Google Analytics, Lexbase and Personal Information: A good bad example

Lexbase have been in the news a lot. Controversially, their online search tool allows you to find people who have, or have had, convictions in your local area. It was down for several months – after intense criticism – and is now up again. Its most recent incarnation boasts a new interface and has provoked the same discussion; that a criminal record can allegedly be shown entirely out of context or, in some cases, incorrectly. Even if this is all publicly available information it’s something which is troubling for many, not least Beatrice Ask; the Swedish Justice Minister.

Not only is it an example of a contentious interpretation of Swedish law surrounding ‘yttrandefrihet’ it’s also a great example of what not to do when you install Google Analytics on your site.

The problem is this – run a search on Lexbase using someone’s name, personal number and a geographical area and all this data is sent to Google Analytics. This entirely breaks the terms and conditions of using Google Analytics which explicitly state that you cannot do this. The documentation includes the following:

“…prohibits sending personally identifiable information (PII) to Google Analytics (such as names, social security numbers, email addresses, or any similar data)…..”

Here’s what that looks like – you can see my test search terms being sent as an event, including a personal number and name:

Sceenshot from GA Debugger showing PII being passed on to Google Analytics

Lexbase’s use is an extreme example but sending PII via Google Analytics is something I see happen with websites  on a fairly regular basis, usually accidentally or in ignorance of the terms and conditions. Not only does doing so probably betray your website’s policy statement on your website, and therefore your customer’s trust, but it also risks your Google Account being terminated by Google.

Counting Zeros – Basic Google Analytics Troubleshooting

I’ve been spending more time in the Google Analytics Product Forums – it’s a great way of staying sharp, many of the questions and issues are similar to the ones my clients face, and it’s nice to be able to contribute to the community there.

One of the most common questions I tend to see there is something along the lines of ‘Why can’t I see any data’. Here’s some tips to get started, if you’re staring at a bunch of zeros:

1. Is your tracking code installed correctly?

Check to see that the Google Analytics tracking code (GATC) is actually on the page and deployed throughout your site. Use a tool like Screaming Frog to scan your site to verify that you have a near as damn it 100% deployment of code.

The Analytics Helper extension for Chrome is a quick way of seeing if there is GATC code on the page.

If you see code, then check to see whether it’s firing – I use Google Analytics Debugger all the time. That will tell you exactly what’s being fired off. Top tip – the Debugger tool uses the Developer Console to display results; when it’s open right click on the console and select ‘Preserve Navigation’ – that way you can look back over a series of calls to GA.

2. Do your UA-XXXXXXXX-X numbers match?

Pretty basic this one, but check that the GATC UA number on the website matches that of the profile your’re looking at. That’s the number which appears in your GATC on your website – compare it with the UA number beside the profile name in the main accounts list in your GA interface.

3. What filters are in place?

If you suddenly see data flatline, but the code’s still on the page and firing, then check out any recent filters you’ve configured. A poorly constructed regular expression, or selecting ‘exclude’ rather than ‘include’ can kill your data. I recommend always have a test profile where you can experiment with new filters before launching them on your main reporting profile – remember, if you break your data, you can’t go back and restore it.

4. Have you been patient?

If you’ve just installed the code then take a deep breath and wait – in the support documentation Google says it can take up to 24 hours. It’s usually faster than that, but there can still be a time lag. Check the Real Time reports – they should show data immediately (assuming that the Tracking Status Information says ‘Tracking Installed’).

If you’re expecting a low volume of visitors anyway, then there may simply be no visits to track – assuming you’ve filtered out your own visits on your main reporting profile use your test profile, or another unfiltered profile, to check that your own visitor data makes sense.


Want to learn more? You’re welcome to talk to me about GA, we can also meet up at the forthcoming Digital Knowledge Day 2013 – Stockholm 10th April, Sweden.

Digital Knowledge Day 2013: Don’t work in silos

First, I need to be upfront and say this is a shameless plug for a conference which Search Integration, the awesome consultancy I work for, is hosting in April.

So, with that bias declared, I can go on to say that this is not a conference to miss; especially as the theme ‘Don’t work in silos’ is still, and so, relevant in digital marketing today. This resonates with me, as I’ve been in too many projects and jobs where, despite a underlying acknowledge of the importance of the web, there is still a lack of a coordinated approach in presenting a company digitally. As my colleague Christofer Brugge writes on our blog, it can be as simple – though as critical – as colleagues not talking to each other, or not having an adequate awareness of each other’s activities. Digital Knowledge Day 2013

In a previous job I worked at a university where the system for presenting the education we offered (essentially our product) was ring-fenced by a blinding complexity of multiple databases, restricted windows of opportunity for developing content and a general disconnect with the marketing efforts of the university. This was a ‘silo’ effect – however much the website was developed the database content sat in the background driving the customer experience no matter how much it was dressed up. This effect rippled through the organisation, where the even the offline catalogue format was steered by the database as well.

With that kind of experience in mind, the line up of speakers we’ve got is particularly cool – they worked with the silo effect head on, and come up trumps. They include Tracy Green, Head of Online Services for the UK parliament (fresh from implementing a new content management system across several websites, as well as launching an award winning mobile intranet site – what’s not to like?), Olle Ahvne, Marketing Communication Manager for Ericsson (again, an award winning strategist) and Karin Zingmark, PR manager for Viasat (who has led Viasat’s explosion onto social media). We’ve put together so many awesome speakers – check them out!

(BTW, for you non-Swedes, the conference will mostly be in Swedish).

What: Digital Knowledge Day 2013

When: April 10th 2013

Where: Hotel Rival,Stockholm, Sweden

Complete your booking here!

A curse on your pages: Facebook madness from Nestle and Beneful

Beneful, a Nestle subsidary which manufactures dog food, is in trouble:


If you’re making dog food, and people start saying that your product is killing their dogs then that’s about as bad as it gets.

You may remember that Nestle provided social media consultants with one of the best, and most public examples about how to mis-manage a Facebook page with the palm oil debacle – here’s one of many blog posts which covered this. One of the key problems with this was censorship, and the deletion of posts.

How to avoid treading this road again? Address the situation head on, speak to the customers, try and gauge the scale of the problem maybe? Nope. Instead, in the face of a growing storm Beneful posts things like the following:




While, at the same time, their Facebook page is starting to fill up with things like:

Screen Shot 2013-02-03 at 8.59.28 PM

Which gets a response something like this:

Screen Shot 2013-02-03 at 9.00.52 PM

It seems like Beneful are going down the same road which Nestle trod a few years ago – threaten your page fans with deletion and remain anonymous as possible; further, don’t address the issue directly.

And here’s why this is, generally, a bad move:

Screen Shot 2013-02-03 at 8.58.05 PM

Cultivating a following of ‘influencers’ works both ways – have a good product, be respectful and people will love you for it. Start disrespecting your followers, particularly by censoring posts, and you’ve not only set your Facebook page on fire, but are merrily throwing petrol onto the flames.

And here’s the impact on existing customers:

Screen Shot 2013-02-03 at 8.52.19 PM

It’s easy to criticise, but here some things I would do in this situation:

– Stop being so goddamn anonymous; use your first name when you moderate or comment; try and give people a name to which they can talk to. Anonymity makes it much easier to be offensive, and get treated the same.

– Use Facebook to acknowledge the current situation; for every hater, there’s a bunch of people who, for the time being, think your brand’s okay. Directly acknowledging the problem, sharing the facts and describing the action you are taking is the mature thing to do. Posts about skyping dogs, in the face of news articles about less than healthy hounds make it look like you’re ducking the issue. Regardless of whether the product is fine, or not, you need to be proactive with the facts – even if it is bad news. Who else can help?- can the FDA support your claims, for example, that everything is okay?

Nestle’s Digital Acceleration Team became a bit of a big deal a while ago – so why does this feel like familiar territory for Nestle?

Update: Beneful are clearly using copy and paste – they seem to spend more time informing people they are removing their posts that actually doing anything meaningful. Community management, this is not.

Thanks to the awesome Corporate Bollocks Facebook page for tipping me off about Beneful.


Swedish Kommuns and Facebook: Another dirty review

Back in February I published a post which gave a quick, and rather dirty, review of Swedish kommuns and their use of Facebook. I like dirty, so here’s a follow up which looks at the same pages, almost a year later, in December 2012. Before we go any further:

What’s wrong with the data:
– Not every page is included (Nor have I looked at multiple pages from the same Kommun), there’s around 90 Kommun pages included here
– Some kommuns have only opened their walls for posts in the last few months (but well done you!)
– I can miscount stuff
– Facebook’s supposed chronological presentation of posts can sometimes be a little odd
– Page owners can have removed posts

And what’s wrong with the method:
– I don’t include comments on posts from the page
– I don’t control for posts from the same person (there’s more than one page which has a lot of ‘engagement’ from one or two special individuals)
– I have not looked at how quickly questions, comments etc are responded to
– I’ve not even attempted to second guess the stratgy behind the page (is it to drive traffic to a website, or encourage people to comment; for example?)
I have instead looked at the number of likes (yes, yes I know – not a sharp metric), the average monthly growth of likes and the number of posts by page friends.

Here’s a bubble chart showing number of likes on December 19th 2012:

Facebook likes - Swedish Kommuns

Clicking on the image will take you to the clickable chart.

Some clear winners, on the basis of likes. I’ve set up Facebook pages myself (and currently manage one for a client) and likes have never been the best metric. But, but, but – when you’re a kommun you know, roughly, how big your target group is – the number of citizens in your kommun. If your likes are a miniscule percentage of that number, then maybe you’re not really communicating optimally via Facebook.

Here’s a bubble chart of kommuns showing the average monthly growth, of likes, in 2012:

Average monthly growth - Kommun Facebook pages

Click on the image to visit the clickable chart.

Averages are dangerous things, but these data give an idea of what the best, and worst, growth looks like. These data give you no idea whether all the growth is restricted to a few months, or is spread throughout the year. Update: Basically, what I’m saying here is that some pages, indeed, a whole bunch of them; have simply not grown much at all. Others, like Karlstad are gaining likes at an impressive rate. GIven that many kommuns said in a recent survey that they lacked resources to adequately manage social media, this is not altogether surprising.

Bubble chart showing the number of posts left by page friends in 2012:

Facebook page posts - Swedish Kommuns

Bonus – correlation between posts and likes…make of this what you will:

FB likes vs posts

I’ve removed Karlstad’s page as including it sends all the other pages scurrying into the bottom left hand corner while it sits high up in the top right – no question for them that more likes is associated with more posts. There are outliers where lower numbers of likes are associated with still high numbers of posts but, generally, on the basis of these data you could roughly predict the number of posts by looking at the number of likes. Remember, as well, that these data are skewed by pages which may have only opened their walls in the last few months as well as pages which have received lots of attention from one or two people.

What can you do with this data?
No deep insights, I’m afraid, as I’ve got no idea what your particular kommun strategy with Facebook is. However, as most kommuns are using social media to have a dialogue with their citizens (Slide 7, Sveriges Kommun & Landsting) then presumably a low level of posts from page friends is an indication this target is not being met (though you may rock with comments on your posts – I didn’t look at that). Check out what the rock stars from this list are doing and remember, developing a social media presence takes time, patience and more than a little enthusiasm.

You can find my list of data here – if you notice any errors, sorry, entirely my fault and if I can find the time I’ll make this less dirty and more shiny. Here’s the raw data from the last time I did this.

Google Analytics IQ Exam – Revising and Passing

I recently left my previous job at Malmö University and started a new role as an analyst with Search Integration – this is an awesome opportunity for me, Search Integration are a great company and the hook for me was being able to work 100% with web analytics. A huge bonus is being able to work with Brian Clifton, whose book ‘Advanced Web Metrics with Google Analytics‘ got me into the field of web analytics in the first place – definitely a Matrix moment!

Part of my new journey with Search Integration was to formalise my Google Analytics knowledge by taking the Google Analytics IQ test. I’m happy to say that I passed first time, with a score of 94%. There’s already some great posts about the IQ test out there – for example, I recommend checking out the very candid ‘How to  fail the Google Analytics Exam‘; but I thought it would be useful to share my insights about taking and passing this exam.

Passing the Google Analytics IQ Test
You can find a good description of what the test is like here, so I won’t bother going over that ground again. Here’s my top tips for passing, and getting maximum benefit from the exam:

1. You need to be familiar with Google Analytics, and using it on a regular basis, before taking the exam. You need to have a basic understanding of implementation and, crucially, interpreting the GA reports – this will give you a strong foundation to build on; otherwise your revision is going to be very abstract. Despite having worked a lot with GA I was very glad that I had revised – there were definitely areas I needed to brush up on.

2. Google’s IQ exam presentations are great. Watch them several times and make notes. I’d put aside at least 2 full days to go through them. Incidentally, right click on the links and open them in a new tab to get the full screen view – I noticed a few people were missing this on some forums and squinting at the preview size. My approach was to watch the presentations, pausing them often and referring to the Google Analytics help site and Brian’s book as supplementary material.

3. Use the awesome Google Analytics Test site – it’s an unofficial community driven labour of love led by Eric Fettman. I would say that most of the questions are harder, and require more lateral thinking, than the GA IQ itself. Beast yourself on their site. To get maximum value from the questions pay attention to the answer sections – for all the questions, not just the ones you got wrong. I found the site particularly helpful in identifying my weak spots.

4. Don’t get cocky. My revision was good, and I’ve a lot of experience in using Analytics but there were definitely a few questions which got me scratching my head. I made sure that I was particularly confident with the following sections:

– Cross domain, and sub-domain tracking
– Attribution rules (and dudu’s blog has one of the best visual explanations I’ve seen)
– Adwords integration with GA
– How cookies work in GA
– Ecommerce tracking
– Filter functionality

Really, the subject is so broad that it’s hard to point at particular subjects and say you must understand these more than others. However, a big epiphany for me – in my journey as an analyst – was really getting to grips with what a visit is, and how Google Analytics measures them. Also, really work hard at understanding the difference between a dimension and a metric – this will pay back big time.

5. As my driving instructor said to me, ‘don’t learn to drive to pass the test, learn to be a good driver’. Aim to exceed the level of the GA IQ test with your revision and you’ll not only feel very confident as you answer the questions, but also as you continue to use it in your job – which is the whole point.

My revision benefited from the following resources:

– The IQ Exam presentations
Brian Clifton’s book and blog
The Google Analytics help site (this gets better and better – good explanations, often with examples)
The Google Analytics Developer site (I love this site, but you don’t need to dive too deep into it – it will help with GATC customisations and cookie understanding; the dimensions and metrics reference guide is also awesome))
Google Analytics YouTube channel (check out the web analytics TV – benefit from web ninja wisdom and lots of real world cases)

There’s a ton of good analytics blogs, but here’s a start:
Advanced Web Metrics
Occam’s Razor
Google Analytics Blog
Analytics Talk

I guess my main take away is to make your understanding of Google Analytics as real world as possible – if you’ve got a access to a range of accounts representing different types of websites then you’ve got a great start. Otherwise pay particular attention to the examples and demonstrations in the resources I identify above.

Good luck, and feel free to get in touch (via comments below) if you’ve got specific questions, or comments, about the IQ exam.

The best Facebook analytics report (that you’re probably not using…)

Let’s take a step back, for a minute, from the usual slightly spitty excited talk about measuring social media by ‘engagement’, ‘reach’ or ‘virality’. All good stuff, to be sure, but sometimes – as a website owner, you just want to know how many visits are coming to your site from the Facebook page you lovingly maintain. It’s your bread and butter metric – it makes sense, and you know where you are with it. Chances are, most of your conversions take place on your website, that’s a good enough reason to want insight into your traffic sources.

Facebook domain insights is a great tool – among the reports it gives you are insights into visits to your website, from any link out there on Facebook. Here’s an example from a website, showing visits coming its way from Facebook:

Graph showing links clicked on from Facebook, back to a website

The number of clicks sent to a website, from links appearing in the news feed, page or profile walls. (Click for a larger view)

If you’re running a web analytics tool, you’ll can also get a similar report from your referring sites report – but what you get from insights, which you don’t get from your own website, is a report showing how many times links about your site were shared, regardless of whether they were clicked or not:

Daily shares from a Facebook page

The number of times people included a link to the website in a status message or wall post (Click for a larger view)

Ok – so where’s the insight? What’s the action I could take?
The first time I looked at the graphs above, my immediate question was ‘what’s responsible for the spikes?’. Finding out that answer will give me insights, which I can use to adjust my content strategy. Insights for domains has a ton of other useful reports, for example, it will show the success of any ‘like’ or ‘send’ buttons you have on your website.

The next question is, are the trends in the data above suggesting that the strategy that I have for Facebook is working, or not? In this instance, I can say that the prioritized purpose of my employer’s activities on Facebook is not to acquire traffic to the website  but to provide a platform for conversation with students (though, obviously, it’s nice when content is shared). However, as we increase publication of content on Facebook – and start using it to reach a larger audience we’ll definitely be using these reports to see what impact it’s having – we’ll be looking for an upwards trend. Whether this traffic does anything valuable to us, or for them, on our website is another question….

Talk to your webmaster, give them buns and get the meta tag from Facebook added to your website – enjoy.