# MVP Experiment Plan: [Product or idea]

> A plan for testing a startup's riskiest assumption with the smallest
> possible product before building the full thing: the assumptions
> ranked by risk, the hypothesis in falsifiable form, the MVP type
> chosen, the metric and success and kill criteria fixed in advance,
> the run plan, results, and the persevere, pivot or stop decision.
> Structure aligned with the build-measure-learn loop and minimum viable
> product practice described in Eric Ries, *The Lean Startup* (2011).
> Replace every `[bracketed]` value and every row marked *Example*.
> The example is fictional.

<!-- Guidance: How to use this template
- Test one assumption per experiment. If you change the price, the
  segment and the channel at once, you will not know what worked.
- Write the success and kill criteria before you run anything, and get
  a co-founder to sign them off. Criteria set after the data arrives
  always pass.
- Pick the cheapest MVP that can falsify the assumption. Code is often
  not needed for the first test.
- Prefer behaviour (payment, sign-up, repeat use) over opinion ("would
  you use this?").
- The example comes from Verrowind (fictional), which tested whether
  independent bicycle shops would pay for stock forecasts before
  writing forecasting software.
-->

## Document control

<!-- Guidance: Keep one plan per experiment and link them from a running log so the team can see what has been learned. -->

| Field | Value |
|---|---|
| Document ID | [EXP-NNN] |
| Version | [0.1] |
| Status | [Planned / Running / Concluded] |
| Owner | [@founder] |
| Reviewers | [@co-founder], [@advisor] |
| Last updated | YYYY-MM-DD |
| Classification | Internal |
| Related | `startup/customer-discovery`, `business/lean-canvas` |

### Revision history

| Version | Date | Author | Change |
|---|---|---|---|
| 0.1 | YYYY-MM-DD | [@founder] | *Example:* Criteria agreed before launch |
| 1.0 | YYYY-MM-DD | [@founder] | *Example:* Results and decision recorded |

## 1. Assumptions and riskiest assumption

<!-- Guidance: List the assumptions your plan depends on, then score each for impact (if wrong, does the business fail?) and evidence (how much do you already know?). The riskiest assumption has high impact and little evidence. Desirability usually comes before feasibility and viability. -->

| ID | Assumption | Type | Impact if wrong (1 to 5) | Evidence today (1 to 5, 5 = strong) | Risk = impact x (6 minus evidence) |
|---|---|---|---|---|---|
| ASM-01 *(Example)* | Independent bicycle shops will pay monthly for stock forecasts | Desirability, viability | 5 | 1 | 25 |
| ASM-02 *(Example)* | Shop point-of-sale data can be exported weekly | Feasibility | 4 | 3 | 12 |
| ASM-03 *(Example)* | Shops can be reached at trade shows at low cost | Viability | 3 | 2 | 12 |
| ASM-04 *(Example)* | Forecasts beat the owner's own reorder habit | Desirability | 4 | 2 | 16 |

Riskiest assumption to test now: *Example:* ASM-01.

## 2. Hypothesis

<!-- Guidance: Write a falsifiable statement: who, what behaviour, how much, by when. If no result could prove it wrong, rewrite it. -->

We believe that [customer segment] will [observable behaviour] at
[threshold] within [period], because [reason from discovery].

*Example:* "We believe that independent bicycle shops with 1 to 3
locations will pay a 150 USD pilot fee for six weeks of hand-made stock
forecasts, with at least 6 of 12 invited shops paying within two weeks
of the offer, because 9 of 15 discovery interviews named overstock of
slow-moving parts as a top-three problem."

## 3. MVP type

<!-- Guidance: Choose the cheapest form that tests the hypothesis. Name what you will fake or do by hand and how you will tell participants. Be honest with participants about what is manual if they ask, and never take money for something you cannot deliver. -->

| MVP type | Tests | Cost and speed | Chosen |
|---|---|---|---|
| Landing page or fake door | Interest, message, channel | Very cheap, days | [ ] |
| Concierge (service by hand) | Willingness to pay, value delivered | Cheap, weeks, does not scale | *Example:* [x] |
| Wizard of Oz (looks automated, run by people) | Value and usability of the future product | Medium | [ ] |
| Pre-sale or letter of intent | Willingness to pay | Cheap | [ ] |
| Single-feature product | Retention on the core job | Higher, needs code | [ ] |

*Example:* Concierge MVP. A founder builds each shop's forecast in a
spreadsheet from its exported sales data and sends a reorder list every
Monday for six weeks. In parallel, a fake-door page tests the channel
(ASM-03) as a separate experiment, EXP-002.

## 4. Metrics, success and kill criteria

<!-- Guidance: One primary metric, decided in advance, with a success threshold and a kill threshold. Between them is "inconclusive": say what you will do then. With small samples the result is directional, not proof; state the sample size and the range of plausible values. -->

| Item | Value |
|---|---|
| Primary metric | *Example:* Shops paying the 150 USD pilot fee, out of 12 invited |
| Success (persevere) | *Example:* 6 or more of 12 pay |
| Kill (stop or pivot) | *Example:* 2 or fewer pay |
| Inconclusive zone | *Example:* 3 to 5 pay: rerun with 12 more shops and a changed offer |
| Secondary metrics | *Example:* Shops acting on at least half of weekly reorder lists; pilot shops asking to continue after week 6 |
| Guardrail | *Example:* Founder time under 10 hours a week |
| Sample and duration | *Example:* 12 shops, offer open 2 weeks, pilot 6 weeks |
| Budget | *Example:* Founder time plus 300 USD for data cleaning tools |

## 5. Run plan

<!-- Guidance: Who does what, when, and how data is recorded. Agree how participants' data is handled and deleted. -->

| Step | Action | Owner | Date |
|---|---|---|---|
| 1 | Invite 12 shops from discovery list with the same written offer | [@ceo] | YYYY-MM-DD |
| 2 | Collect fees and sales exports | [@ceo] | YYYY-MM-DD |
| 3 | Send weekly forecasts for 6 weeks, log usage | [@cto] | YYYY-MM-DD |
| 4 | Exit interview with every shop | [@ceo] | YYYY-MM-DD |
| 5 | Results review against criteria | Both founders | YYYY-MM-DD |

```mermaid
flowchart LR
    A["Riskiest assumption ASM-01"] --> B["Build: concierge MVP"]
    B --> C["Measure: shops paying, usage"]
    C --> D["Learn: compare with criteria"]
    D -->|"6 or more of 12"| E["Persevere"]
    D -->|"3 to 5"| F["Rerun with changed offer"]
    D -->|"2 or fewer"| G["Pivot or stop"]
    F --> B
```

## 6. Results

<!-- Guidance: Report the numbers against the criteria written in section 4, without moving the goalposts. Add what surprised you. -->

| Metric | Criterion | Result *(Example)* |
|---|---|---|
| Shops paying the pilot fee | 6 or more of 12 | 7 of 12 (58%) |
| Plausible range for the paying rate | Information only | About 32% to 81% (95% Wilson interval, n = 12) |
| Shops acting on at least half of reorder lists | Secondary | 5 of 7 |
| Asked to continue after week 6 | Secondary | 6 of 7 |
| Founder time | Under 10 hours a week | About 8 hours a week |

*Example:* Separate experiment EXP-002 (fake door): 36 of 400 visitors
from trade-show follow-ups booked a demo (9%, about 6.6% to 12.2% at
95%).

## 7. Decision and next experiment

<!-- Guidance: State persevere, pivot or stop, and why, in one paragraph. Name the next riskiest assumption. If you pivot, write a `startup/pivot-memo`. -->

| Item | Value |
|---|---|
| Decision | *Example:* Persevere: success criterion met |
| What we learned | *Example:* Shops pay for the outcome; owners trust a list more than a dashboard |
| Next riskiest assumption | *Example:* ASM-04, forecasts beat the owner's reorder habit, tested with a 12-week accuracy comparison |
| Follow-on work | *Example:* Price test and automated data import, feeding the revenue assumptions in `startup/financial-projections` |

## 8. References

<!-- Guidance: Cite the method and any internal research the hypothesis relies on. -->

- Eric Ries, *The Lean Startup* (Crown Business, 2011): minimum viable
  product and the build-measure-learn loop. Publisher and book
  information: https://theleanstartup.com/book
- Related templates: `startup/customer-discovery`,
  `startup/pivot-memo`, `business/lean-canvas`,
  `business/value-proposition-canvas`, `product/ab-test-plan`,
  `product/opportunity-solution-tree`.

## Standard and licence note

Structure aligned with the minimum viable product and build-measure-
learn practice described in Eric Ries's *The Lean Startup* (2011). The
book is named and cited only; no text is copied and all wording here is
original. Official text: the publisher's edition, purchase or licensed
access via https://theleanstartup.com/book. The example company and its
figures are fictional. Template licence: MIT.