We previously conducted this test to quantify the impact of IP address anonymization on Google Analytics geolocation accuracy. We explored this in order to understand how the potential change for privacy reasons could impact our data quality. This is especially relevant for geo-targeted advertising and personalisation. We’ve since revisited the data and revised the technique, the output from which is shown in this post.
After switching on IP anonymization in Google Analytics using GTM, we observed an average 174 km city-level discrepancy in the attributed geolocation. But don’t panic, this number is highly distorted. Most sessions were still attributed to the same country/continent after switching on IP anonymization. When it comes to the city level accuracy, 67% of the IP-anonymised visits were attributed to the same city, and 14% of the visits were attributed to the adjacent cities within 50 km (about the distance from London to Luton). Therefore, despite the average distance being 174 km apart, 81% of the visits have less than 50 km discrepancy (or no discrepancy at all).
In addition, it’s also worth noting that the number of sessions with ‘(not set)’ value in their city field (which means GA cannot identify which city is this visit coming from) increased by 21.65% after IP anonymization.
IP anonymization – Overall Discrepancy
IP anonymization – Discrepancy by Device
Mobile visits are generally more accurate comparing to desktop/tablet visits.
IP anonymization – Discrepancy by Continent
IP-anonymised visits from Americas (the continent) appear to have higher accuracy compared to Europe. On average, only 56% of the visits from Europe were attributed to the same city. Whereas for visits from Americas the average city level accuracy is 80%. Distance-wise, mis-located visits from Americas also have fewer discrepancy distance than Europe.
Visits sample size from Africa, Asia, and Oceania are not large enough to provide sufficient confidence for this assessment. Therefore please treat accuracy figures regarding those continents with caution.
Note:
The test was conducted on our site using GTM including 2267 sessions, from 2018-03-16 to 2018-04-15.
Majority of the visits comes from UK and US. Therefore figures regarding other countries are less accurate and may vary a lot.
Discrepancy distance mentioned in this assessment is the distance between cities. So it doesn’t go deeper than the city level (not which street/blocks the visitors actually are).