Google Analytics is now recording Post Code data for visitors to websites across Europe.
Here’s an example for the UK, where over the last few days roughly 5% of ‘city’ traffic has been recorded with a postal code rather than a city name (the percentage varied by account):
(Note that this only contains the first portion of postcodes, which gives a smallish region, but not enough data to personally identify someone.)
The data is being recorded across other European countries too. For example here’s a snapshot of some postal code data being recorded in Germany – home of the toughest data protection laws in Europe. (Across a few accounts, German postal code data was being recorded for roughly 1.5% of all country sessions):
Similar data appears to now be flowing into accounts across many European countries, as pointed out by Benoît Perrotin:
@danbarker got postal code data showed up also on Italy, Portugal, Spain, Ireland, Belgium, Nederlands
— Benoît (@benoitperrotin) August 30, 2014
The data appears to have begun trickling in on August 27th, with a much greater flow on the 28th:
Positives & Negatives
There are some big positives of this, but also a few negatives:
- This is great for direct mailers, whose businesses are very much focused around postal regions.
- If it’s accurate, it allows you to judge response from particular regions.
- Allows you to attribute sales to catalogues that you previously may not.
- It’s probably good for charities, political parties, and other campaigners.
- Many of these businesses have a ‘local’ focus, for example political parties tailoring messaging by postal code, and using local volunteers.
- It’s good for any business with retail outlets. Rather than the arbitrary ‘city’ names, that often included small towns, postal codes are
- It also means you can match up your data more easily with other sources:
- Returns data for retailers.
- Third party demographic data.
- Population data, to understand your traffic in areas vs the actual size of the population.
The first caveat is, we do not know how this data is being collected, or why it seems only to cover a percentage of traffic.
The second caveat is, it’s unlikely that the data here would be completely accurate & comprehensive. That being the case, you’d only ever likely get a sample for any given area, and it would be difficult to tell whether those samples were evenly sized by region (eg. if I’m told I have 100 visits from postcode A and 200 from postcode B, does that mean I actually got double the number of visits from the latter, or just that fewer visits from the first area were correctly classified?)
The biggest caveat on all of this is that – at present – the data is being recorded in the ‘City’ field within Google Analytics. That makes things a bit of a mess: Some of the data is still recorded as city/town names, some is now recorded as postal codes. That means firstly that the data can’t be used with much confidence (eg. if you see a postal code, you have to ask “is that all of the data for that region, or is some of it grouped under a town name somewhere?)
Update: A final caveat is: Martin Macdonald spotted an oddity with an ‘ME14′ postcode appearing to be very popular. On further digging, the same postcodes seem to appear again & again among the top 5 for different accounts. Speculation (from @scottjlawson & @davecatley) is these may be large internet exchanges/providers.
How to Find the Data
The simplest way to reach the data is to navigate to:
- Audience > Geo > Location.
- Click the ‘City’ Primary Dimension (just to the top left of the main table listings)
- Use the search filter box, placing the following filter into it: “[0-9]” (including braces, excluding quotes). This essentially says ‘show me any results that contain any number’, which matches most postal codes across Europe (and of course excludes City names, which do not contain numbers)
Alternatively, the following Google Analytics Custom Report will give you a quick snapshot of Sessions listed by Postal Code and Country:
My hope is that this data will remain in Analytics. Ideally it would be in an additional dimension (‘Postal Code’) rather than being shoe-horned into the ‘City’ field, and would cover the US and other regions.
If you have any thoughts about any of this, do share them with me (@danbarker) on Twitter, or leave a comment below.