Postal Code Data Now in Google Analytics

Google Analytics is now recording Post Code data for visitors to websites across Europe.

Rachel McCombie of Air Experiences was the first to spot the change.

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:

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:

Positives:

  • 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 Caveats:

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:

Thoughts?

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.

Do English People want Scotland to Stay Part of the UK?

Here are the results of 2 surveys. The first is one that’s been run quite a lot by polling organisations over the last couple of years. The second one is much rarer. Here are the descriptions:

  • Survey 1: A survey of 1,000 people in Scotland, carried out over the web, asking the simple quesiton “Should Scotland be an independent country?”
  • Survey 2: A survey of 500 people in England, again carried out over the web, asking the same question: “Should Scotland be an independent country?”

I thought it was strange the second question had not been asked more often, so thought I would run a poll myself.

Scottish Results

Here are the overall results of 1,000 people in Scotland being asked the question “Should Scotland be an independent country?”

As you can see – very, very close. So close in fact that – if you look at those ‘+/-4.6%’ error bars, it’s literally too close to call based on the 1,000 people surveyed.

(I’ve included error bars, so you can see the margin of error (and whether they are meaningful), and broken them down by age & gender too.)

Scottish Results by Age

Here are the results split by age, for the 2/3 of respondents where the age was known.

As you can see, the largest ‘Yes’ lean is among those 35-44; the largest ‘No’ lean is 18-24. This is interesting, but, as these are small groups of respondents I would not draw any conclusions based on this. (see the error bars, for example)

Scottish Results by Gender

Here are the Scottish results, split by gender:

Again, that’s leaning toward ‘Yes’  for male, and ‘No’ for female, but too close to call.

English Results

I ran exactly the same survey across 500 people in England – asking the question “Should Scotland be an independent country?” (worded exactly, I believe, as the official ballot question).

This time, the results were very, very different:

English Results by Age

Splitting this out by age, the results are very similar across all brackets (albeit note the error bars again here – these are very small samples in each group, so far from exact).

English Results by Gender

Again, splitting by gender we see a similar picture: English people do not want Scotland to be an independent country.

Caveats

It’s always worth caveating this kind of survey (in fact that’s true of almost all data). This is not an election. I did not ask “If you were to vote today, how would you vote when asked the question ‘Should Scotland be an independent country?'” – I simply asked the actual ballot question Scottish voters will be asked.

You’ll notice that I surveyed 500 people in England here, and 1000 in Scotland (note ‘in England’, ‘in Scotland’ rather than English/Scottish). The reason for that was I started by surveying 500 people in each. The “England” results were so conclusive I stopped. There was no clear winner in Scotland, so I ran for another 500 responses. Again, too close to call.

The final obvious caveat is: I haven’t surveyed Northern Ireland or Wales here. If you’d like me to do that, feel free to add a comment on the post. And – if you’d like me to survey more people in England if you feel doing so would alter the outcome – feel free to drop me a note too.

Overall Summary

The polls here are snapshots of 2 particular audiences. Based on those audiences, we can say:

  • The audience within Scotland are not sure whether they want Scotland to be part of the United Kingdom or not. The results are quite literally too close to call. (Voting one way or the other is another matter, where actual risk & proactive effort are both involved)
  • The English audience on the other hand are are very, very, very much swayed one way: they want Scotland to remain part of the UK.

Do share this with others if you think they may be interested.