How does Google determine user location and how accurate is it?

Although Google’s advertising products will try to track users very granularly especially on mobile devices, for GA4 Google will mainly rely on users’ IP address. This means:

• It will be as accurate as IP locations generally are (so not very inside Australia itself as an example).
• Any method will be subject to people using VPNs, the use of which is likely to grow a lot because of censorship around the world.
• For your actual leads/customers, you are better off collecting proper location (at the level that makes sense for your org) as part of your website forms and using that data. However for anonymous users, it’s still very useful to compare performance and engagement of geographic regions as a rough guide.
• Sometimes Google won’t be able to map an IP address, in which case the location would show in reports as “(not set)”

Major dimensions to check out

• Country: This tends to be pretty accurate worldwide and is useful as a baseline. You might find you have overseas audiences you weren’t expecting, which can be an opportunity or possibly just a filter to add to your regular reports to make sure they focus on your actual target countries.
• Region: This would be the administrative region in your country so for Australia it would be the state, for China the provice etc.
• City: This is the most granular location and depends heavily on how well-mapped your country’s IP addresses are. For Australia, a lot of people’s ISPs would show you as being somewhere else (you can check out where websites think you are here) so take this with a grain of salt. You can also disable granular device tracking in your property which will stop Google’s estimate of user city. It’s probably a good idea to disable this for at least your non-target countries.
• Continent: This is pretty useless for most orgs, better to break this data down as per the next section.

Grouping your locations

A lot of orgs have their own location classifications (eg. sales regions) and it’s definitely worth breaking up your data this way. Universal Analytics let you do an upload of custom geographic data but unfortunately that’s not one of the options for GA4 so you would need to do it outside of GA4. The easiest place for this is in Looker Studio where you can use the CASE statement to create a custom column that turns the existing geographic columns into your own regions. You can also do this in BigQuery if you have the BigQuery import on. You could potentially do it in GA4 if you geolocate the user yourself, calculate the region and send it with your GA4 hits as a custom dimension but that’s a lot of work for the same result.

One more thing to beware of

Make sure your geographic data is not just a population density map. It’s not actionable or insightful to see that the top cities visiting your Australian website are Sydney and Melbourne because they’re just the biggest 2 cities. (You can check out this Facebook group for more examples). Now if your spread differed from the population this would be worth people knowing. But if your spread matches the population, you’re better off comparing conversion rates or website engagement but not total traffic.