Being data-driven is one of the core skills of a product manager. Product analytics refers to collecting and analyzing data to understand how users interact with your product. Data helps you answer the “What” part of the equation. It is a quick way to collect valuable insights from users. Be it identifying user friction points by looking at heatmaps or identifying patterns across cohorts or understanding funnel drop-offs, all fall under the purview of product analytics.

I use the principle of working backward to see what data needs to be collected. Identifying and defining key metrics is the first step. For example, let’s say the metric you want to track is DAU. Then you further define an active user as someone who opens the app and reads an article. Now that you have this information, you should look at all possible flows through which a user can read an article and then add events to the flow.

The data is collected by inserting pieces of code at key locations where users’ actions will lead to meaningful insights. This is called adding instrumentation. Every time a user does that action, data points are collected in a format that analytical tools can interpret. It is important to have an events list stored in a format that is accessible to all. It could either be as simple as an excel sheet or a built-in analytics inspector within your product. Using tools like Google Analytics, and Amplitude makes it very easy for anyone to quickly look up data given they know the right events. Here is a sample event tracker document for reference.

It is also very helpful to know the kind of data stored and where to access it from. This keeps you self-reliant in scenarios where the analyst may not be available. Keeping a document with feature names and their corresponding table names has helped me immensely.

There are numerous tools in the market but Amplitude and Google Analytics are two of my favorites. Amplitude is easy to use and understand whereas Google Analytics is incredibly powerful. You can add numerous data sources and then query to create dashboards. Metabase is yet another commonly used tool which allows one to use visual building blocks to write a query.  Learning these tools is quite easy given the user-friendly UI and availability of tons of resources online. They come with their own guided tours as well when you first launch the product. But the main task is to deeply understand what each of these metrics mean. In Google Analytics, you can hover over most pre-defined metrics to understand what it exactly means. You can reuse most of the pre-defined metrics but also create custom ones.

You can create custom dashboards and create slices based on input parameters which is exceptionally helpful for business teams for daily reports and basic root cause analysis.

Pro tip: I personally have a dashboard with all the inputs that could be factors for why an important metric is falling or going up. This makes correlating the causes a much faster exercise.

Amplitude lets you create cohorts which are synonymous with segments. You can do different types of analyses from acquisition to retention without needing the help of an analyst. I won’t go into the basic functionality details but here are some features that I really like in Amplitude.

Persona chart : As the name suggests, it groups users into clusters based on the similarities of their event behavior. So users who behave the same way will belong to the same cluster. You can choose the number of clusters you want to create. So let’s say you want to see new users who come back to your app even after 2 weeks. So you can see in the chart below there are 287167 new users in total in the time selected and 27% of them were active after 2 weeks. Each cohort shows you the % of people who are in the target segment. You can sort by the cohort who has the highest percentage and look at the events these users are performing and then optimize your flows to enable other users to do these events at the same frequency. That's a growth hack right there.

Pathfinder: Pathfinder is a great tool to show you how users are using your product. You can select the starting and ending events to see the different paths the user takes between the steps in a given time period. This can be very useful when optimizing funnels and flows.

In the diagram below you can see that 91% of users land on the main landing screen as the second step and then flow to different paths.

Experimentation: One of the easiest ways to set up and monitor experiments is through Amplitude. Setup the primary and secondary metric and configure control and treatment variants. You can select which segment of users see which variant. Finally you can easily see the results of the experiment, whether it was statistically significant or not and you can also roll it back.

The market is flooded with powerful tools. The skill lies in not knowing how to use the tool but understanding the data deeply. You should spend the most time not looking for things yourself but knowing what to look for.        

PS: Knowing a little SQL goes a long way.

‍