Lean Validation Board
The lean validation board (Validation Board in the original) is a one-page grid on which a team tests a product idea one assumption at a time. The team writes the target customer and their problem on it. It picks the riskiest assumption and the cheapest experiment that could disprove it: interviews, an offer presented with its price or a service delivered by hand. The team writes the minimum success criterion before the experiment. The verdict decides what comes next: a disproved assumption opens a new column, a confirmed assumption moves the team on to the next one. The board comes from the Lean Startup Machine workshops. It applies the principles of Eric Ries's Lean Startup.
Goal
The lean validation board serves a team that must prove a customer has a problem before it builds a solution. It holds the team's hypotheses about the customer, the problem and the solution. It names the assumption that the project depends on and that nobody has checked. It sets the cheapest test that could disprove that assumption and the minimum result that will confirm it.
A team takes the assumptions in its plan as true. It tests what is easy to test, then it builds. Grace Ng names the mistake that makes such a test useless: without a criterion set in advance, every positive result looks like validation.
The board supports one decision per experiment: does the tested assumption hold? The team delivers the filled-in grid and the experiment log. The grid shows a sponsor one column per pivot and the assumptions that the team has invalidated or validated. The log gives the evidence for each verdict.
Usage
When to use it
- New product or service whose customer and problem are still assumed: test the assumption whose failure would sink the project before anyone writes a line of specification.
- No data on the target segment: the experiment produces the first data.
- Internal service whose adoption is not assured: treat the target employees as the board's customers.
- Sponsor convinced in advance that the demand exists: a criterion written before the experiment lets a count decide rather than an opinion.
- Development budget request that needs evidence: the committee sees the validated assumptions, the criterion of each and the log.
- Limited discovery budget: an experiment runs on about ten participants in two weeks at most.
- Team spread over several sites: the grid fits on a shared digital whiteboard.
When not to use it
- Question that existing data can settle: there is no experiment to design. Apply hypothesis formulation and testing.
- Need and solution set by law or by contract: there is nothing to validate with customers. Track the project's assumptions with assumption and constraint analysis.
- Assumption about the repeated use of a live version: an interview or an offer cannot measure lasting use. Release a minimum viable product.
Description
The zones of the board
The board reads from top to bottom. The top zone tracks the project's hypotheses from one column to the next. In the bottom zone, the team designs the current experiment. The results zone files each tested assumption under invalidated or validated.
| Zone | What the team writes there | Rule |
|---|---|---|
| Columns | Start, then one column per pivot. The original template provides four pivot columns. | A pivot column carries only the hypothesis that changes. |
| Customer hypothesis | The target segment. | Narrow enough for the team to reach about ten of its members. |
| Problem hypothesis | "I believe [customer x] has a problem [achieving goal y]." | A specific difficulty that the customer meets. |
| Solution hypothesis | The team's answer to the problem. | Empty until the problem is validated. |
| Core assumptions | Any assumption whose failure sinks the project. | Listed with the whole team and a few people from outside the project. The team clears the zone after each experiment. |
| Riskiest assumption | The core assumption with the highest uncertainty. | One per experiment. |
| Method | Exploration, pitch or concierge. | The cheapest one that could disprove the assumption. |
| Minimum success criterion | The weakest result that the team accepts as validation: "6 people out of 10 [exhibit a given behaviour]". | Written before the experiment. |
| Result | An Invalidated area and a Validated area. Each holds six numbered boxes. After each experiment, the team files the note with the tested assumption in the area that matches the verdict. | Invalidated: pivot on at least one hypothesis in the column. Validated: test the next riskiest assumption. |
The board works at two levels and gives each level its own word. The customer, problem and solution hypotheses describe the project. The core assumptions are the conditions that these descriptions depend on. The riskiest assumption comes from this second list.
Writing the customer and the problem
The team fills in the Start column first. You can reuse the customer and the problem from your own Product Vision Board. Grace Ng proposes a form for the problem: "I believe [customer x] has a problem [achieving goal y]." The customer is a segment that the team knows how to reach: "self-employed physiotherapists in the canton of Vaud" rather than "health professionals". Among the design mistakes, Ng lists a problem framed too broadly or seen from the company's side ("customers don't use our service enough"). She also lists testing a sentiment ("customers like the idea") in place of a specific problem.
The original template prescribes leaving the solution cell empty until the problem is validated. A team that writes its solution at the start designs its experiments to confirm it. On a platform that serves two customer groups, the template has the team validate the riskier side first.
Choosing the riskiest assumption
The team then lists the core assumptions: all those whose failure would sink the project. Ries calls the elements of the plan that everything else depends on leap-of-faith assumptions. They are also the riskiest.
Among them, the team picks the one with the highest uncertainty. Ries gives the same rule: test the riskiest assumptions first. Only one assumption goes into the experiment. If the team tests two assumptions together, a failure does not say which one fell.
A team on its own overlooks the assumptions it takes for granted. Ng has the whole team and a few people from outside the project list the assumptions. These outsiders resist confirmation bias better. The participants then pick the riskiest assumption by dot voting.
Choosing the method: exploration, pitch or concierge
The original template asks for the cheapest way to test the assumption. It offers exploration, pitch and concierge. Grace Ng describes them in two articles from 2014.
| Method | Question tested | What the customer does | Cost to the team |
|---|---|---|---|
| Exploration | Does this customer have this problem? | Describes in an interview how they work and what they struggle with. | Low: about ten interviews. |
| Pitch | Does the problem matter enough for the customer to commit? | Responds to an offer presented before the product exists: signs up, leaves contact details or pays in advance. | Low to medium: an offer page with a price or a pitch delivered in a meeting. |
| Concierge | Can the team deliver the expected service? | Receives the service that the team delivers by hand. | High: team time for each customer served. |
The pitch and the concierge present a solution. They therefore come after the problem is validated. Exploration opens every column where the problem is still to be proven. See Interviews for how to run an exploration interview.
Ries counts the smoke test among the forms of minimum viable product: an advertisement or a page that measures interest before the product exists. The pitch on the board is a smoke test. Ries also describes the concierge. In his Food on the Table example, the founders served a first customer by hand for USD 9.95 a week. Ries points out that this personalised service is a learning activity. A successful concierge covers a handful of customers. Its success says nothing about the cost of serving a thousand.
Writing the minimum success criterion
The minimum success criterion is the weakest result that the team accepts as validation. The team writes it before the experiment and dates it. For Ng, defining success is the decisive step before an experiment. A criterion lowered after the result turns a failure into a validation.
Ng gives the form: "I expect [x number of people] out of the [y number of people in the experiment] to [exhibit behavior z]." The criterion counts people who act: six physiotherapists out of ten who name a problem unprompted, four out of ten who sign up for a paid trial. Ries contrasts these actionable metrics with vanity metrics, which rise whatever the team does. A visit count or an "I like the idea" settles no assumption.
"If you cannot fail, you cannot learn," Ries writes. Ash Maurya asks for falsifiable hypotheses: the team defines in advance the conditions under which the hypothesis is proved or disproved. "Physiotherapists are interested in the project" cannot fail. "Six physiotherapists out of ten name missed appointments among their top three worries" can fail.
Running the experiment and reading the verdict
Maurya limits an experiment to two weeks. The team keeps the experiment log off the board: one row per experiment with the assumption tested, the method, the criterion, the result, the verdict and the learning. On the board, the team keeps only the note with the tested assumption and files it in the area that matches the verdict. The original template then prescribes clearing the experiment zone.
The verdict follows a rule written on the board. An invalidated assumption forces a pivot on at least one hypothesis in the column: the team opens the next column and writes the hypothesis that changes in it. A validated assumption moves the team on to the next one in the same column. The team draws up a new list of core assumptions and picks the riskiest.
Pivoting on what the experiment taught
For Ries, a pivot is a structured course correction that tests a new hypothesis about the product, the strategy and the engine of growth. For Maurya, a pivot with no learning behind it amounts to trying things at random to see what sticks. The new column therefore builds on the log of the invalidated experiment.
The board's rule covers a single experiment. Ries rejects any formula for the decision to pivot: human judgement stays part of it. He gives this decision to a meeting that reviews the results over time. The meeting recurs every few weeks to every few months. In practice, the board's verdict says whether an assumption held. The meeting decides whether the strategy changes.
Neighbouring techniques
The pivot-or-persevere meeting belongs to Eric Ries's build-measure-learn loop. Ries plans this loop in reverse: what to learn, then what to measure, then what to build. On the board, the team writes what it must learn (the assumption) and what it must measure (the criterion) before it builds. The board records each turn of the loop.
A minimum viable product (MVP) released to early adopters measures the use of a live version. For Ries, an MVP ranges from the smoke test to the early prototype: the pitch and the concierge are MVPs. The team uses the board before any version is released to choose the assumption to test and the cheapest test.
Hypothesis formulation and testing tests a null hypothesis and an alternative hypothesis against an existing data set. The analyst sets the significance level before reading the data. The board creates its own data through the experiment. Its threshold counts people out of about ten. Ng considers statistical significance secondary on so narrow a segment. Both techniques fix the decision rule before the result.
Real options tell the team when to commit: at the last responsible moment. Before that commitment, the team uses the board to test the assumption the commitment depends on. The link between the two techniques is a practitioner's reading: an option still open leaves time to test the riskiest assumption.
Ash Maurya's Lean Canvas adapts the Business Model Canvas and describes the whole plan on one page. In Running Lean, Maurya describes the sequence: document the plan, find its riskiest parts, then test them. The board is the tool for the last step.
AI considerations
To list the core assumptions, a language model takes the role of the outsider that Ng recommends. From a description of the project, it proposes core assumptions that the team had not written. It flags those that cannot fail and rewrites them as counts: "X people out of Y [exhibit a given behaviour]". It shares the team's bias as soon as the team shows it the solution: the analyst gives it the problem alone.
To prepare the exploration, it drafts an interview guide from the problem hypothesis and spots the questions that suggest the answer. To analyse the interviews, it groups the interview notes by theme. The analyst then counts the people who meet the criterion again by hand, since the verdict rests on that count.
The answers that a model invents for a fictional customer remain guesses. The board collects the behaviour of real customers. The criterion and the verdict stay with the team. Asked whether an assumption is validated, a model tends to build a favourable story from ambiguous results. The criterion written in advance neutralises that story. Interview notes identify people. They fall under the Federal Act on Data Protection (FADP). The analyst anonymises them before sending them to an AI service hosted by a third party.
Examples
A team in the canton of Vaud is preparing software for self-employed physiotherapists. After two explorations, its board shows one pivot column and both outcomes of a verdict.
| No. | Column | Assumption tested | Method | Criterion | Result | Verdict | Learning |
|---|---|---|---|---|---|---|---|
| 1 | Start | Missed appointments rank among their top three worries. | Exploration: 10 interviews | 6 of 10 | 2 of 10 | Invalidated | 7 of 10 name insurer billing. |
| 2 | Pivot 1 | Insurer billing takes them more than two hours a week. | Exploration: 10 interviews | 6 of 10 | 8 of 10 | Validated | |
| 3 | Pivot 1 | They pay CHF 40 a month. | Pitch: offer page with price | 4 of 10 sign up for the paid trial | Pending |
The Pivot 1 column rewrites only the problem. The new problem comes from row 1 of the log, where seven physiotherapists out of ten named insurer billing.
Visualisations
The filled-in board keeps the assumptions and the verdicts. The log keeps the evidence. The tables of zones and methods are the board's legend. The two arrows on the filled-in board carry the verdict: an invalidation opens a column, a validation stays in the same column.
Cost
| Phase | Level | Justification |
|---|---|---|
| Preparation | Low | A session of one to two hours to write the customer and the problem, list the core assumptions with the whole team and a few people from outside the project, then vote. |
| Execution | Medium | An experiment of two weeks at most on about ten participants. The concierge costs the most, since the team delivers the service to each customer. |
| Documentation | Low | The board itself. The experiment log holds one row per experiment. |
Tooling
The original template prints as a poster of about 91 by 61 cm. Each box takes a single note. The team writes it in capitals and in five words at most. The template is free to use for entrepreneurs, consultants and executives. A software company must buy a licence for it.
A digital whiteboard (Miro, Mural and their equivalents) replaces the poster for a distributed team. A spreadsheet holds the experiment log: one row per experiment with the assumption tested, the method, the criterion, the result, the verdict and the learning.
For the pitch, a page builder (Webflow, Carrd) and a form (Typeform, Google Forms) carry the offer, the price and the sign-up. For exploration, a video-conferencing tool records the interviews with the interviewees' consent. The concierge needs no tool of its own: a phone, a calendar and an invoice are enough.
Sources
The board comes from the Lean Startup Machine workshops. It applies the principles of Eric Ries's Lean Startup. The texts of Grace Ng and Ash Maurya refine the design of the experiments.
- Lean Startup Machine, Validation Board, 2013: the zones of the board, its instructions, the verdict rule and the licence terms.
- Grace Ng, A Guide To Validating Product Ideas With Quick And Simple Experiments, Smashing Magazine, 10 April 2014: exploration, pitch and concierge; the form of the problem hypothesis and of the criterion; the design mistakes of an experiment; statistical significance on a narrow segment.
- Grace Ng, Why Lean Startup Experiments Are Hard to Design, Lean Enterprise Institute, 7 May 2014: the criterion as the decisive step; the list of assumptions drawn up with the whole team and a few outsiders; dot voting.
- Eric Ries, The Lean Startup, Crown Business, 2011: the introduction to Part Two for planning in reverse; chapter 4 "Experiment" for the experiment that can fail; chapter 5 "Leap" for leap-of-faith assumptions; chapter 6 "Test" for the forms of MVP, the smoke test and the concierge; chapter 7 "Measure" for the order of the tests and vanity metrics; chapter 8 "Pivot (or Persevere)" for the definition of the pivot and the meeting.
- Ash Maurya, Running Lean, O'Reilly, 3rd edition, 2022: the overall process from the documented plan to the test of its riskiest parts.
- Ash Maurya, How to Identify a Lean Startup, 15 September 2010: the falsifiable hypothesis.
- Ash Maurya, The Art of the Scientist, 9 June 2016: the pivot grounded in learning.
- Ash Maurya, Why Lean Canvas versus Business Model Canvas, 22 February 2024: the Lean Canvas and what it takes from the Business Model Canvas.
- Ash Maurya, The Art of Experiment Design, 14 April 2024: the length of an experiment.

