In this article, I'll go through the why and how you can start a product analytics MVP that is impactful to your organization.
I'll cover best practices on product analytics, starting with the why, what to track, documenting a tracking plan and analysis.
Why product analytics?
You manage a product that enables communication, listening to music, managing documents, sharing files, or other essential function; how do you know your product is fulfilling its purpose/mission?.
How do you know your product is fulfilling its purpose?
WHAT
- Which features are your customers using?
- How many active customers are using your product on a monthly basis?
- How long customers take to complete key tasks?
- Which features are correlated with higher account LTV?
- How frequently/recently are customers using the product?
- Are customers breaking ToS and sharing their username?
- Based on product usage which customers have higher risk of churning?
The answer to the questions above can inform decision-making across the organization, help you design a better product, and make your customers happier.
What exactly can you do with product analytics data?
DRIVE ACTION
- Increase satisfaction and customer retention
- Understand app and feature usage
- Prioritize development
- Empower Customer Success & Sales
- Uncover pain points (UX, copy)
- Monitor functionality
- Customer segmentation
- Predict churn based on usage patterns
- Deployment monitoring
- Feature testing (A/B testing)
Take a peak at how our friends at Gitlab do it, it is throughly described in their Handbook:
The Product Analytics MVP
Independent of your strategy, the product analytics MVP goal is to answer basic questions any product owner should know.
Process
A product analytics MVP should deliver value quickly with low to moderate effort, as well as an answer to the core three questions about your product:
- Who is using the application?
- What key tasks are customers performing?
- How frequently are customer using it?
TAKE ACTION
- Title a document with a one phrase motto of why your application/company exists
- Create a table with all the key tasks a customer can perform in your app and break the action from the subject
| Event/Action | Subject |
|---|---|
| Open App | User |
| Close App | User |
| Start | Workout; User |
| Stop | Workout; User |
| Pause | Workout; User |
| Complete | Workout; User |
Documentation
Documentation is often overlooked, and it is one of the most critical aspects of product analytics implementation. There are several reasons behind creating documentation for tracking.
- It enables data analysts to know which data to look for to answer questions without reverse-engineering or going through flows.
- It facilitates implementation by providing application developers with the exact code to implement (copy-paste), easier on them = more tracking implemented.
- It facilitates business continuity in case members of your team leave the company.
Here are three most common approaches for tracking plan documentation.
| No Documentation | Static Documentation | Test Driven |
|---|---|---|
| Just don't do it. | Document what you track and how. Tracking specifications help engineers, sales, marketing, product owners, and other analysts. Use Confluence or Notion, even Excel (if you really have to). | DevOps mature organizations take a test-driven approach to tracking, integrating analytics in their CI/CD pipelines. Use Iteratively |
TAKE ACTION
- Create a user story for each action and subject listed on the action & subject table.
Below is an example of what the user story should look like. Add screenshots to provide additional detail.
AS
The user starts a workout
THEN
Fire the following code
dataLayer.push({
event: "track-event",
event_category: "ux",
event_action: "start",
event_workout: "workout",
obj_workout: {
schema: "iglu:com.company/workout/jsonschema/1-0-0",
id: "", // << WORKOUT ID >>
action: "start",
},
obj_user: {
schema: "iglu:com.company/user/jsonschema/1-0-0",
id: "", // << USER ID >>
},
});
WHERE
<< WORKOUT ID >> = The unique workout id e.g. 2187271 (integer)
<< USER ID >> = The referencable backend unique user id e.g. joaocorreia (string)
TIP: Add tracking in small increments, ease your development team into analytics vs. one massive set of user stories.
Tools
Although we love Snowplow, we know it is not the right tool for everyone. You have to find a tool that fits your organization's analytics capabilities, and resources.
The main factor is your organization's analytics maturity, determined by:
- The existence of a data analytics/science team with SQL knowledge
- A modern data stack
- Leadership vision of data as a critical driver to the success of the business
- The volume of data your app can generate
Without a data team, your choice should fall with analytics solutions that provide an interface out of the box.
If, on the other hand, you have a data team hungry for data, SnowcatCloud (Snowplow) is probably a good fit.
Other Ideas for Product Analytics Data
Difficulty: Easy
- Active Users (Percentage of active users. How do new features release impact the number of active users)
- Cohort Customer Retention
- Visualize Customer Journey funnel
- Feed customer behavior/traits into Sales & Marketing with Hightouch
Difficulty: Medium
- Time to Task Completion (How much time does it take to perform key tasks the application was design to facilitate?)
- Segment Customers and Activate Customer Success (model your customer base on key behaviors and empower your marketing to help them progress in their customer journey)
Difficulty: Hard
- Churn Modeling (Create a probablistic model and empower customer success to target customers at high risk of churning with exclusive offers)
- A/B Testing or Feature Flagging
Product Analytics KPIs
Once tracking is implemented, you'll need to create key performance indicators (KPIs). KPIs are well-defined metrics that quantify the occurrence of a particular event/action.
You can create KPIs either through the analytics tool interface or SQL (we dbt because it keeps the complexity on the SQL side and allows you to enrich this data with other information.
Use a visualization tool (Tableau, Looker, Mode) to visualize the number of events over time and slice by additional dimensions (country/zip/phone model, user traits).
TAKE ACTION
- Create a KPI glossary with the metrics that matter to you and your team and agree on their exact definition.
| Metric | Definition |
|---|---|
| Daily Active Users | Users who perform at least one workout in a given day |
| Total Users | Number of users with at least one app start in the last twelve months |
| % Daily Active Users | Daily Active Users / Total Users |
| Monthly Active Users | Users who complete at least three workouts in a given month |
| % Monthly Active Users | Monthly Active Users / Total Users |
| # Workouts Started | Number of workouts started |
| # Workouts Completed | Number of workouts completed |
| User Workout Recency | Number of days since last workout |
Conclusion
It is not difficult to start a product analytics MVP, and the value you can deliver with product analytics data can be of strategic importance to your business.
Define the actions that matter, select a tool that fits your organization, document your tracking plan, define KPIs in writing and finally analyze your data. Add additional dimensions and slice to drive insights.
Keep the initial scope small and incremental, focus on measuring the key tasks your product was designed to accomplish, and remember:
We don't track everything; we have a strategy Tracking without a strategy is not product analytics and can be distracting us and our team from what matters. We track purposefully and strategically.
There are no silver bullets, we need budget for tools & people We will need tools and smart people with data to do meaningful product analytics, which means we'll need a budget, even if small.
We will need developer time and focus Any product analytics implementation requires developer time and focus. We build a cooperative environment with our developer team.