Increase in impressions reported by Google Analytics

[MargotLoco2]

Because Universal Analytics (and Google Webmaster) only stores SEO data for 3 months, I've been downloading SEO data (from the Acquisition tab of Analytics) to get a record of how impressions, clicks, CTR etc are changing in the long term (our business is seasonal, so these long-term patterns are important).

Today, I downloaded data for September, and found a very large increase in the number of impressions compared to previous months.

I looked back at the data for August, which I've already downloaded, and found that Analytics is now reporting much higher numbers of impressions than I have in my downloaded data. The total number of impressions has roughly doubled, and the increase for individual URLs varies, with some increasing by a factor of 10. The number of clicks has also increased, by about 15% in total. Because of the 3 month cut-off, I could only look back as far as the 11th of July, but the impressions for the end of July are also much higher than in my downloaded data.

I've noticed that Analytics has changed some other details in its reporting of SEO data. For example, the impressions and clicks data is no longer rounded. Could this increase in impressions be a result of those changes? Has anyone else experienced something similar?

We can go ahead and use the new data but it will throw our analysis off for past months (which have the lower numbers). If others have experienced something similar it would be good to know, so that we can adjust our historical numbers accordingly.

[Guest]

This post is deleted!

[jamoboggins]

It's not that impressions have increased month-on-month, it's that data from August (which we initially downloaded at the beginning of September) is now showing much higher impressions etc data than when we initially downloaded it. This throws into doubt all the previous data (which we now cannot access because Analytics only goes back 3 months).

[katemorris]

That data should be straight from Search Console when you link your GA account with Search Console. Can you compare your reports from GA in August with the same data in Search Console? Is that different? What about what you see in Search Console vs what you see on screen in GA? Let's start there.

[jamoboggins]

Hi Kate,

Many thanks for the response. Margot is away this week so I'm picking this up in her absence.

The August Impressions and Clicks data in Search Console is slightly different to the SEO data in GA (it appears to differ by up to ~8% in either direction), but appears generally consistent with the current data in GA.

The GA and Search Console data are both much higher (around 3 X) than what we have in the historical data we'd previously downloaded for August.

The August Impressions data we previously downloaded shows daily impressions, and each day is rounded to the nearest thousand (i.e. each daily number ends with 000). The new data in GA and Search Console appears to be no longer rounded at all. Surely this must be related.

Any further thoughts appreciated!

Thanks

Jamie

[katemorris]

Can you add screen shots of your report download settings? Is this report automatic? Is is from a GA dashboard or pulled by hand every month?

I suspect something might be going wrong with the report pulling from GA.

[jamoboggins]

We've been pulling the data from GA as follows:

Acquisition > Search Engine Optimisation > Landing Pages > Export CSV

We've not set up a dashboard so I guess it's "pulled by hand". I've checked and the number of impressions is the same (at least now) regardless of whether it's the Landing Pages or Queries data that is exported.

We followed exactly the same process when we first downloaded the data and so the data has definitely changed.

In the Search Console the data we've been using to cross-reference is in Search Traffic > Search Analytics.

It seems to me that we're unlikely to get a definitive answer on why it has changed and so we may need to simply start again with the past three months of data, and maybe set up a report so that we are 100% sure of the data export process. But any further advice would be gratefully received!

[katemorris]

Yeah, there had to be something off with the dates pulled or something like that. It's always possible that the data came out wrong but more than likely something was missed in the report pulling. Human error and all. I've done it so many times myself.

If I might recommend, if you have the resources, pull this data from Search Console directly, rather than GA. Using their API, you can pull it directly: https://developers.google.com/webmaster-tools/?hl=en

This might be helpful: http://searchwilderness.com/gwmt-data-python/

[jamoboggins]

Thanks for the further response.

However I think there has been a bit of confusion - we have already pulled the data directly from the search console by exporting the CSV.

So the discrepancy still remains, unless all of the historical data that we pulled (for every month back to April) was pulled incorrectly.

We are likely to automate the extraction of data in future to try and avoid any human error (thanks for the link, which will be helpful as we work out how to do this), but we're fairly sure that there wasn't human error this time. This is due to the fact that the data was previously rounded (in both GA and the search console) to the nearest thousand. When this rounding stopped all the impressions numbers jumped significantly, and that's the issue we are trying to get to the bottom of.

[katemorris]

Ooohh!!!! Yes, I did misunderstand. I think the discrepancy here is that the CSV download only downloads part of the total data, the first thousand rows to be exact.

For example, in an account I have access to right now, for the last 30 days Search Console shows 35,145 clicks and over a million impressions. The download shows, upon summation of the data, 404,923 impresssions and 20,309 clicks.

You can't use the download to use as an overall view. The API should give you more accurate numbers.

[jamoboggins]

Thanks, but the explanation still doesn't quite make sense because the discrepancy occurs for historical months in the downloaded data. So even if the CSV only downloads (for example) the top 1000 landing pages, it doesn't explain why the same download showed different data later. The top 1000 landing pages in that period should not have changed.

Anyway I think we will start using the API to extract the data in future as this seems to be more reliable regardless, so thanks for the help.

[katemorris]

I think the API will help, but for the same date range, no filters, etc, the data shouldn't have changed. BUT Google has been known to edit their Search Console data, or they have in the past when they found discrepancies. There are any number of reasons why, but I am sorry we couldn't nail it down for you. I really do think the API will help. Best of luck!