Growth Loops · Session 03 · Dates TBD
Build the next useful action into the product
Ship AIPhoenix, ArizonaPress → to begin
More activity is not automatically more value.
A product that makes customers click ten times to finish one job can look engaged while becoming worse.
Customers can reach first value. Tonight asks what a healthy return looks like and how the product makes the next valuable action easier to discover.
Act one
An active customer is not someone who opened the app. It is an eligible customer who repeated a behavior at a cadence that matches the job.
Defaulting to daily active users makes a monthly planning product look sick and a high-friction daily product look healthy.
| Meaning | Illustrative measure | |
|---|---|---|
| Frequency | How often value recurs | Weeks with at least one useful share |
| Depth | How much of the core job completes | Useful results per active workspace |
| Breadth | How widely value spreads | Eligible roles completing their useful action |
| Quality | Whether the action worked | Successful shares without reversal |
Choose the dimensions the product can improve without forcing activity. More of all three is not always the goal.
Exercise: write the active-use contract
define active = eligible account + core action + window
entity: workspacebehavior: useful report sharedwindow: at least once in a natural weekexclusions + time zone documented
Write who is eligible, the exact event conditions, the interval, exclusions, and whether multiple actions increase the count. Ask another team to calculate it from the definition.
If two people produce different denominators, the metric is not defined yet.
Act two
Compare what retained and slipping customers do, then investigate why the difference exists before turning it into product pressure.
| Numerator | Denominator | |
|---|---|---|
| Templates | Eligible accounts using a template | Accounts with a repeatable report job |
| Collaboration | Accounts with collaborator action | Multi-user accounts |
| Scheduling | Accounts scheduling a useful report | Accounts with recurring need |
| Export | Accounts completing an export | Accounts whose workflow leaves the product |
| Core action | Accounts repeating useful share | All eligible active accounts |
Putting ineligible customers in the denominator makes specialized features look weak and sends the team toward the wrong fix.
Retained customers may use a feature because they are retained — not the other way around.
Behavioral separation suggests where to investigate. Sequence, customer job, segment mix, and experiments are needed before claiming causation.
The strongest candidate is both behaviorally distinct and connected to a customer reason you can articulate.
Exercise: inspect the retained behavior
retained cohort vs slipping cohort
same eligibility + age window appliedscheduled sharing separates before week-4 outcomedifference concentrated in recurring-report teams3 customer conversations queued
Compare action sequences and feature adoption, then narrow the finding to the segment for whom the behavior matches a real job. Record alternative explanations beside the hypothesis.
Do not announce a magic feature. Leave with a candidate behavior and the evidence still missing.
Act three
The intervention should help the customer resume an unfinished job, not manufacture a session for the dashboard.
Saved state, templates, collaboration, and progress can create a reason to return because the product remembers the work already done.
Next value
Engagement theatre
An intervention can raise sessions and lower trust at the same time. Measure dismissals, opt-outs, errors, and unsuccessful returns.
Live build: make one next action obvious
ship next-value path → inspect eligible journey
trigger: saved recurring job, no schedule setsurface: in-product empty statesuccess: next useful share completedguardrails: dismissals + errors verified
Choose one behavior and one eligible segment. Build the smallest in-product or lifecycle intervention, then walk through completion, dismissal, ineligibility, and error states.
The intervention is complete only when it can stop. Suppression and exit rules are part of the feature.
| Definition | Decision evidence | |
|---|---|---|
| Active use | Entity + core action + natural window | Two people calculate the same number |
| Adoption | Feature use among eligible accounts | Retained and slipping cohorts compared |
| Hypothesis | Behavior + customer reason + alternatives | Sequence and segment caveats recorded |
| Intervention | One next-value path with exit rules | Value and trust events verified |
Commit all four to 03-engagement/. Session four follows these cohorts through time to see where repeat value holds or fades.
Engagement is value with a reason to recur.
Next
Next session chooses the retention definition that matches the product, reads cohorts across age and calendar time, and turns bright spots and decay into ranked hypotheses.
shipai.club · Free, always