Your Canvas Is Not Evidence

A few years ago, I opened a shared drive with a startup team and found what looked like a very well-run company. There was a Value Proposition Canvas, a Business Model Canvas, an ICP definition, a customer journey, several workshop boards and enough digital Post-its to wallpaper a small office.

What I could not find was a decision.

Nothing had been stopped. No customer segment had been dropped. No assumption had been killed. The roadmap looked remarkably similar to the one from three months earlier. The team had done a lot of work, but somehow none of the work had cost them anything.

I have thought about that moment quite a bit, not least because I have spent years using frameworks myself. I have taught them, built them, put them into presentations and stood in front of rooms explaining why this particular collection of boxes would help everyone see the problem more clearly. So when people roll their eyes at yet another canvas, I have some sympathy.

Frameworks are rightly in bad standing in parts of the startup world. There is an entire workshop industry built around the idea that if you put enough smart people in a room with enough Post-its, clarity will eventually emerge. Sometimes it does. Sometimes what emerges is a very colourful wall.

The problem is not the framework. The problem is what we expect it to do.

A framework cannot create evidence. It can only give evidence somewhere to go.

A canvas is not evidence. It is where you put the evidence you went out and got.

That distinction is why, despite everything I just said, I turned the ideas in Human Signal into six practical tools and decided to give them away for free.

I do not think of these tools as answers. I think of them as checklists against cowardice.

Founders are exceptionally good at asking difficult questions right up until the moment the answer threatens something they have spent eighteen months building. Then our standards become surprisingly flexible. A customer saying “This is really interesting” starts sounding like demand. Three enthusiastic pilot users become validation. A meeting with a large corporate becomes traction. If the corporate has a recognisable logo, it may even become a slide in the investor deck.

A good framework makes this kind of negotiation harder. Its job is to define the question before you know whether you will like the answer.

Take the Signal Stack, one of the tools. At its core, it separates what people say from what they actually do. Someone telling you they love your product is a signal. Someone returning every Tuesday morning and eventually paying for it is also a signal. They are simply not the same signal.

This sounds painfully obvious when written down. It becomes much less obvious when payroll is due in six weeks.

That is what a framework can do. It can force you to distinguish opinion from behaviour, interest from commitment, noise from signal. It can establish the rules before your brain starts renegotiating them.

And I suspect this matters more now than it did a few years ago. AI has made building dramatically cheaper. A weak signal no longer has to slow you down for very long. You can change the landing page tonight, reposition the product tomorrow, add the missing feature over the weekend and launch another experiment on Monday. When building is cheap, continuing to build becomes the easiest response to almost any problem.

Unfortunately, there is no framework that can make the decision for you.

A retention curve can show that customers keep leaving. It cannot make you accept that they do not care enough. An ICP exercise can show that the segment you started with is not responding. It cannot make you stop selling to them. A demand test can show that people praise the idea but refuse to pay for it. It cannot stop you from explaining why this particular test was probably flawed anyway.

Even a tool with a clear deadline becomes useless if nobody owns the decision. Without a human being willing to say yes, no or stop when the time is up, it is just another line in another document. And the world already has enough documents.

This is where frameworks end and judgment begins.

There is a simple test I like for this. Think about the last canvas, scorecard or framework you completed. Then ask yourself what changed because of it.

Did you kill a feature? Change your ICP? Rewrite the proposition? Stop an experiment? Move resources? Decide not to build something?

If nothing changed, perhaps you did not create a decision tool at all. You created a very nicely formatted record of a meeting.

That is not entirely useless. But it is not evidence, and it is certainly not progress.

The irony is that founders have never had more access to frameworks. There are thousands online, hundreds in books and, thanks to AI, an effectively infinite supply of new ones. Ask for a “proprietary seven-step framework for validating B2B SaaS demand” and you can have one before your coffee gets cold. Give it a pyramid and a catchy acronym and you are halfway to a consulting business.

Structure is no longer scarce. Judgment is.

The willingness to look at what customers actually did, compare it with what you hoped they would do, and allow the difference to change your next move is still painfully human.

That is why I am giving away the six practical tools built from the ideas in Human Signal. Not because I think another set of worksheets will save anyone, but because the right structure can make it harder to hide from the evidence you already have.

Frameworks do not create judgment. They make it harder to avoid using it.

So download them, print them, put them into Notion, run them through AI, cover a wall in Post-its if that makes you happy.

Just make sure that, occasionally, something dies.

The tools are free because they are worth nothing until they cost you a decision.

Download the Human Signal tools