Skip to content
All articles
Strategy8 min read

A Business Model Canvas Is Not Evidence: How to Test the Nine Boxes

Turn a completed Business Model Canvas into a ranked set of business hypotheses, tests and decision thresholds.

For: Founders, innovation teams and strategy leaders

A team can complete a Business Model Canvas in an hour. That does not mean it has a business model.

The nine blocks create a compact explanation of how value is intended to move through a system. In a new model, most entries are assumptions. In an established model, familiar statements may still be unsupported or outdated.

The practical question is:

Which assumption would damage the model most if it were wrong, and what evidence could change our decision?

Start with one coherent model

A canvas should represent one model option at a defined level.

Weak scope:

Our company in the future.

Usable scope:

A regional subscription service for preventive maintenance of packaging equipment at small factories over the next 18 months.

Do not combine several pricing models, unrelated segments and alternative channels on one canvas. The result may look rich while describing a business that cannot exist.

Trace one customer across the canvas

Take a priority segment and ask:

  1. What important problem or progress defines the value?
  2. How does the customer discover, evaluate, buy and receive it?
  3. What relationship is required before and after purchase?
  4. What creates revenue?
  5. Which activities, resources and partners deliver the promise?
  6. What costs follow from those choices?

If the answers do not connect, the model is incomplete even when every box contains text.

Use evidence labels

Label each important claim:

  • Observed: supported by direct evidence.
  • Assumed: plausible but not tested.
  • Contradicted: current evidence points against it.
  • Unknown: no meaningful investigation yet.

Examples:

ClaimLabel
Five customers paid for a three-month pilotObserved
Buyers will pay €1,500 per month at scaleAssumed
Every interviewed plant has remote-compatible equipmentContradicted
Sensor supplier volume pricingUnknown

The labels do not have to be perfect. Their purpose is to prevent team confidence from becoming invisible proof.

Find cross-block assumptions

The most dangerous assumptions often connect several blocks.

For a maintenance subscription:

One technician can support 18 plants if remote monitoring prevents one in three visits.

This claim affects:

  • Key Activities;
  • Key Resources;
  • Cost Structure;
  • Value Proposition;
  • Revenue Streams.

If remote diagnosis does not prevent travel, the service may miss its response promise and lose its target margin at the same time.

Cross-block assumptions deserve more attention than isolated wording decisions.

Prioritise by consequence and evidence

Use a simple matrix:

Consequence if wrongEvidence strengthAction
HighWeakTest first
HighStrongMonitor
LowWeakTest cheaply or defer
LowStrongRecord and move on

The team should not test what is easiest merely to create activity.

Replace “research” with a decision test

Weak next step:

Research willingness to pay.

Better:

Offer a paid eight-week pilot to ten qualified plants. Continue if at least four sign, gross delivery cost remains below 60% of pilot revenue and response performance stays within the promised range.

The better test includes:

  • a defined population;
  • behaviour with a cost or commitment;
  • operating evidence;
  • a threshold;
  • a decision.

Interviews remain useful, especially before an offer is clear. They are not the same evidence as payment, adoption or retention.

Mini-case: the forecasting subscription

A team believes independent restaurants will pay €99 per month for a reviewed cash-flow forecast.

Its canvas includes:

  • accountants as a partner channel;
  • automated bank feeds as a key resource;
  • analyst review as a key activity;
  • 80 accounts per analyst as a cost assumption.

The team proposes a large marketing campaign.

A better next step is a concierge pilot with varied real accounts. Measure:

  • data cleaning time;
  • analyst time;
  • forecast usefulness;
  • continued payment;
  • referral behaviour.

This one test informs customer value, channel credibility, capacity, costs and revenue.

Keep versions and alternatives separate

Date every canvas. Keep:

  • current-state model;
  • target option A;
  • target option B;
  • evidence and decision log.

When new evidence appears, record what changed. Otherwise the team may update the visual while forgetting why the decision changed.

Practical next step

Select five claims from your current canvas. For each, write:

  1. evidence label;
  2. consequence if wrong;
  3. smallest credible test;
  4. decision threshold;
  5. owner and review date.

Start with the weak-evidence claim that connects the most important blocks.

References

  1. Strategyzer. “The Business Model Canvas.” Official tool page (opens in a new tab).
  2. Osterwalder, A., & Pigneur, Y. Business Model Generation. Wiley, 2010.

Continue with the tool

Use the Business Model Canvas guide to build a coherent model and rank its critical assumptions.