Claim and evidence
Write a short doc for every test and log the result, including failures, so learning compounds and is shared.
No dated example yet. Treat this tactic as a hypothesis.
Primary
The company or founder describes the tactic and its result themselves: own blog, talk, filing.
Working · 2026-11 Standard practice; tooling ranges from a spreadsheet to experimentation platforms.
What the numbers show
No published number for this tactic has passed our word-for-word check yet, so there is no chart.
Why it works
Written hypotheses and results prevent repeated tests and turn losses into inputs for the next round. Sharing failures builds a culture where testing is safe.
How to run it
- Write hypothesis, metric, audience and duration before launch
- Record the outcome and the learning
- Share wins and failures with the wider team
- Tag and store docs in one searchable place
- Review old learnings when scoring new ideas
Ways to run it
Design options to test, not promised results.
- Option AOne-page test doc
A short template filled before launch: hypothesis, metric, audience, duration and expected decision.
- Option BSearchable learning library
Store every doc in one tagged place so the team can find past tests before repeating one.
- Option CShare the losses too
Present failed tests alongside wins in team reviews so testing feels safe.
Timeline
- 12016 first documented: Sean Ellis, SaaStr Annual talk transcript
- 22026-11 status checked: Working
Dated examples
No dated example yet. Treat this tactic as a hypothesis.
Does it still work?
Standard practice; tooling ranges from a spreadsheet to experimentation platforms. source
Risks
- Docs go stale if nobody owns the archive
- Learnings without context mislead later teams
Sources
- saastr.com
“It's not just about sharing the wins, but it's also sharing the things that didn't work.”
✓ quote checked word-for-word on the page - saastr.com
“Capture that learning and talk about that learning.”
✓ quote checked word-for-word on the page
Change log
- 2026-11-20 — harvest v0.1 (ops-ai)