BY JOSHUA AJIEH
Users may leave because they never reached a useful outcome, the experience failed to meet their expectations, or the problem they solved does not occur often enough to bring them back. A falling retention curve shows the behaviour. It does not, by itself, explain the cause.
Start by tracing the first journey. Then compare users who returned with those who did not, talk to both groups and test one explanation at a time.
First, define what returning should mean.
A daily habit, a weekly workflow and an occasional purchase need different retention measures. Someone who does not return tomorrow has not necessarily abandoned a product they only need next month.
Write down the action that represents useful repeat use and the interval in which you would reasonably expect it to happen. Measure that behaviour alongside visits or app opens.
- Useful action: What does the user do when the product helps them accomplish their goal?
- Expected interval: When would that need naturally arise again?
- Cohort: Which group of new users are you following, and have they had enough time to return?
Separate activation from repeat value.
A signup is an account creation event. Activation should represent a meaningful first outcome. Retention asks whether users come back to receive value again.
These are related questions, but they require different evidence. If users leave before their first useful outcome, investigate the path to activation. If they reach that outcome and still disappear, investigate its usefulness, quality and relevance to a recurring need.
For example, imagine a tool that helps founders analyse customer interviews. Creating an account is a signup. Producing a usable synthesis could be a candidate activation event. Analysing another set of interviews when the next research task arises could represent repeat value. This is an illustrative example, not a universal activation definition.
Trace the whole first journey.
Follow the journey from the promise that attracted the user to the outcome they experienced. Look for the point where expectation, effort and value stopped matching.
- Arrival: What brought the user here, and what did the message lead them to expect?
- First task: What were they trying to accomplish?
- First outcome: Did they complete that task, and was the result useful?
- Next need: When would they need that outcome again, and what would bring them back?
Examine drop-off between these steps. Where available, check errors, loading delays, failed actions and support feedback. Pair the behavioural data with conversations or direct observation.
Compare users who returned with those who did not.
Compare users from similar signup periods who have had the same opportunity to return. Then examine differences in their goals, acquisition sources, first actions and outcomes.
- Did returning users arrive with a different problem to solve?
- Did they reach a useful outcome that other users missed?
- Did they receive help or bring existing data into the product?
- Did they have a more frequent reason to use it?
- Were some acquisition sources attracting a poor audience fit?
Treat these differences as clues. An action associated with returning does not prove that making everyone take that action will improve retention. It gives you an explanation to investigate.
Ask users about what actually happened.
Speak with people who returned and people who left. Ask about their recent experience rather than whether they like the idea or might use a future feature.
- What was happening when you decided to try the product?
- What were you hoping to accomplish?
- Walk me through what you did and what happened next.
- Which part helped, and where did you get stuck?
- When did you next face the same problem?
- What did you use or do on that occasion?
If someone says the product was useful, ask for a specific example of what it helped them do. Positive feedback and repeat behaviour provide different kinds of evidence.Read why customer praise is not the same as product demand.
Choose a test that matches the evidence.
Do not redesign the whole experience from one interview or one funnel chart. Write a working hypothesis, identify the evidence supporting it and choose a focused test.
- If users cannot complete the first task:test a clearer or simpler path to that outcome.
- If the outcome is disappointing:investigate quality, reliability and whether it solves the intended problem.
- If expectations do not match the experience:test a more accurate promise with the intended audience.
- If the need is occasional:assess retention over an appropriate interval before trying to manufacture daily use.
For each test, define who it is for, what will change, what behaviour you will measure and how long users need to demonstrate a result. Record what would support or weaken your hypothesis before you begin.
Check whether reach is also part of the problem.
Acquisition and retention can fail together. A campaign may attract people whose needs do not match the product, while the experience may also make it difficult for suitable users to reach value.
Compare outcomes by audience and acquisition source before concluding that the whole product needs rebuilding.Explore how to distinguish a product problem from a distribution problem.
Your retention investigation checklist.
- Define useful repeat use and its expected frequency.
- Identify the first meaningful outcome.
- Locate drop-off before and after that outcome.
- Compare users with similar opportunities to return.
- Interview returning and non-returning users.
- Choose one explanation and run a focused test.
The aim is to understand where value breaks down and what evidence would justify your next change. More features, reminders or traffic should follow that investigation.
Work through your next decision.
Stop Building, Start Validating offers a practical framework for investigating weak traction and choosing what to test next. If you want help examining a live product problem, explore the Product Diagnosis engagement.