The Proof Grid: Nine Kinds of Evidence Every Offer Needs

Forty-six days

For years, runners and writers treated a mile in under four minutes as a wall. Some called it beyond what a human body could do.

On 6 May 1954 Roger Bannister ran it in 3 minutes 59.4 seconds.

Forty-six days later, another runner beat his time. More followed. Today it is a standard for good male middle-distance runners.

Bodies did not change in forty-six days. What changed was one piece of evidence: the thing was now known to be possible in the world.

That is a kind of proof most sales pages never offer. They skip straight to who else did it, when the reader has not yet accepted that it can be done.

Why testimonials are not enough

Ask most businesses for their proof and they hand you testimonials. Screenshots, star ratings, a quote with a first name.

That is social proof, the tendency to take other people's actions as a guide to what is correct. It is real and it works.

It also answers exactly one question: has someone else done this? A reader has two more.

Before that question comes a bigger one. Is this even possible now, in this market, with these rules? After it comes a more personal one.

Fine for them, but can I do it?

There is a second weakness. Testimonials are the winners. Readers know that, even if they could not name survivorship bias, the mistake of judging by the cases that made it through.

So they discount them.

The Proof Grid

Lay the three questions down the side of a page.

Possible in the world. Something changed that makes this achievable. A new rule, a new tool, a shift in how people buy.

Possible for others. Named people have done it. Here is how and what happened.

Possible for me. Someone with my limits, my budget, and my starting point can do it.

Now put three tenses across the top: past, present, future.

The past column explains why it was hard before, or how it used to work. The present column shows it working now, with dates. The future column gives a reason to think it will last.

Three by three is nine. The future column matters more than it looks. Nobody buys a path they expect to close, so a page that proves it worked last year and stops there leaves the reader's largest doubt untouched.

Grid Fillers: how to find evidence for each cell

Offers that feel airtight are rarely better at collecting praise. They have covered more cells. We call the people who build them Grid Fillers.

They start from the claim and work down. Take the headline promise and ask: what would have to be true for this to be true? Each answer is a cell that needs evidence.

"You can book twenty roofing jobs a month from ads." What has to be true? That people search for roofers.

That ads reach them at a cost that leaves margin. That a shop your size can answer the phone fast enough.

They use math for the world row. Market size, search volume, published benchmarks. A number from a public source is not an opinion, and it is the cheapest proof there is.

They match the "me" row to the reader. The most similar customer, not the most impressive. We cover why in the three stop-beliefs.

They count kinds, not pieces. Ten screenshots are one kind of proof. A named case, a public statistic, and a plain explanation of the mechanism are three.

Risk is measured in one. One source, one type, one tense. Each reads as fragile, however strong it is alone.

"I am new. I do not have nine kinds of proof."

You may have two customers and no case studies. The grid is more useful to you than to an established firm, because six of its nine cells do not need customers at all.

The whole top row is about the world. Public data fills it. Our own article on what a click costs by industry is top-row proof anyone can cite.

The "others" row can use public cases that are not yours, as long as you say so. "Here is what a published study of 2,000 accounts found" is honest and strong.

What you may not do is invent. A made-up testimonial is a legal problem as well as a moral one. The FTC requires that claims be truthful and backed by evidence, and it has rules against fake reviews.

The same applies to results. If a customer's outcome was unusual, say what is typical. A cell left empty and admitted is safer than a cell filled with something you cannot support.

What to read next, and why

Four-minute mile (encyclopedia entry): The record, the dates, and how fast it fell once someone had done it.

Social proof (encyclopedia entry): The research on why other people's choices persuade us, and where it stops working.

Survivorship bias (encyclopedia entry): Why a wall of winners convinces less than you expect.

Advertising FAQs: A Guide for Small Business (FTC guide): What counts as evidence for a claim, in the regulator's own words.

So the simple takeaway is this: draw a three by three grid, put your proof in the cells, and look at which ones are empty. Start with the top row.