Jared Zhao
CEO
Good retention fuels strong lifetime value (LTV) and gives you the confidence to invest heavily in growth. Bad retention, on the other hand, is a sign of a leaky bucket—a SaaS startup’s worst nightmare.
💡
In the rest of this article, we’ll explore how to do Cohort Retention analysis using the data that you’ve already been collecting since the earliest days of your product.
Tracking retention goes far beyond vanity metrics like signups or website traffic.
Product Health—High churn or declining retention often points to issues with usability, onboarding, or core value. Segmenting by how far customers made it in your product can pinpoint where to improve.
Customer Lifetime Value (LTV)—Strong retention boosts LTV, letting you spend more to acquire customers and scale faster.
Cohort Retention measures retention over time, grouped by the period a customer started (a “cohort”). Grouping by cohort enables you to evaluate retention as it changes, allowing you to see if you’re moving in the right direction.
To improve retention, you need to understand what sets your best customers apart. Segmenting users by attributes like acquisition channel, persona, or product usage reveals where value is created—or lost.
Take Uber: do riders stick after their first trip, or after several? What patterns drive loyalty? Segmentation helps you prioritize features, target valuable users, and align your team around faster growth.
Takeaways:
CSV and Google Sheets are fine for quick testing but don't deliver lasting value. We should consider replacing them with a fully set-up sample data source to guide users faster.
Connecting a live data source should be frictionless. Add options like booking support directly from the connection page, extending free trials, or offering incentives to encourage it.
The “Aha” moment for new Athenic AI users probably happens somewhere around 15 questions mark:
💡
Identify the user and action:
User: The unit you’re measuring retention for (e.g., Uber → riders; Athenic AI → Teams). Usually it’s whoever is billed, but sometimes it’s another entity.
Action: the main way users get value (e.g., Uber → rides completed; Athenic AI → questions asked).
Create 2 Datasets
Users Dataset (1 row per user):
User IDActions Dataset (1 row per action taken):
User ID,timestamp
For now, just select these basic columns (Simple Datasets). If the data is messy, you can build Advanced Datasets using SQL later.
Create Cohort Retention Charts
Now, we’re ready to start creating our Cohort Retention charts. The best way to format your questions is:
To get the most from retention analysis, you need to find what separates your best-retained users from the rest. Adding segmentation data in Athenic AI lets you compare retention across different user behaviors.
For example, we looked at retention by data source type, projects created, team size, and number of questions asked. We added minimal supporting datasets to our Knowledge Graph to compare users with live vs. static data sources, and those who asked 1 vs. 15+ questions.
Retention analysis isn’t just another metric — it’s the clearest blueprint for building a product users can’t live without. By deeply understanding cohort retention and segmenting precisely, you uncover where your product truly delivers value — and where it falls short.
The insights uncovered through cohort retention empower you to create "Aha!" moments earlier and more frequently, driving your customers towards lasting engagement. Keep listening closely to your retention data—it has the power to guide you toward the product your customers can't imagine living without.
Sign up for an Athenic AI account to start building your own Cohort Retention charts, or book a demo to speak with us directly.
