If you read this, chances are that Web analytics and Mobile application analytics is something that you are familiar with.
The internet abounds with presentations, blog posts or white papers that deal with the notion of value creation from the data you can collect online.
At Futurice we create software. A lot of software in fact. Therefore, Digital analytics is a topic where we both have a lot of vested interest and a lot of ideas how to get it done right.
Traditionally, the focus has been on the analysis part of the data. You collect all this data from your site or mobile app and your first step typically is to organize into a set of reports and dashboard - getting the data out.
This is a very important step, but quite often you’ll find out that getting the data out isn’t delivering the full value everyone in the business is hoping for.
Solution is to have a human (that weird person called the analyst) look at the data and make sense of it. This data exploration revolves around 3 main themes:
- Verifying business assumptions
- Building hypothesis on why things aren’t what they should have been, testing the hypothesis with the available data and/or designing tests if the required data isn’t there
- Turning the test results into recommendations to improve the situation
This is obviously the most valuable part of the data utilization, the one that everyone talks about.
But here comes the catch: in order to do all these great things, you need good data!
In the recent years, tools like Google Analytics have greatly simplified the data collection process. Deploying tracking codes is no longer as IT focused, labor intensive as it used to be, giving to many the deceiving impression that the data collection is a walk in the park.
We think differently.
In fact, at Futurice we believe that there is as much value created by collecting the data in the right way as there is to use it the right way.
For your data to deliver business insights, it must reflect your business intent and you must get it right the first time. While you can perform the analysis of the data at a later time (and repeat the exercise several times), you can only collect data once and data not collected initially will be lost forever.
So, when we design a site or a mobile application, we want to design its measurement strategy as well.
Measurement strategy is another word for the Key Performance Indicators (KPIs), Metrics and Segments that will be used for the analysis part.
So how do we come up with that Measurement strategy? We use a methodology. Methodologies are great because they give consistent results.
Our methodology isn’t new or secret. It’s all about taking a step back, stating the business goals that the site/application will help fulfill, breaking these into smaller measurable actions that happen within the resource and matching them to a metric/KPI. Segments can be defined beforehand in cases where subset of data are obvious but thanks to the tools flexibility, they also can be created afterward and can be revisited.
Once the metrics and KPIs are defined, what is left to do is merely coding their capture into the tool of your choice
You can of course look at this data collection planning as a necessary unexciting part of the process but I see it very differently. There is a lot of value from that part of the process.
Firstly because it forces to revisit the business aim of the resource with a critical eye – what is it that we are trying to do with this site/application? Is it realistic? Do we have a market?
Secondly because it forces to think about performance – how will we know what works and what doesn’t
Thirdly because it forces to think about processes – how will we document all of this so that we remember why we did it that way? How will we make sure that this will keep working as our site/application evolves? How will we account for the business change?
Lastly because it is the foundation for good data that allows the analysts to do their magic
At Futurice, we devised a graphic representation of this value creation that we call the “hour glass model”. Here's how it looks like :
The good thing about this model is that it also gives us purpose : we make the measurement strategy as a core part of our development project because we want good data so that our clients can get actionnable insights and that our designers and developers can see the performance of their work in action. In addition, it's a powerful visual reminder that getting the data pouring in is only providing half of the value !