Rory Sutherland tells a story about HS2.¹ The brief went to engineering firms, so the question became: how do we reduce travel time between London and Manchester? They optimised for speed. They built models. They produced a single defensible answer nobody could argue with.

What if the brief had gone to Disney? Disney would have rewritten the question. How do we make the journey so enjoyable people feel stupid going by car? Free Wi-Fi. A bar car. A ball pit for kids. Multiple answers, all subjective, none provable with a spreadsheet, all closer to what humans actually want.

The engineers' question had one right answer. Disney's had dozens. That is why it never got asked. Open-ended questions require judgment. Judgment can be challenged. Numbers cannot. So we close down the solution space until what remains is a single figure that nobody can argue with — not because it is the best answer, but because it is the safest.

Climate technology has done this at industrial scale.

47% emissions reduction. Sub-$100 per ton. 3.2 gigatons of CO₂ avoided by 2040. I have now looked at more than a hundred climate companies. Nearly all of them lead with a number. The number is always accurate. Always defensible. And it almost never lands.

The numbers matter. But they answer the wrong question. "How do we prove our impact?" produces whitepapers, spec sheets, and investor decks that other engineers nod along to. "How do we give someone a reason to pay attention in the first place?" produces loyalty, talent, press, and the kind of recognition that compounds rather than decaying after a funding announcement. The sector chose the first question — unanimously, instinctively — because every audience it needs to convince rewards defensibility. Investors. Regulators. Procurement boards. Grant committees. The entire communication apparatus tilts toward proof.

Proof is not persuasion.²

The founder rehearses a perfect pitch for the technical audience and goes blank when a journalist asks "but why should anyone care?" The marketing hire arrives, reads the website, the deck, the LinkedIn, and spends three months trying to reconcile three different companies into one. The enterprise buyer understood the product in the meeting and could not explain it to their procurement committee the next morning. Anyone who has sat through that silence — the moment between a technically correct presentation and the room's failure to move — knows the feeling. These are not failures of intelligence. They are failures of architecture.

Sutherland has a name for this: closing down the solution space. Pretend an open-ended question has a single right answer. Solve for that answer. Now nobody can blame you, because you did not make a decision — you followed the data.³ The climate sector has closed down its entire communication to the version that survives a board meeting. It works in the boardroom. It fails everywhere else. A Series B company that spends six months with a story that does not convert has burned through a quarter of its runway before the narrative problem is even named.

A number without a story is a fact. A number inside a story people already care about is proof. The difference is architecture — the structural work that gives someone a reason to look at the number in the first place. Lead with the proof before anyone cares, and the proof just sits there. Accurate, defensible, ignored.

I do not think anyone has fully solved this yet. The companies that come closest tend not to lead with climate at all — they lead with cost, performance, or status, and let the climate case arrive after the audience is already paying attention.⁴

The question was never how to prove impact. It was how to make impact worth paying attention to. That question has many right answers. Which is exactly why almost no one is asking it.