Acquisition · trust signals · 67,000 users · Programme across 10 domains
About ten testimonial tests changed the priority, not conversion
Formats, placements and messages changed. About ten experiments produced no reliable winners for this audience.
Source: supplied OPW growth-system deck and speaker notes. Anonymized case; not attributed to a named app or BelovedAgain. The source does not provide confidence intervals or p-values. Results are specific to this tested flow.
What we tested
Member testimonial blocks against pages without testimonial content, with different formats, placements and messages across 10 domains. The programme covered approximately ten experiments.
The example
The documented example involved 67,000 users. The programme reported no reliable winners. Small step-level movements did not justify rollout.
The decision
We deprioritised further testimonial testing for this audience. Flat results freed traffic, design time and development capacity for other opportunities.
This is not proof that testimonials never work. It is evidence about this audience and this programme.