.svg)
Retention analysis measures how many users return to your product over time, and which factors influence whether they stay or leave. It is one of the most important analytical methods available to product teams because retention is the single metric most directly connected to long-term revenue growth. This guide covers what retention analysis is, how to read retention data correctly, how to run it in Amplitude, and how to turn retention insights into product decisions that actually improve the numbers.
Acquisition gets the attention. Retention builds the business.
A product that acquires 1,000 new users a month but retains only 10% of them is running on a leaky bucket. No amount of acquisition spend fixes a retention problem, it just makes the bucket leak faster.
Retention analysis is how you find the leaks, understand why they exist, and identify what keeps users coming back. Done correctly, it is one of the most actionable analytical methods a product team has.
What Is Retention Analysis?
Retention analysis measures the percentage of users who return to your product after their first visit or a defined starting action over a set period of time.
A retention chart typically shows a cohort of users, grouped by when they first signed up or completed a key action, and tracks what percentage of that cohort came back on day 1, day 7, day 14, day 30, and so on. The result is a curve that starts at 100% on day zero and drops as users stop returning.
The shape of that curve tells you a lot about the health of your product. A curve that drops steeply and flattens near zero means you are acquiring users who do not find lasting value. A curve that drops initially but flattens at a meaningful percentage means a core group of users has found genuine value and keeps coming back. That flattening point which is the retention floor is one of the most important numbers in product analytics.
Why Retention Analysis Matters
Retention is the compounding metric. A product with strong retention grows faster with less acquisition spend because the existing user base stays and grows. A product with weak retention requires constant acquisition investment just to maintain a flat user base.
Beyond the business case, retention analysis is diagnostically powerful. It surfaces questions that no other analysis answers: Are users who signed up from a specific acquisition channel retaining better than others? Do users who complete a specific onboarding action in the first week retain at twice the rate of users who do not? Is a recent product change improving or hurting long-term engagement?
These are the questions that drive the most impactful product decisions, and retention analysis is what makes them answerable.
Types of Retention Analysis
There are three main types of retention analysis. Understanding which one to use for which question is what separates actionable retention work from generic reporting.
N-day retention measures whether a user was active on a specific day after their first visit. Day 7 retention, for example, measures the percentage of users who were active exactly on day 7. This is useful for understanding early engagement patterns and identifying the critical window in which users either find value or leave.
Unbounded retention (also called rolling retention) measures whether a user was active on or after a specific day, meaning a user who returned on day 9 counts as retained at day 7. This gives a less strict but often more realistic picture of whether users are still engaged with the product.
Bracket retention groups days into windows: week 1, week 2, week 3, month 2 -- and measures whether users were active at least once within each window. This is useful for products with natural usage cadences that are not daily, like weekly tools or monthly reporting products.
Amplitude supports all three types. Choosing the right one depends on how your product is actually used -- a daily habit app should use N-day retention, a weekly workflow tool should use bracket retention.
How to Build a Retention Analysis in Amplitude
Building a retention analysis in Amplitude starts with defining two events: the start event (what defines the beginning of the user journey, typically sign up or first session) and the return event (what counts as a meaningful return, could be any session, or a specific action like using a core feature).
Navigate to the Retention Analysis chart type in Amplitude. Select your start event and your return event. Choose your retention type (N-day, unbounded, or bracket). Set the time range and the cohort grouping by week or by month depending on your product's growth pace.
The resulting chart shows your retention curve. The key number to look for is the retention floor, the percentage at which the curve flattens. If your day 30 retention is 25% and holding steady, that means 25% of users who sign up are still active a month later. If the curve is still falling at day 30 with no sign of flattening, you have a product-market fit problem that no amount of feature work will fix without addressing the core value proposition.
How to Read Retention Data Correctly
Reading retention data correctly requires going beyond the overall curve and asking four questions.
Where does the curve flatten? The retention floor is your most important number. It tells you what percentage of users are genuinely getting ongoing value from your product. A floor of 0% means no one is coming back long-term. A floor of 20% means one in five users has found real value.
How does retention differ by acquisition channel? Segmenting your retention chart by initial_utm_medium or initial_utm_source often reveals significant differences in user quality across channels. Users from organic search may retain at 30% while users from paid social retain at 8%. This has direct implications for where you invest acquisition budget. Cohort analysis by acquisition channel is the companion analysis that gives you the full picture.
How does retention differ by onboarding behaviour? This is where retention analysis becomes most actionable. Segment your retention curve by whether users completed a specific onboarding action set up their first project, invited a team member, connected an integration, and you will almost always find that users who completed that action retain significantly better than users who did not. That action is your activation metric. Getting more users to it faster is your highest-leverage retention lever.
Is retention improving or declining over time? Compare retention curves for cohorts from different time periods -- users who signed up in January versus March versus May. If later cohorts retain better, your product improvements are working. If they retain worse, something has changed that is hurting the experience for new users.
The Connection Between Funnel Analysis and Retention
Funnel analysis and retention analysis answer different but connected questions. Funnel analysis tells you where users drop off in a specific sequence of steps. Retention analysis tells you whether users who completed that journey come back over time.
Used together they give you a complete picture of your product's health. The funnel shows you the acquisition and activation problem. Retention shows you the engagement and value problem. Teams that only look at one miss half the story.
A common pattern: funnel analysis reveals that 40% of users drop off during onboarding. Retention analysis reveals that users who complete onboarding retain at 35% while users who do not retain at 4%. The combination tells you that fixing onboarding is not just an acquisition problem -- it is your single most important retention lever.
Do you want to learn more on analyzing conversion drop offs? Watch this video:
Using Feature Analysis to Understand Retention Drivers
Once you have identified that certain user behaviours correlate with better retention, the next question is which specific features are driving that correlation.
Feature analysis in Amplitude lets you identify which features are used by your highest-retaining users, which features correlate most strongly with long-term engagement, and which features are being used early by users who later churn. This is the analysis that tells you what to build more of, what to improve, and what to deprioritise.
The combination of retention analysis (showing you the retention curve), cohort analysis (showing you which groups retain best), and feature analysis (showing you what drives that retention) is the analytical foundation for product decisions that actually move the metric.
Common Retention Analysis Mistakes
Using the wrong retention type for your product. N-day retention on a weekly tool produces misleadingly low numbers because users are not expected to return every day. Match the retention type to your product's natural usage cadence.
Not segmenting the data. An unsegmented retention curve tells you that something is happening. Segmented retention curves tell you who it is happening to and why. Always break retention down by acquisition channel, onboarding behaviour, plan type, and user type before drawing conclusions.
Confusing correlation with causation. Users who complete a specific onboarding action retain better -- but does completing that action cause better retention, or do users who were already more motivated complete it and also retain better? Retention analysis surfaces correlations. Experiments are what confirm causation. Use retention insights to generate hypotheses, then run experiments to validate them.
Ignoring the retention floor. Teams focus on improving early retention (day 1, day 7) without checking whether the retention curve ever flattens. A product with improving day 7 retention but a floor near zero is not getting healthier -- it is just losing users more slowly.
Turning Retention Insights Into Product Decisions
The output of retention analysis is not a report, it is a prioritised list of product interventions.
If your retention floor is too low, your core value proposition needs to be stronger or clearer. If your retention differs significantly by acquisition channel, your acquisition strategy needs to be rebalanced toward higher-quality sources. If users who complete a specific action retain dramatically better, your onboarding needs to guide more users to that action faster. If a recent product change correlates with a drop in retention for a specific cohort, that change needs to be investigated before it affects more users.
Each of these insights points to a specific intervention. Each intervention is a hypothesis. Each hypothesis is an experiment. That is the loop that makes retention analysis compound into real product improvement over time.
Want to make your experimentation programme compound?
If you are identifying retention insights but your experimentation programme is not building on them fast enough, the Experimentation Growth Engine automates hypothesis generation, prioritisation, and result evaluation so every retention finding feeds directly into your next experiment.
👉See the Experimentation Growth Engine
Want to talk through your retention setup?
FAQ
What is retention analysis in product analytics?
Retention analysis measures the percentage of users who return to your product after their first visit or a defined starting action -- over a set period of time. It tracks how a cohort of users behaves over days, weeks, and months, showing you what percentage remains active and where users stop returning. It is one of the most important analytical methods for understanding long-term product health.
What is a good retention rate for a SaaS product?
Benchmarks vary significantly by product type, pricing model, and target market. For B2B SaaS, day 30 retention rates above 40% are generally considered strong. For consumer apps, day 30 retention above 20% is often considered healthy. The more useful question is whether your retention floor is positive and whether it is improving over time -- both are better indicators of product health than comparison to a generic benchmark.
What is the difference between N-day retention and unbounded retention?
N-day retention measures whether a user was active on a specific day after their first visit -- for example, exactly on day 7. Unbounded retention measures whether a user was active on or after a specific day -- so a user active on day 9 counts as retained at day 7. N-day retention is stricter and better for daily habit products. Unbounded retention gives a more realistic picture for products that are not used every day.
How do you improve retention in a SaaS product?
The most reliable retention lever is identifying the onboarding action that correlates most strongly with long-term retention -- sometimes called the activation metric or the "aha moment" -- and optimising onboarding to get more users to that action faster. Beyond activation, retention improvements typically come from improving core feature value, reducing friction in recurring workflows, and personalising the experience for different user segments.
What is the difference between retention analysis and cohort analysis?
Retention analysis specifically measures whether users return to your product over time -- it focuses on the re-engagement question. Cohort analysis is a broader method that groups users by a shared characteristic and tracks how any metric evolves for that group over time. Retention analysis is typically a specific application of cohort analysis -- you are tracking the retention metric for cohorts defined by their sign-up date or first action.
How do you run retention analysis in Amplitude?
In Amplitude, navigate to the Retention Analysis chart type. Select your start event (typically sign up or first session) and your return event (any session or a specific meaningful action). Choose your retention type -- N-day, unbounded, or bracket -- based on your product's usage cadence. Set the time range and cohort grouping. The resulting chart shows your retention curve by cohort, which you can then segment by user properties to understand which groups retain best and why.
