How To Build a Cohort Retention Chart From Raw SaaS Event Data

A cohort retention chart shows how consistently groups of users return after a shared starting event, such as signup or activation. Built from raw SaaS event data, a reliable cohort retention chart helps product teams separate acquisition volume from lasting usage, identify lifecycle drop-offs, and choose specific onboarding or product improvements.

What a Cohort Retention Chart Measures

A cohort retention chart measures the share of a starting population that completes a defined return action during each elapsed period. Rows represent cohorts, columns represent periods since the cohort began, and each cell contains retained users divided by the original cohort size.

Cohorts and observation windows

A cohort is a group that shares a time-based or behavioral starting condition. A monthly signup cohort might contain every user whose first account creation occurred in January. An activation cohort might contain users who completed a meaningful milestone during the same month. The choice changes the question being answered.

The observation window is the time range in which a cohort can be measured. A January cohort may have six complete monthly periods by July, while a June cohort may have only one. Comparing a mature cohort with a recent cohort without marking unavailable periods creates a false impression of poor retention.

User, logo, and revenue retention

User retention counts people. Logo retention counts customer accounts or companies. Gross revenue retention, commonly abbreviated GRR, measures recurring revenue retained before expansion, while net revenue retention, or NRR, includes expansion and contraction. A product-led SaaS team may need all four views, but they should not be combined in one chart.

Measure	Denominator	Useful question
User retention	Users in the starting cohort	Do individual users return?
Logo retention	Accounts in the starting cohort	Do customer organizations remain active?
GRR	Starting recurring revenue	How much starting revenue remains?
NRR	Starting recurring revenue	Does the installed base expand or contract?

For product behavior, start with user or logo retention and define the return event precisely. A page view may be too weak for a core workflow, while a successful report export, collaboration action, or weekly project update may better represent continuing value.

Design the Retention Definition Before Building the Chart

A valid cohort retention chart starts with a written retention definition that names the cohort event, return event, time cadence, and comparison window.

Choose the cohort event

Use a signup cohort when you want to measure the complete journey from account creation. Use an activation cohort when signup volume includes many users who never reach the product's value moment. Activation cohorts are often more useful for diagnosing product experience because they compare users who crossed a meaningful threshold.

Document the rule in one sentence: “A user belongs to the weekly activation cohort for the week in which the user completes three tracked projects.” The rule should be deterministic, testable, and based on events that your instrumentation captures consistently.

Choose the retained action

Retention requires a return action, not merely the presence of an account. Define whether retention means logging in, viewing a dashboard, completing a core workflow, inviting a teammate, or producing an output. The action should represent value and should not change halfway through the analysis.

Our team has found that broad login retention can hide weak product usage. When we tested a core workflow event instead, the resulting chart exposed a clearer gap between users who experienced value and users who only returned to inspect the account.

Choose cadence and comparable periods

Weekly cadence suits products with frequent usage, while monthly cadence suits workflows that naturally occur less often. Avoid mixing weekly and monthly percentages in one chart. For each cohort, calculate period zero, period one, period two, and later periods using the same duration and timezone rule.

Use a retention specification checklist before querying data:

• Starting event and timestamp field
• Unique identity key
• Return event and qualifying conditions
• Weekly or monthly cadence
• Timezone and date boundary
• Minimum cohort size for interpretation
• Latest complete period included

Build a Cohort Retention Chart From Raw Event Data

Building a cohort retention chart from raw events requires four derived fields: cohort period, activity period, elapsed period, and distinct retained-user count.

Start with the event table

A practical event table contains at least user_id, event_name, and occurred_at. Add account_id for logo-level analysis, event properties for qualifying actions, and an identity mapping table when anonymous visitors can later become known users.

Filter the starting event first. For each user, select the earliest valid starting timestamp. Trimming the starting data to first occurrence prevents one user from entering multiple cohorts.

Derive the four calculation fields

1. Cohort period: truncate the user's first qualifying timestamp to a week or month.
2. Activity period: truncate each qualifying return event to the same cadence.
3. Elapsed period: calculate the number of cadence units between cohort period and activity period.
4. Retained users: count distinct users for each cohort and elapsed period.

The core formula is:

Retention percentage = distinct retained users in cohort period and elapsed period / distinct users in the original cohort × 100

For example, suppose a monthly cohort contains 100 users. If 64 distinct users complete the retained action in month one and 42 do so in month two, the cells are 64% and 42%. A user who performs the action five times in month one still counts once.

Worked retention table

Cohort	Starting users	Month 0	Month 1	Month 2	Month 3
January	100	100%	64%	42%	35%
February	80	100%	61%	39%	Not complete
March	120	100%	68%	Not complete	Not complete

Do not replace “Not complete” with zero. The value is unavailable because the cohort has not had enough time to reach that period. Keeping unavailable cells blank protects the chart from recent-cohort censoring.

For SQL, group by cohort period and elapsed period, then use a distinct count of the user identifier. For a spreadsheet, export one row per user and activity period, remove duplicate user-period combinations, and build the summary from that clean table.

Create the Chart in Excel and a Product Analytics Platform

Excel works well for a controlled audit, while a product analytics platform is preferable when event volume, identity rules, or cohort refreshes make manual preparation unreliable.

Excel implementation

Assume a clean activity table contains columns A through E: user ID, starting date, activity date, cohort month, and activity month. A cohort month formula can use =DATE(YEAR(B2),MONTH(B2),1). An activity month formula can use =DATE(YEAR(C2),MONTH(C2),1). The elapsed month formula is:

=(YEAR(E2)-YEAR(D2))*12+MONTH(E2)-MONTH(D2)

Create a pivot table with cohort month in rows, elapsed month in columns, and distinct user ID as the value. Divide each retained-user cell by the starting-user count for its row. Apply a color scale to create a heatmap, using blanks for incomplete periods. Add a line chart that plots each mature cohort across elapsed periods when the audience needs curve comparison rather than cell-level detail.

Platform implementation

A product analytics platform should automate identity resolution, event filtering, distinct counts, cohort refreshes, and segment comparisons. The setup still requires human review: confirm the starting event, inspect raw examples, verify timezone handling, and compare platform counts against a small spreadsheet sample.

Founder OS can support this workflow by capturing product actions through Product Tracking, connecting events to User Profiles, and allowing behavioral User Segmentation. A team can create an audience for users who completed activation but did not perform the retained action, then use the Product Onboarding Tool to test a targeted experience. GTM Report can connect the resulting behavior to broader growth and revenue questions.

When we tested manual spreadsheet refreshes against automated event reporting, the main difference was not the formula. It was consistency: automated updates made it easier to notice a changed event definition before a monthly review used the wrong comparison.

Read Retention Curves and Find the Drop-Off to Fix

Retention curves identify where usage declines, but they do not prove why the decline occurred; product teams must combine the pattern with behavioral evidence and controlled follow-up.

Read rows and columns separately

Rows show the experience of individual cohorts over time. A steep decline between period zero and period one suggests an early activation or value-discovery problem. A gradual decline suggests ongoing usage friction, insufficient habit formation, or a use case that is not frequent enough.

Columns compare cohorts at the same lifecycle age. If newer cohorts have higher period-one retention than older cohorts, an onboarding or product change may be promising. If every cohort drops at the same elapsed period, investigate the workflow or customer need associated with that point.

Recognize common curve shapes

• Sharp early drop: inspect activation completion, setup time, first-value events, and onboarding guidance.
• Flattening curve: a stable core of users may have found durable value; identify what those users do differently.
• Improving recent cohorts: compare release dates, acquisition sources, plan types, and activation quality before attributing improvement to one change.
• Irregular spikes: check batch jobs, imports, billing cycles, and tracking changes before treating them as user behavior.

Use segments to generate hypotheses. Compare retained and non-retained users by first feature used, time to activation, invitation behavior, account size, acquisition source, and plan. A segment difference is a prioritization signal, not causal proof.

What surprised our team was how often the highest-retention users did not use the most features. They reached one repeatable value moment quickly, which made time-to-value more actionable than raw feature count.

Avoid Incomplete Cohorts and Misleading Retention Data

Data-quality checks determine whether a cohort retention chart describes customer behavior or measurement noise.

Audit identity and duplicates

Check whether one person can have multiple user IDs across devices, workspaces, or anonymous and known sessions. Decide whether the analysis is user-level or account-level, then apply the same identity rule to the numerator and denominator. Deduplicate repeated events within each user and period.

Audit event consistency

Compare event names, properties, and volume before and after releases. A renamed event can make retention appear to fall even when usage is unchanged. Missing client-side events, blocked scripts, offline activity, and server-side events that arrive late can all create gaps.

Audit cohort maturity

Mark each cohort-period cell as complete only when the full observation window has elapsed. Filter out cohorts below the minimum size chosen by the team. Small cohorts can still be useful for investigation, but their percentages should not be presented with the same confidence as larger cohorts.

Run these checks before publishing:

• Does period zero equal the starting cohort size?
• Are retained counts distinct by user or account?
• Can retention exceed 100% because identities or denominators differ?
• Are incomplete cells blank rather than zero?
• Did event volume change after instrumentation or release changes?
• Do a few raw user timelines match the summary output?

Our team uses a fixed validation sample before sharing a new chart: inspect several users from the numerator, several from the denominator, and several excluded users. That small audit often catches timezone and identity mistakes faster than reviewing a long query.

Turn Cohort Insights Into a B2B SaaS Growth Workflow

A retention workflow turns one chart into a repeatable cycle of measurement, segmentation, intervention, and review.

Use a five-step operating loop

1. Instrument: define and capture signup, activation, core-value, invitation, and return events.
2. Measure: refresh the cohort retention chart using complete and comparable periods.
3. Segment: compare retained and non-retained users by behavior, account attributes, plan, and acquisition source.
4. Intervene: improve onboarding, messaging, education, or product flow for a clearly defined segment.
5. Review: compare the next mature cohorts while documenting the release date and changed exposure.

Connect retention to revenue questions

Retention becomes more commercially useful when paired with account and revenue attributes. Compare logo retention by plan, user retention by account size, and retained behavior by acquisition source. Do not infer revenue retention from user retention alone, because one retained user may represent very different commercial value from another.

Founder OS is designed for teams that want to shorten the distance between raw behavior and action. Install Product Tracking to capture events, use User Profiles and User Segmentation to define behavioral cohorts, review onboarding performance with the Product Onboarding Tool, and connect product patterns to GTM Report. The result is a workflow in which retention monitoring supports both product decisions and revenue conversations.

A practical implementation sequence is to start with one activation event, one retained action, one cadence, and one mature comparison. After the definitions are validated, add account-level views, acquisition attributes, and targeted onboarding experiments.

Week	Implementation focus	Output
1	Identity and event audit	Trusted starting and return events
2	Cohort calculation	Validated retention table
3	Behavioral comparison	Prioritized drop-off hypothesis
4	Onboarding or product intervention	Documented experiment and follow-up cohort

Start building a reliable cohort retention chart with Founder OS by instrumenting product events, creating behavioral segments, finding retention drop-offs, and connecting onboarding improvements to GTM and revenue insights. Start free to reach your first insight quickly, or book a demo to map the workflow to your B2B SaaS product.