Why this matters
Growth ideas arrive faster than a small team can test them, and the loudest usually wins. The experimentation loop makes ideas compete on evidence, and the experiment hypothesis template fixes what is written before anything is built, so the result can be read whichever way it falls.
Small businesses also have little traffic. The experimentation cadence guide notes that low-traffic stages often cannot reach a result in useful time; writing the card is where that is discovered.
The coworking space and its ideas
Hivemill Coworking is a fictional coworking space in a market town selling day passes and monthly memberships. Day pass visitors are its best source of members, but few convert.
Hivemill Coworking: figures and research
| Measure (last six months) | Value |
|---|---|
| Day pass visitors (first-time) | 2,520, about 420 a month |
| Of those, started a membership within 30 days | 151, about 6% |
| Day pass price | £18 |
| Monthly membership price | £165 |
| Members who cancelled within their first two months | 22 of 151 |
| Research | Finding |
|---|---|
| Exit card left on desks (64 replies) | 26 said "I'd join but the first month feels like a big commitment"; 14 said they did not know membership included meeting room hours |
| Front desk notes | Visitors often ask whether today's day pass "counts towards" joining |
| Day pass visitors receive one receipt email and nothing else |
| Idea | Proposed by |
|---|---|
| A: Credit the day pass price against the first month if the visitor joins within 14 days | Front desk lead |
| B: A three-email follow-up after a day pass, explaining membership benefits | Marketing freelancer |
| C: Redesign the website homepage | Owner |
Scoring and choosing an idea
Score each idea on impact, confidence and ease, with a reason beside every score. Impact is judged against the step from day pass to member, not a local measure. Confidence comes from the research, not from who proposed the idea. Ease covers build, sign-off and anything that could delay launch. Multiply the scores to sort the list, then check the top: does it address the biggest problem the research found, and can the traffic detect its effect?
The card then follows the template's order; a field left blank is a reason not to start.
Task
- Score ideas A, B and C on impact, confidence and ease, with a one-line reason for each score.
- Choose one idea and say why the scores, the research and the traffic support it.
- Write the experiment card: name, evidence, problem, change, hypothesis sentence, expected effect, primary metric, guardrails, unit and audience, run time, decision rule, risks and owner.
- Check whether the traffic allows an A/B TestControlled comparison of two variants. and adjust the design if not.
- Say what happens after each possible result.
Worked answer
Hivemill Coworking: ICE scores out of ten
| Idea | Impact | Confidence | Ease | Score | Reason |
|---|---|---|---|---|---|
| A: day pass credit | 7 | 7 | 8 | 392 | Touches every day pass visitor at the decision; 26 of 64 replies and front desk notes point to commitment; a till rule and a sign |
| B: follow-up emails | 5 | 4 | 8 | 160 | Addresses the 14 who missed meeting rooms; evidence is thinner; easy to write |
| C: homepage redesign | 3 | 2 | 2 | 12 | Works on acquisition, not the day pass decision; no research behind it; weeks of work |
Choice. Idea A tops the list and matches the biggest research theme. Idea B is cheap and low risk; ship it after A ends so the two results stay separate. Idea C waits until research points to the homepage.
Hivemill Coworking: experiment card EXP-01
| Field | Entry |
|---|---|
| Name | Day pass credit towards first month |
| Evidence | 6% of first-time day pass visitors join within 30 days; 26 of 64 exit cards cite first-month commitment; repeated front desk questions |
| Problem | Visitors who like the space see joining as a large new outlay on top of what they just paid |
| Change | At checkout and on a desk card, offer the £18 day pass as credit against the first month if they join within 14 days. Control: no offer |
| Hypothesis | If we credit the day pass against the first month, then the share of day pass visitors who join within 30 days will rise, because the first month will feel like a continuation rather than a fresh commitment |
| Expected effect | From 6% to at least 12%; a smaller lift does not repay the credit given away |
| Primary metric | First-time day pass visitors who start a membership within 30 days ÷ first-time day pass visitors |
| Guardrails | Cancellations within two months must not exceed the current 22 in 151 (about 15%); day pass sales must not fall |
| Unit and audience | Randomised 50/50 by visitor at first booking, using the booking reference; returning visitors and corporate passes excluded |
| Run time | About 355 visitors per arm for this effect at 80% power and 5% significance, so roughly 710 visitors, or 8 whole weeks at about 420 a month, then 30 days to read joins |
| Decision rule | Ship if the variant reaches 12% or more with guardrails intact; iterate on the offer wording if positive but below 12%; abandon if flat or negative |
| Risks | Staff mention the offer to control visitors; a local event changes the mix. Brief the front desk and note events in the log |
| Owner | Front desk lead; manager reads the result |
Traffic check. The card works only because the expected effect is bold; a modest lift would need far more visitors than Hivemill sees in a year. That is why the team tests a meaningful change rather than a tweak.
After the result. A win ships the offer to everyone and moves idea B up the backlog. An iterate or abandon goes into the learning log with what the team now believes about commitment. Either way, the conversion rate, CVRShare of visits or clicks that convert., from day pass to member becomes a KPIA metric tied directly to an objective. reviewed weekly.
Common mistakes
- Scores without reasons. They cannot be challenged.
- Judging impact on a local metric. A homepage that looks better does not touch the day pass decision.
- An optimistic expected effect. It shrinks the sample on paper and the test never reaches a result.
- No guardrail on member quality. A credit could attract members who leave after a month.
- Writing the decision rule after the data. Then any result can be argued into a win.
Bring the card into the first growth plan capstone, or see the module in order on the beginner growth marketing level page.
