Experiment library
Improve SaaS trial-to-paid conversion · method v1
Activation-triggered trial follow-up
Send help to trial users at the moment they stall before the activation event, and compare paid conversion against users who get the standard sequence.
- Hypothesis
- Trial users who stall before the activation event convert less often because they never reach the value moment, not because the price is wrong.
- Change
- A short, specific follow-up (email or in-app) triggered when a user has not reached the activation event within a set number of days.
- Who is included
- New trial accounts started during the test window that meet the eligibility rule (for example, not sales-assisted).
- Compared with
- Eligible trial accounts that receive the existing onboarding sequence only.
- Assignment
- Random 50/50 split on account ID at trial start.
- Measurement
- Paid conversion per arm among matured cohorts, counting only collected first payments net of refunds within 30 days.
Worked example
400 trials a month at 8% paid gives 32 customers. If half receive the follow-up and their paid rate rises to 10%, the treated group yields 20 customers against 16 in the comparison. At that volume you need roughly two to three months before the difference is distinguishable from noise.
Evidence you need
- Trial start dates per account
- Activation event timestamps
- First successful payment per account (Stripe or billing export)
Implementation steps
- Define the activation event and stall window in writing
- Assign accounts at trial start and log the arm
- Build the follow-up for the treated arm only
- Wait for every cohort to mature past the trial length
- Compare paid conversion and first-month refunds per arm
Cost components
- Copywriting and setup time
- Email or in-app messaging tool
- Analyst time to reconcile payments to accounts
Stop if
- Unsubscribe or complaint rate in the treated arm above twice the comparison
- Refund rate in the treated arm above the comparison by more than 3 points
Scale if
- Treated paid rate higher across at least two matured monthly cohorts with refunds no worse
Does not apply when
- Fewer than about 100 trials per month
- Products where activation cannot be tracked per account
- Sales-led trials where a rep already follows up
Adapt this experiment to your business
Your values stay in this browser until you sign in, then become a private protocol that records this method and version.