Last Wednesday I attended the conference day of the ninth Google Analytics User Conference (GAUC) in Amsterdam. It’s the biggest Google Analytics event in Europe. There were a lot of interesting speakers. This post contains my five key takeaways from the conference. For more details about the speakers, visit http://www.gauc.nl/programma-conference-day/.

1: Dimension labeling##

With Google’s Universal Analytics, Google gave us the option to add custom data (custom dimensions & metrics) to the standard analytics data. As a web analyst, you are responsible for the data people act on. You are also the one that should make this as easy as possible for them. The example given at the GAUC was about a blog. Say you want to see if a read article is old. You’ll probably do this by adding a custom dimension with a publication date. But you shouldn’t stop there:

• Publication date: a great way to see when a article was published. But you’ll need to calculate the actual days an article is old.
• Amount of days an article is live: you instantly see how many days an article is old, but you don’t really know if this is a new, old or even archived one.
• New, old or archived: your data instantly shows if read articles are new, old or archived. No thinking required.

The examples above show that you should always think about what the data is for, and how to make it useful in the easiest way. Keep that in mind when adding data to any report.

2: Set your KPIs carefully##

KPIs add great value to any data set. In Google Analytics, you can set goals to track important events and pageviews, or use e-commerce data to analyse your website’s performance. When setting up your KPIs, make sure that they have a quantitative and qualitative part.

A bad KPI example

Conversion rate (CR) is not really useable as a KPI. For example: your website’s CR increased from 10% to 15%. A nice increase of 50%. In the same period, website traffic is down with 50%. What does this give us?

• Period 1: 2000 visits, 10% CR, 200 conversions
• Period 2: 1000 visits, 15% CR, 150 conversions

The image is clear: a better CR does not result in more conversions.

A good KPI example

Conversions are great for KPIs. They have a quantitative and a qualitative part. Both are hidden in the bad example above:

• Visits (quantitative): how large was the volume of visits where conversions could have happened?
• Conversion Rate (qualitative): how well did these visits perform?

In the same two periods as before, you would have seen that conversions were down. Looking deeper into the data (the quantitative and qualitative part), the drop in visits and better CR would have appeared. Less conversions, better performance of your website. Interesting if you’d ask me.

3: DIY APIs###

One talk was about adding API data to analytics data. A frequently used example is weather data. Add this information to see how sunny or cloudy weather impacts your website’s visitors. But the most awesome thing was the mention of Kimonolabs (or the alternative import.io). This tool allows you to turn any website into a data feed. Their video does a great job of explaining how it works:

kimono: a 60 second introduction from Kimono Labs on Vimeo.

Awesome right?

4: Measuring job interviews with UA’s Measurement Protocol###

The Measurement Protocol may be the most powerful feature of Universal Analytics. It basically allows you to setup a URL to send data to Google Analytics. This way, any system that can send GET or POST requests, can also send data to Google Analytics. A great example was given during the GAUC. It was about a vacancy website.

Normally, you’d optimize your data by website behaviour, for example job applications. This is nice and all, but the real question is: do these people get invited for a job interview? You can measure this with the Measurement Protocol:

• Step 1: Google Analytics is loaded on a webpage and generates a client id.
• Step 2: The user sends an application by filling out a form.
• Step 3: Add a hidden input field to the form and store the Google Analytics client id (step 1) in it. You can read how to het the client id in the google analytics guide.
• Step 4: Store the client id with the job application on the server.
• Step 5: As soon as an agent changes the state of an application to ‘invited for job interview’, send an event to Google Analytics via the Measurement Protocol and use the client id of the application (step 4) as the client id in this request.
• Step 6: Because of the last non-direct attribution model, and the same client id, the invited for job interview event will be attributed to the source the user had when applying for the job. Keep in mind that if a user had a new source between these two actions (the application and the invited for job interview event) the newer source will be attributed.

Looking at a sample data set:

• Adwords traffic: 10.000 visits, 200 applications, 25 interviews
• LinkedIn traffic: 10.000 visits, 100 applications, 50 interviews

Without the Measurement Protocol solution in place, you’d think Adwords had the best performing ads. With the new insights, LinkedIn is the better one. A great use of the possibilities of Universal Analytics.

It really got the analytics geek inside of me thinking. Potentially, you could use this to setup your own measurement system for apps (Google has a different way of tracking for apps and websites) to send the app behavior data to analytics just as website data. Make the app’s data mimic the website data, merging the data, and the insights, into one Google Analytics property.

5: Enhanced e-commerce for blogs###

Yes, enhanced e-commerce reporting for a non e-commerce website. It’s possible. A recurring theme throughout several talks was creativity. Be creative when implementing analytics into any site. This is how Simo Ahava came up with the crazy idea of using Universal Analytics (awesome) enhanced e-commerce report features for his own blog. He basically looked at the available product fields and used them to measure his blogs performance:

• Product: a blog article.
• Product view: a view of an article title.
• Product add to cart: a minor scroll on an article page.
• Product checkout:
  • Step 1: a 25% read.
  • Step 2: a 50% read.
  • Step 3: a 75% read.
• Transaction: a 100% read and time on article of 60 seconds.
• Revenue: word count.

A great and very creative use of Universal Analytics. You can read the details on his blog post. And yes, we’ll be implementing this on GEEK.

These were my 5 takeaways from the Google Analytics User Conference in Amsterdam. There was a lot more awesome stuff. It was an inspiring day, and I hope you’re inspired as well now you’ve read this article. If you have any questions or suggestions, don’t be afraid to ask.