I’ve been asked more times than I can remember, “How do you calculate the real average position?” The good news is that everyone asking realizes that taking an average of an average will get you in trouble. The bad news is they still don’t know how to actually calculate the true average position. In this post, I’m going to run through the methodology and a dummy example in Excel.
Let’s start by looking at the dummy data.
We have date, campaign, impressions, and average position. The end goal could be to calculate average position by date or by campaign.
The reason we can’t just take an average of average position, basically, is the differences in impression volume. We need to account for the impression volume by calculating a weighted average instead of an average of an average.
The first step is to multiply average position and impressions in the raw data, creating a metric I usually label “Avg Pos Index”. Fill this formula down for all rows.
We can now make a pivot table off of this data and create a calculated field for our true average position. It’s simply the sum of the average position index divided by the sum of impressions.
And that’s it! Now we can easily see the true average position by date or campaign just by changing the layout of the pivot table.
A couple of parting thoughts:
- I ran through this example in Excel, but the same methodology will work in Tableau, SQL, or whatever tool you are using.
- PLA’s will wreck your average position calculation. You obviously don’t want to include them as PLA’s don’t have a position in the same sense as regular paid search ads, but they are typically still included in search reports that pull average position. Just make sure you remove PLA’s by filtering them out from your average position pivot/dashboard.
Interested in working with us? See our open analytics roles here.