quality of other data sources. We’ll see if they can make GWT more accurate over time.


  • HarrisNeifield

    No problem Neptune, happy to share the data. I agree, it’s not the clear answer we want but it’s something. We do have to consider it at least.

    And yes I agree that someone could cherry-pick from this data, but marketers have always been able to find cherry-picked data to support the stance they want to take if they’re willing to do that. At SEER, we’ve got great clients who we build trust with and when we see questions about the data we’ll put it out there in a post like this and always present data in an accurate, fair and not misleading way.

    I see some potential uses of the paid and organic report but as noted above, there are some questions about the data and we do have to be careful with how we use it.

  • Martin Röttgerding

    When the paid+organic report (let’s call it POR) first came out we also dug into the data and found many inconsistencies. You could say the report combines data from SQR and WMT, but it’s actually a little more complicated. Here’s a translated summary of our findings:

    Comparing search query report with paid+organic, we found
    - the POR has a few queries that are missing in the SQR
    - the SQR has many queries that are missing from the POR
    - all queries with clicks where included in both
    - impression numbers where a little higher in the SQR (about 5%)

    Comparing WMT with POR:
    - WMT impression numbers are “rounded”, like 226 became 200, 435 became 320, 12 became 30, 8 became <10. POR numbers are much more accurate and precise.
    - In WMT you see similar queries combined and their numbers aggregated. This should lead to fewer queries shown in WMT. However, in our case we've found 50% more organic queries in WMT than in the POR.

    - Source (in German):

  • HarrisNeifield

    Thanks Martin. We found discrepancies at the query level as well, but more data in the SQR than the POR which matches what you found. It’s good to see that we’re matching up. We did not dive into impression data in this post because of what you noted – GWT impression numbers are rounded anyway.

    It’s good to see that data from another source matches what we’re seeing. Thanks for sharing the data!

  • Johann Colombano-Rut

    The thing is, the discrepancy between the number of clicks is not enough to rule on the usefullness of that report.
    If impressions are underestimated too in the same proportions as clicks,
    then it doesn’t impact the click-through rate much, and since it means
    there was in fact more clicks and impressions, the precision of the CTR
    is even a bit better !

    But in the case of an overestimated number of impressions, then it would be another matter entirely since it would impact the CTR way too much…

    I also found the same discrepancies in clicks as you, but personnally haven’t been able to measure the differences between impression numbers. So in my view, the jury’s still out !

    Do you think the organic search impressions given in the paid and organic report follow the same trend as the clicks ?

  • HarrisNeifield

    Hi Johann,

    I don’t recommend analyzing impressions, and therefore also CTR, at all as impressions are rounded by GWT. Impression data is unreliable. We stuck with clicks for this analysis.

    It’s good to hear that you’re seeing the same discrepancies in clicks as SEER data shows. Thanks for sharing!