Note- This post aims to help beginner PMs understand the thought process behind documenting and setting up analytics.
I have received a lot of - "My current priority is to set up analytics on our products and I don't know where to start" type messages from PMs for so long that it has driven me to this point. I figured I have to do something about it. So...here goes.
Product analytics is a specialised application of business intelligence that helps a product manager evaluate product defects, identify opportunities for product improvements and detect patterns of usage or capacity of your product. It is you trying to figure out what your customers are doing on your product and using that as a way to inform optimisations and expansion opportunities.
The main goals for setting up analytics generally are-
Understanding your customers
Figuring out customer pain points that some of them have learnt to manage for optimal conversion.
Improving activation.
Actively measuring bottom-line results/funnels.
With analytics you are-
Asking questions about specific parts of the product (I want to see the number of customers that have done x and y over a period of z)
Collecting answers and trying to interpret them to come up with hypotheses, insights and experimentation ideas.
So what are the core elements when setting up analytics?
Actions- Basically every button/link with a CTA. Might also include forms being filled and or pages being reached/loaded/triggered.
Attributes- This comes in different shapes and sizes depending on the products and it is solely tied to the states your users/customers need to achieve that is the most important to your company. So if a customer has not completed their KYC, their attribute can be set as “status” = “unverified”, and when KYC has occurred that can change to “verified”. It helps with creating correct cohorts and understanding each cohorts issues.
This brings me to the main mental models that are important to set up analytics for your company. Here is a fun list-
Funnels: Funnels are a set of steps/actions that your users have to carry out to be able to get a specific job done by your product. A simple example is an onboarding/Activation funnel that may include actions like-
Sign up (by filling in all the information)
Go through onboarding/education cards
Submit KYC
Get approved
Make the first deposit.
It typically can be ~ 3 steps and might also be more. With that, you can glean information about-
The hardest or possibly complicated parts of your onboarding
Friction points that may be a result of engineering bugs
So the most important things you should consider to be able to test out funnels are-
What goals can customers achieve using my product, or what goals are most important to us that our customers/users need to achieve?
What actions[buttons, pages etc] help them achieve these goals?
Cohorts- This refers to a group of users/customers with similar characteristics. In the product analytics sense, this is more directed and focused on customers with similar BEHAVIOURS(read characteristics) that is as a result of them using your product. A simple cohort can be created using (Continuing from the example above)-
Date of last login.
Status of their KYC.
Approval status
Or a group of those things together.
The main purpose of this is to-
Understand behavioural patterns of certain user groups
Figure out how to solve problems for a particular group of users/customers by hyper-focusing on their pain points.
Painting a picture of what your customers might be like (demography, size - this is in the case of companies, etc).
Cohorts + Funnels- With your understanding of both funnels and cohorts, you can possibly piece this one together. Knowing and measuring different cohorts funnels and paths, you can piece together what might be difficult for what kind of customers. This might be fairly explanatory but if it is not, send me a DM on Twitter here to ask for clarity.
Everything can be a query if you understand the elements at play (The actions, possible funnels etc)- This basically means analytics can answer any and each of your questions(So far the elements you need are tracked)
Tools for analytics
Here is a small list of some of my favourite analytics tools-
Mixpanel- Here is a video of a member of the Mixpanel team answering certain questions using Mixpanel and a video on demand product as an example- https://www.loom.com/share/1b2b3f67c7dc4b9d9a7f72aae5569f63?t=341
Segment
Google analytics.
Stakeholders that matter as it relates to analytics-
Growth and marketing
Product
Sales
Product design.
In final words, track every action that probably has a button.
Follow me on Twitter, reach out to collaborate and pick my brain here.