Trial mechanics experiment
Experiment: trial-length test
Trial-length tests are operationally complex but high-signal. The 7-day vs 14-day vs 30-day decision changes activation timing, support load, and conversion math. The framework below isolates the variable and names the only honest metric: trial-to-paid conversion at 60 days, not 14.
Min sample size: 300+ trial signups per variant for a 25%+ conversion-rate lift detection. Below 300, qualitative beats quantitative.
Duration: 60+ days from test start, since the primary metric is 60-day conversion. Cannot be rushed.
Verified · editorial policy
Hypothesis structure
Changing trial length from [CURRENT] days to [VARIANT] days will [INCREASE / MAINTAIN / DECREASE] trial-to-paid conversion by at least [EXPECTED LIFT] because [SPECIFIC REASON].
If you cannot complete this template, you do not have an experiment — you have a guess.
Variant design
Only the trial length changes. Same onboarding emails, same in-product nudges (adjusted to fit the new timeline). New signups split between control and variant.
Primary metric
Trial-to-paid conversion rate measured at 60 days post-signup. Not at trial end; at 60 days. Some conversions happen post-trial-expiry.
Secondary metrics (watch but do not decide on)
- Time-to-activation (does longer trial delay or accelerate activation?).
- Support-ticket volume per trial user (longer trials = higher support).
- Cancel rate within the trial (early signal of fit).
Procedure
Step 1
Decide on the variant lengths
Most useful comparisons: 7-day vs 14-day, 14-day vs 30-day. Skipping intermediate steps (7 vs 30) produces wide effects that are hard to attribute.
Step 2
Adjust onboarding emails to fit the variant timeline
A 7-day onboarding sequence does not work for a 30-day trial; the cadence must change. Document the adjustments so you can replicate.
Step 3
Track each cohort through 60 days
Conversion at trial-end is the half-story. The full story includes conversions in the 2-4 weeks after trial expiry.
Step 4
Compare 60-day conversion rates
If variant produces higher 60-day conversion, ship it. If lower, keep control. If within noise, the test was inconclusive.
Step 5
Watch support-load economics
Longer trials = more support per trial user. A 14-day trial that converts at 12% with low support might beat a 30-day at 14% with double support load.
Self-deceptions to avoid
- Reading trial-end conversion as final. Some users convert after expiry; only 60-day numbers capture the full effect.
- Ignoring support cost. A higher-converting variant that doubles support load may have worse unit economics.
- Running on too small a sample. Trial-length tests require real cohort sizes; sub-100 per variant is noise.
What success looks like
Variant produces 20%+ higher 60-day conversion AND comparable support load. Or maintains conversion while reducing support load.
Related benchmark
See the directional range for trial to paid conversion to calibrate the expected lift in your hypothesis.
Frequently asked
- Should I default to 7-day, 14-day, or 30-day trials?
- Depends on time-to-value. Products where users hit the 'aha' moment in under 60 minutes do well on 7-day trials. Products requiring data setup, integrations, or workflow change benefit from 14-30 days.
Other experiments
Test on a page that is already pointed in the right direction
A/B tests on a misaligned page produce two losing variants. The diagnostic labels the alignment problem first; the test optimizes within the right alignment.