Swedish Government’s cookie opt-in banner has severe impact on their Google Analytics data collection


This is the data which Google Analytics collected, from the Swedish government’s website before, and after, they introduced a banner asking for visitors to accept cookies:

Impact on Google Analytics data of banner asking for opt-in - severe reduction in tracked visitors
With the introduction of the banner, at the end of June, 80-90% of their data disappears – just as we’ve seen in the UK, on the Information Commission’s website. Thanks to the Swedish authorities for sharing this data.

Why a banner?
Here’s what they say, on the banner:

Banner on Swedish government website Why does this matter?
The EU directive which this is based on is throwing the baby out with the bath water. In an attempt to protect our online privacy they have taken a crude approach to an issue which is more complicated than their directive recognizes. In this particular instance, the use of Google Analytics, there is not a privacy issue and the data is used to improve the website.

What’s the precedent for this?
In Sweden, right now, there is none. The PTS, the organisation responsible for making the directive a reality in Sweden, has nothing particularly specific (In Swedish); certainly nothing which directly requires a banner. Uppsala’s county website also sports a similar banner.

But there’s still some data being collected – all is not lost, right?
Wrong. One of the joys of the analytics data we can collect is that, from the start, we get a bucket of visitor data which represents everyone who has visited our site. Our job as analysts is to segment out different types of visitors and figure out how the site is performing for them. The opt in banner segments the data into people who accept cookies. I would hazard that the people who accept cookies are a rather unique demographic, who probably don’t represent most of your other visitors. In my humble opinion, your data is screwed from the start.

You could use this as an argument for the cookie directive. ‘See’ you can say, ‘no one wants cookies on their computer’. I would say that the lumping together of cookies as all universally bad has been lazy legislation; it does not reflect reality. Cookies which track visitor activity for improving the website are  a little different to those tracking your activity across the web. Without being able to gather data on what our visitors are doing on our site – which content works, which buttons get clicked, for example – we’re flying blind and the users’ experience will suffer.

I own a website  (in Sweden) – what should I do?
Check out the PTS regulations, they say  the following:

PTS guidance for website owners
In essence, ‘you don’t have to change your website right now’. Maybe while we wait, we’ll get a browser opt-in option which could be satisfactory for the EU, as  Peter Hustinex (European Data Protection Supervisor) hinted at in a recent presentation. It’ll be interesting to see what comes out from Google on this.

Here’s some additional reading:

E-Consultancy reflects on the ICO’s banner and the implications for future use of cookies.
Brian Clifton’s most recent post on the implications of the EU directive for Google Analytics
IAB (Sweden) – Recommendation on cookie use (In Swedish, but English translation available on their site).

About these ads

5 thoughts on “Swedish Government’s cookie opt-in banner has severe impact on their Google Analytics data collection

  1. The trouble is the GA co-opts the first party site to place the _utma cookie.so it is visible by Google and the first party site. It uniquely identifies a visitor and has a 2 year expiry. Google may say that it does not track behaviour with it, but it cannot be proven. If it only needed to detect unique visitors surely a session cookie, or one with a short expiry would be sufficient.

  2. I do not agree. This is a non-problem. What methods to gather usage/response/quality data did you use before posh tools like Google analytics came up? You did surveys. So – go back to doing surveys. In 99.9% cases those who really need the results can pay for them (usually it’s a commercial thing – at least when tuning a site/complex web system). I think the real problem is that some people take for granted the easily retrievable loads of data that we have now. It used to be harder to gather these numbers – but in the same process the gathered numbers had a completely different kind of quality to them. If you want to base assumptions on loads of data you need a set of tool that scientists still struggle to come up with. If you want to make your client happy – work hard and this directive will not be a problem. Because even with GA you need to work hard to really understand whats going on at the other end of the network. What i’m trying to say is that the devil likes statistics becuase he’s all in the details. Please also note that the directive only have impact in the public web. As soon as you users log on to anything, you have you development usage data as readily as in any other application.

    • Hi Dany,
      Thank you for your comment.

      Actually, I agree with your sentiment. Many website owners do not use usability studies or surveys – which is a shame, as these give insights to your website’s function in a way that nothing else can. However, if you want to gain the best possible insight into your website’s function then you also complement your usability and survey data with analytics data. (Incidentally, several survey tool vendors now offer an integration with web analytics tools). If you’ve got a well visited website, then the volumes of traffic, and therefore the robustness of your data, means that you can make data based assumptions on website performance very quickly. A challenge with usability testing and surveys is the long time they take to complete. If you’re managing a website, and reporting website performance to management then you’ll be reporting a series a key performance indicators; these could be, for example, keyword success, forms completed, campaign performance or engagement with key pages or content – without analytics data you’ll have no idea how these are performing and will not be able to make rapid changes to your website, or online campaigns.

      With regards to the quality of data, if you’ve got 10 000 visitors a day to your website and you’re doing A/B tests of link anchor text, for example, then you’re going to have a robust data set very quickly! To get quantitative data from a usability test is a challenge – you’ll get the low hanging fruit very quickly, but the more subtle stuff will be harder to achieve. Also, search engine optimization is going to suffer – how do you optimise your site when you don’t know how people are finding it?

      I also agree with you that the devil is in the detail – it’s easy to drown in the reports from your analytics tool. Sure, the numbers are exciting but are they are actionable? That’s why many of us work with reporting, and analysis – our job as web managers is to distill these reports in something useful, and actionable; give me a day with a web editor or marketing manager and I can show them how to do this.

      I entirely disagree with your assumption that if people need the data then they can pay for it. This will not be the reality for many organisations, both in the public and private sector. Cookie based analytics tools are effective, operate in real time and give answers to business questions – without this data many will struggle to match their website with customer needs. It’s too easy to make decisions on how you ‘think’ visitors behave on your website – analytics tools allow us to test ideas immediately, against a potentially huge data set.

      I think your comment about data collection from logged on visitors is an interesting one – but if the impact of this directive were to make more websites require people to log on before using services then it’s back fired somewhat in protecting online privacy and anonymity.

      Do you think it’s possible to manage a site in the dynamic way that analytics data usually supports, by only relying on survey or usability test data?

  3. Pingback: New recommendation on cookie use – browser settings rather than banners « University Usability

  4. Pingback: EU Cookie Law Sucks - But Good News - Page 4

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s