Epic Hypothesis (Lean UX)
Frame epics as falsifiable if/then hypotheses with tiny discovery experiments and 2-4 week validation measures — so you can kill bad ideas before committing engineering.
You wrote the epic as "Build a dashboard with five charts," sized it at 8 weeks, and the CEO already told the board you're shipping it. That's not an epic — that's a commitment you'll regret. A hypothesis says: if we do X for this persona, we expect outcome Y, and we'll test it with a prototype in two weeks.
Who it's for: PMs framing major initiatives, product leads managing uncertainty, teams running Lean UX or continuous discovery, PMs who keep getting asked to "validate" features they've already built, founders avoiding premature commitment
Example
"Frame our upcoming Slack integration as an epic hypothesis" → If/then statement (integration for trial users → 20% activation lift) + 3 tiny experiments (Figma prototype, concierge Slack DMs to 10 users, landing page CTA) + 4-week validation measures (activation rate, qualitative "saved me time" quotes) + decision criteria for build/pivot/kill
CLAUDE.md Template
New here? 3-minute setup guide → | Already set up? Copy the template below.
npx ccplaybooks add epic-hypothesis# Epic Hypothesis
Frame epics as testable hypotheses using an if/then structure that articulates the action, target beneficiary, expected outcome, and validation method. Use this to manage uncertainty by making assumptions explicit, defining lightweight experiments ("tiny acts of discovery"), and establishing measurable success criteria before committing to full build.
This is not a requirements spec — it's a hypothesis you're testing, not a feature you're committed to shipping.
## The Framework (Tim Herbig / Lean UX)
**If/Then Hypothesis:**
- **If we** [action or solution on behalf of target persona]
- **for** [target persona]
- **Then we will** [attain a desirable outcome or job-to-be-done]Get new playbooks like this one
One email a week with new Claude Code workflows. Free, like everything here.
No spam. Unsubscribe anytime.
What This Does
Converts an epic into a testable if/then hypothesis with lightweight validation experiments and 2-4 week success measures. Forces explicit assumptions ("we believe X") and falsifiable outcomes ("we know it's valid if Y"), so you can kill bad bets before full engineering commitment.
Built on Tim Herbig's Lean UX hypothesis format. Pairs with problem-statement, proto-persona, jobs-to-be-done, and user-story-splitting.
Quick Start
mkdir -p ~/Documents/EpicHypothesis
mv ~/Downloads/CLAUDE.md ~/Documents/EpicHypothesis/
cd ~/Documents/EpicHypothesis
claude
Provide the epic idea, target persona, existing discovery evidence. Claude drafts the hypothesis, suggests tiny experiments, and defines validation measures.
The Three-Part Structure
- If/Then Hypothesis — If we [action] for [persona], then we will [outcome]
- Tiny Acts of Discovery — 2-3 lightweight experiments (prototype, concierge, landing page, Wizard of Oz)
- Validation Measures — Within [2-4 weeks] we observe [quantitative + qualitative outcomes]
Experiment Types
- Prototype + user testing — Clickable mockup with 5-10 users
- Concierge test — Manually deliver feature for 10 users
- Landing page test — Describe feature, measure interest/signups
- Wizard of Oz — Present as automated, deliver manually
- A/B test — Test minimal version against control
Tips & Best Practices
- "Then we will" must be an outcome, not an output. "Users will have notifications" is output. "Users will respond 50% faster" is outcome.
- Falsifiable measures only. If every result confirms the hypothesis, you didn't test it.
- 2-4 week validation window. Longer windows mean you've already built the thing.
- Kill cheerfully. Invalidated hypotheses are saves, not failures.
- Don't frame post-commitment as hypothesis. If the CEO already said yes, you're not testing — you're theater-performing.
Common Pitfalls
- Hypothesis is actually a feature spec ("build a dashboard")
- Skipping experiments and going straight to full build
- Vague success criteria ("users are happy") that can't be falsified
- 6-month validation windows where you've already built the thing
- Treating epics as commitments, then running experiments as theater