Let’s be honest, data and particular analytics data does not have a good reputation in public at the moment. And let’s face it, who likes to be watched and tracked? I guess no one.
So why do I work in analytics? Do I like to watch and track people? No, I don’t like that. It is something that makes me uncomfortable.
Let me go back to the point where I discovered analytics for the first time. That was 2006. I was responsible for parts of a classified ads marketplace, and a significant portion of my job was to decide what we will build. What kind of stuff we should improve and what new features make sense.
My first approach was to listen to the stakeholders, what does IT think, what the management, what the sales team? That gave me at least a list of items. Then I looked into the competition or other e-commerce use cases. The list got more significant.
In the first month, we decided by gut feeling and stakeholder management.
But I found this very unsatisfactory and pretty annoying to discuss these things in bigger groups. It’s like always when it comes to product design everyone has an opinion.
I looked for answers and found the first ones in web analytics. And I can tell you that it was a great discovery. It was like magic. It took some time to get a proper setup with Google Analytics (it was 2006, meaning no segments, no events).
But then we could tell what areas are used most of the time. How our funnel looked like (even when we didn’t know that it was called funnel)? How different segments of users used the platform (logged in vs. anon – in the old days, you did that by using different views with different filters)?
Using this data in the discussions was like finding a new superpower. People won’t argue against data if you present it as a solid case. It opened the way for “better” decisions. At least that was what I thought at that time.
Automatically? Of course not.
I work with analytics data now for 14 years. And I had enough situations where data was viewed in just the right angles to prove a decision already made.
On the other end, I saw teams looking at their data, being unable to make any sense of it, or deriving any action from it. A good friend of mine is still calling this “the men who stare at numbers”. And he is just honest about it.
During the consulting years, I did so many projects where we worked in improving the tracking setup, improving data quality, and making sure that people trust the data. And when all was clean and good, there was a reasonable question: “And now what”? What does the data tell me now
Well, data in itself is like a rock. Lying there, looking solid, not moving, showing some edges. You can create a Zen garden of these stones, rake them from time to time and look at them (we call these dashboards these days).
I am coming back to the first part of tracking people.
If we only tracked people to have data lying around and glanced in high-level dashboards, it’s pretty worthless. There is no sense to track a person to know that X sessions have happened on a website in a month.
My purpose overall is to build products that let users progress with the tasks they have. The products should be accessible for anyone, they should not cause new frustrations, and they should be a joy to use.
High goals, aren’t they. But for me, it makes sense to have them as a north star for all product work.
I do tracking to help me to optimize for these goals. When I track user retention, I do this, because this indicator can help me to figure out if my product works. But I only do it if retention is a useful metric.
I know plenty of services where retention is measured wildly, but if a user comes back every week, does not say anything if the assistance is reasonable or not.
If you are Facebook and your revenue is based on ad impressions and interaction, of course, the user retentions are an essential metric because you want to optimize for it (if this business model is in general, a good idea is a different story).
If you are todo, retention is essential. If you are a fitness app, the retention of users doing other sessions could be a primary indicator.
When you are selling tickets, retention is a minor indicator.
Data makes sense for me if it helps to build excellent services. And that also includes marketing. The service starts with marketing. If you have a unique solution for a known problem, it is beneficial to find people with this problem and point them to your solution. That will make their lives better. If you bring your solution in front of people who don’t care about the problem, you waste their time (and your money).
You will sometimes hear the approach to “track everything; you don’t know what you might need in the future.” Please don’t follow this path. I rarely had an occasion where we were in significant need of long historical data for an event that became just relevant.
Make sure you understand your business models. What drives revenue, what drives costs. Check what metrics are based on each other and can be an indicator of final revenue.
I like to work with the du-Pont schema as a starting point.
And then go deeper into some metrics to see what context I would need there.
Question if you need user-based data?
Product analytics can also work if you don’t have any user information.
A lot of website tracking would even work without session information.
And look at stuff you track right now and if you still need it.
Sunsetting a tracking event is a good thing. It cleans up your data of something no one cares. It frees up your documentation. It makes your data more trustable.
Because it helps me find patterns that can point me to improvements in my services, they need validation outside of analytics. But analytics data can be a game-changer by identifying patterns that would not be visible if you simply talk to people (“I use the service every week.” Well, our data tells us that rarely people do that, so you might just be kind to me).
Every week I sent out posts, products, tweets and projects I came across. Give it a try and just unsubscribe if they are boring. Sign up to get my links