Tag Management Tool for Data Analyst
is this a good idea for your company or your department ?
The issue: The missing data
As Data Analyst, I am often seeing the case that we are missing some data for a proper analysis. In a perfect world, we would track everything that is necessary directly when the website or the application has been created. It would be so easy to directly have data from every important touchpoints and just realized the analysis afterwards. Even better, we could have automatic report that tell us when something is going wrong. The problem that we are facing with missing data is a serious problem for 2 particular reasons :
- Data that are not been retrieved will, most probably, never be retrieved. – Data that are not been retrieved will, most probably, never be retrieved.
- While fixing and retrieving the data, the Product Manager / Owner is blind from some necessary KPI.- While fixing and retrieving the data, the Product Manager / Owner is blind from some necessary KPI.
The major question that come to our mind is: Why are we not doing that?
The answers are pretty simple:
- Most of the company lack the resources for doing proper tracking at the same time than developing the product.
Be careful, when I speak about resources, I am not only writing about developer but basically about everything that work around the development of the product or website.
Why the Product Manager didn’t think about the tracking? Why the developer didn’t implement it during the development phase? Why no one in management asked for this? And more importantly why the data analyst was not giving specification during the development process?
You can see that this problem can come from different causes and everyone has a reason to be blamed. When you face this kind of issue, or any issue that I can recall, you always have to think that everyone is most likely to do the best he can. Therefore it is not the issue coming from people but from time-resource. The developer would like to introduce this tracking but has to pass most of its time correcting or developing the application. The Product Manager may not have been trained properly on tracking and rely on the data analyst that is wondering why there is no new data in the system…. - The application itself (or the website) is evolving and the functionality that we want to track now was maybe a side feature then, or even did not existed and has been developed aside from another team. More and more often, company release MVP with basic metric calculation and want to improve data later, if the basic concept has actually proved itself worthy.
The solution: Tag Management System
The problem now is that you have new data that you would like to integrate into your database but no team to integrate them because the development team has to take care of the bugs. The Product Owner has received your requirement for the tracking but it will most likely be integrated in 2 or 3 sprints (which mean 6 to 9 weeks to wait at a minimum).
The solution there is that you can integrate a Tag Management system into your system and therefore be more flexible of what you can do and track. However, is it really a good idea to do that? Here are the pro and con of that solution:
Pro :
- Shorter time to implement tracking
You will basically shortcut the sprints of your IT team, no need to wait 6 weeks for the data to start being populated - Knowing exactly how the data have been tracked & retrieved
Understanding the logic behind the data is something that I see lacking within the most brilliant data analyst. - Closer relationship with your IT to find smart solution for data retrieval
Better integration with your IT team is always a plus. You get to know each other and more likely, you will understand each other. - Improve technical skills of the person in charge, would keep him up to date to technology change.
This aspect is most likely interesting for the people in charge of the tag implementation. Thanks to that, his job will be more challenging, which is a very important aspect of a job for data analyst. (we like challenges)
As in data science, “there is no free lunch”, in tag management, it is the same. This superb solution comes unfortunately with some big cons:
- Knowing what need to be implemented in the Tag management and what not
As you have a way to shortcut IT implementation and sprint, be prepared that some people will try to stuff your tag management with everything possible. It is not why you can that you should. - Your code is not the best code
At some point in time, sooner rather than later, there is going to be someone in the IT team to discover that the codes you wrote were not the most efficient (if you are good) or crappy (if you are not). As this is not your job, you don’t have time to look for the most efficient solution and that could harm (slow down hopefully) the application (or website). At this point, you need to get even closer to IT, make them understand why you need this solution and cooperate with them to find a process without slowing YOU down. - Need of a specific profile in your team
Let’s be honest, finding someone able to code in C, JS, Java and in Python / VB / R is not that easy. It is hard and expensive, this needs to be considered, otherwise you will be have hard time when this person leave the department (or even worst the company). Documentation and backup solution needs to be prepared before the problem appear. - High expectation & higher commitment
As you may have noticed, the information in a company do not travel as fast and as far as it should. Therefore, it may be that the tracking you have implemented with your Tag manager, doesn’t work because of an update that you never heard about. You will need to find where the issue is coming from and a solution for that issue. It means time and commitment to every change in this application (or website).
Conclusion :
As you can see the tag management implementation doesn’t come free of charge. In my opinion, the hassle is worth the flexibility but you need to be clever when working with this kind of tool. Clear assignment for that tool needs to be decided and the IT and Data analysis Department needs to have good relation with each other, otherwise, the first problem, people will jump to each other throat. You have been warned. 😉