The Cost of Data, part 2: Please, don't track Everything
Steen Rasmussen
Connecting People, Data and Business Outcomes - Director of Data Innovation, Board Member & International Keynote Speaker
When faced with the question, so “what do you want to track?” There is one scary answer which keeps repeating and we hear from customers again and again.?
Everything.?
And even as magnificent as it sounds it is still a horribly wrong answer.?
And it isn’t even because I do not understand the logic and the thinking behind it which most likely goes something like;?
“I don’t know what we should track, and I don’t know what we will need in the future, so if we track everything then we will be future proof, since we have all the data and we can then later decide what we need.”
And the logic is good, except for a couple of things.?
there is a cost associated with tracking everything
First of again there is a cost associated with tracking everything.?
Both in the form of setting the tracking up which is one thing, but also in the form of the actual cost of long term governance and maintenance. So if we conservatively estimate that every single thing we track under the perfect automated setup takes 3 minutes per week to validate and maintain, then if we track just the 200 standard metrics in Universal Google Analytics that gives us 10 hours of governance - every week.?
tracking everything will normally not track everything
Second of and even more critical, tracking everything will normally not track everything.?
Not even close.?
What??
Yes, tracking everything will in most cases track all interactions which might be considered standard. On the big plate of KPI’s there are so many different things you could track that it is seriously overwhelming.
领英推荐
If we add the 300-400 additional things we can now track with event tracking in Google Analytics 4, that easily means we should have one person dedicated full time to do only maintenance.
8 hours every day, no breaks or sick days.?
So the time it takes to track everything correctly is significant.?
And in most cases errors will not be caught within a week.?
redundant tracking significantly increases the risk of losing important or critical data
Spending resources on maintaining redundant data significantly increases the risk of breaking, contaminating or losing data at the important or critical KPI level. especially if combined with the facts from the last post (link) that data quality costs are exponential.??
The point is simple, don’t track everything.?
Make a simple data strategy.?
Define what is critical for the business and start by tracking only what you can use or activate now or plan to use or activate in the future.?
And if you are already tracking everything, it is time to stop, evaluate what you have and throw something away you have never used.
Did you miss part 1?
Founder, SaaS Pimp and Automation Expert, Intercontinental Speaker. Not a Data Analyst, not a Web Analyst, not a Web Developer, not a Front-end Developer, not a Back-end Developer.
3 年Tracking everything is in breach with GDPR. Companies are only supposed to track just the data they need, stop collecting and dispose of the collected data once they are done with it. Collecting data they thought would be useful but turned out not to be is fine. It's ok to be wrong, but there should be regular audits to determine what is tracked and in breach of these tracking minimalism principles. Perhaps we could tell clients that tracking everything will happen in stages. After each stage, they must demonstrate what actions they took based on the data collected in the last stage. If a given data point made no improvement or they did not use the data from that data point then perhaps they should lose the data for that data point. That should force them to see data as something that needs to lead to action and results or they lose it. A company could see itself as wanting a lot of data but end up having to pay extra for dismantling a lot their tracking and disposing of the collected data. Someone in the organisation, a Head of Analytics, the Chief Data Officer or another CXO, will start asking whether it would not be smarter and cheaper to aim for a more nimble implementation and track only what makes a difference.
Founder @ UseItBetter, Analytics & Conversion Rate Optimisation
3 年Just use UseItBetter or Heap to autotrack everything and use GA to track select few metrics that you will be 100% sure of. You will have the best of both worlds.
MBA em MKT digital e Analytics
3 年Willing to track everything is a good way to track nothing. Restrictions are fundamental for keeping focus and providing solutions for real problems, not imaginary ones.
Brand Leadership Community founder - Former IBM, ACNielsen, SaxoBank & IIH Nordic - MeasureCamp CPH organiser since 2017
3 年Redflag at “I don’t know what we should track....". Run a Goals/KPI workshop before proceeding any further...