Skip to content

Archiving old Google Analytics data: what should we save? (2)

Yesterday I explained how we’re about to lose the website visitor data we had in the old Google ‘Universal’ Analytics …forever. So if historical data from the last few years is important to us, we need to download it now.

In my accounts, I can see data from September 2016 (the earliest they still provide) to August 2023 (when the old ‘Universal’ Analytics was switched off, at least for me). However, I’ve decided that for the reports I want, annual data is what I need, so I’ll be downloading data from 2017 to 2022 inclusive. I had the new ‘GA4’ analytics running from well before the start of 2023, so the data from that can take over then. Anyone who didn’t get GA4 running until 2023 (or later) may have some fiddling to do, but I hope they’ve got an overlap.

The first data set I want is simply overall visitor numbers. Sadly, there’s not a default report showing just visitors per month, but we can easily make one up, or we can just use a preset report on more granular data and combine the figures.

To make up a custom report, just go to Customisation > Custom Reports, then make a new ‘Flat Table’ report with “Year” or “Month of Year” as the only “Dimension”, and “Sessions”, “Page Views” and “Users” (or whatever we want) as the “Metrics”. Save the report, and it’ll appear. Now select the date range at the top, in my case from 1 Jan 2017 to 31 Dec 2022.

Then we need to ensure that all the rows in the table are included. Down at the bottom, it will say ‘Show Rows’. This needs to be adjusted to cover all of the table; in my case, it says ‘1 to 10 of 18’, so I set the ‘Show Rows’ to 25.

Finally we click ‘Export’ at the top, and in my case, I selected ‘CSV’, to get the data in its most accessible format. Opening the file in a spreadsheet looks like this:

To use a preset report to just get overall visitor trends, we can select any reports; whether we’re looking at where the visitors came from, the pages they viewed, or whatever, the totals should be the same. So let’s pick one which might contain some useful additional data: I’ve gone for the report on the type of device visitors were using (desktop, tablet or mobile), because I like to see the changes. This report can be found in the old Universal Analytics under Audience > Mobile > Overview.

The next step is to get the years broken down. This can be done by clicking ‘Secondary Dimension’ and typing in (and selecting) ‘Year’. Finally, I happen to have had a ‘conversion’ set up for visitors who spent some time on the site, so let’s add that in too, by selecting it under ‘Conversions’ at the top of the table. Then select the date range at the top, in my case from 1 Jan 2017 to 31 Dec 2022.

Now we click ‘Export’ at the top, again selecting ‘CSV’, to get the data in its most accessible format. Opening the file in a spreadsheet gives all the combinations of device type and year, e.g. the numbers of visitors on a tablet in 2018, etc.

It’s straightforward to see how the data can be pulled around and figures for the different devices can be added to give total visitors (regardless of device) for each year.

Now we repeat the exercise for any other data we want. In each case:

  1. Select the report;
  2. Add ‘Year’ (or ‘Month of Year’) as the secondary dimension;
  3. Ensure all the rows are shown;
  4. Export the data.

I’m saving the reports for ‘Channels’ (under ‘Acquisition > All Traffic’), ‘Source/Medium’ (under ‘Acquisition > All Traffic’), and ‘All Pages’ (under ‘Behaviour > Site Content’). There may well be others that are important though. Note that if the graph shows any freak data, this is our last chance to investigate it, and either make allowances or make notes. For example, many sites may have had a surge in traffic on just one day which can actually skew annual results, which will just hoover up that surge without any explanation.

Best of luck. Don’t put off this job.