How do we use statistics without sounding like every competitor?

A statistic earns its place in your messaging when it belongs to you and prices something the buyer already owns. The market-size figure, the average-cost-of-a-breach figure, the percentage from the industry report everyone in your category quotes ... those belong to the category, not to you, and any competitor can print the same sentence by tomorrow morning. Lead with a number your own work produced, or a number the buyer already reports on his own dashboard, and keep the public figure underneath as support.

Founders reach for the public number for good reasons. It's credible, it's free, and it makes the problem sound big. What it can't do is make the problem sound like his, this quarter, at a size that justifies a budget line.

Why do buyers skip past the number we lead with?

A public statistic argues that the category's problem is real. Your buyer's next question is whether it's real for him. A figure sourced from an analyst report can't answer that, because it was never about him in the first place. He reads it, agrees with it in the abstract, and keeps scrolling for the part that isn't abstract.

Then there's copyability. Every sentence on your homepage carries a switching cost for a competitor who wants to use it. Your customer's language, your own measurements, the details of what you built ... all of that costs real work to copy. A figure from a public report costs nothing. It's the most portable claim on your page, which makes it the weakest one for telling you apart. I call it the Commons Number: a figure held in common by everyone in the category, so quoting it funds the category rather than you.

The third effect only shows up when the number works. A buyer you've convinced the category problem is urgent goes shopping for the category. You paid for his education, and the shortlist he builds has eight names on it, increasingly assembled by an AI engine that can't attribute a shared public figure to any single vendor. Nothing on that page told the engine what to cite you for. That's the same mechanism behind why AI describes your company accurately and still never recommends you.

Cybersecurity is the clearest case, because there the shared alarm has become the pitch. Hundreds of vendors open with the same breach-cost and dwell-time figures from the same handful of annual reports, and a CISO has read all of them before he reaches your site. The statistic that was supposed to create urgency now signals that you sound like the last four vendors who called. We've written about what it takes to stand out when every vendor claims the same things, and the pattern travels well outside that category.

What did we find when we scanned 302 homepages for numbers?

We scored 302 company homepages on the 19-signal Brand Signal Score for the 2026 Messaging Index, capturing the verbatim line from the page behind every score. That gave us something unusual: a set of real homepage sentences we could read rather than summarize. For this article we scanned those captured quotes for hard numbers, then crossed the result against how each page scored on Cost of Inaction and on Owned Language.

213 of the 302 pages quote a hard number somewhere in their strongest evidence. Numbers are not rare and they are not useless: the pages carrying one average 20.45 out of 38 against 16.16 for the pages carrying none. The trouble starts one column over. Of those 213 pages with a number on them, 108 score a flat zero on Cost of Inaction, which means the page prints a figure and still never says what standing still costs this buyer. Across the whole index only 14 companies of 302 earn full marks on Cost of Inaction, while 144 earn full marks on Proof and Evidence. As a field we're good at proving and worst at making it matter.

The 14 companies in the top row all land in the Weak or Invisible bands. Not one of them reaches Approaching. They have the figure, they have no stake and no language of their own, and the number is carrying the whole argument by itself. This is one 302-company sample read as association rather than cause, and the detection is a text scan of quotes our scorers captured, so treat the counts as directional. The direction is consistent enough to act on.

What makes a number ours instead of the category's?

An ownable number comes from somewhere a competitor can't reach. There are three such places, and most companies are already sitting on at least one of them without treating it as marketing material.

1. 1Your own instrument. Something you measure that nobody else in the category bothers to measure. Ours is the Brand Signal Score, which is why this article can tell you 108 of 213 rather than quoting somebody's forecast. Building the instrument is the work; once it exists it produces a fresh receipt every time you run it.
2. 2The buyer's own operation. A number he already reports to his own board, restated in his units. Denied claims per month. Minutes per shift. Days of float. When the figure is one he can check against his own system, the argument stops being about your category and starts being about his quarter.
3. 3Your customer base in aggregate. What changed for the last dozen accounts, stated plainly and small enough to be believable. Small verified deltas beat transformation promises, which is why buyers stopped believing marketing claims in the first place.

Notice what all three have in common. Each one requires you to have decided something: what you measure, whose operation you're built around, which outcome you're accountable for. A Commons Number is attractive precisely because it lets you skip that decision. That's the same avoidance behind describing a pain the whole category describes and calling it problem-centric messaging.

Where should the public figure actually sit?

Keep it. Move it. The industry statistic does useful work as corroboration once you've already made a specific claim, and it does damage as the opening line. Above the fold, your hero has one job: tell a stranger who you're for, what problem you take, and what you believe about it. A borrowed figure spends that space arguing on behalf of everyone in your market. Our hero section guide covers what belongs there instead.

Two rules make the public figure safe further down the page. Attribute it and date it, so a reader and an AI engine can both see it's sourced rather than asserted. And put it next to your own number, in the same block, where the contrast does the work: here's what the field looks like, here's what we measured, here's the gap. A shared figure that frames an owned figure is proof. A shared figure standing alone is decoration.

How do we tell if our proof is doing any work?

Run two checks on every statistic currently on your site, starting with the hero.

1. 1The competitor check. Could your three closest rivals print this exact sentence without lying? If yes, it's a Commons Number and it can't be your lead. This is the Cover-the-Logo Test applied to one line rather than the whole page.
2. 2The unit check. Does the number use a unit the buyer's own operation produces, and does the sentence name what it costs him per week, per month, or per deal? A figure with no owner and no clock is the reason buyers keep calling good products a nice to have.

Most pages fail both checks in the same sentence, which is good news, because it means one rewrite fixes both. Swap the borrowed figure for the sharpest number you can defend, name whose dashboard it lives on, and let the analyst report corroborate you three paragraphs later. If you want an outside read on where your page stands, the Brand Signal Score scores your homepage on the same 19 signals we used for the index above.

“A number every competitor can print is a number that sells the category and leaves the shortlist to somebody else.”

Where do we start?

Start with the hero, because that's where the borrowed figure does the most damage and where one sentence changes the most. Pull the public statistic out, write the sharpest owned number you can actually defend in its place, and see whether the page still argues for the category or finally argues for you. If nothing you measure feels defensible enough to lead with, that's the finding, and it's a positioning problem rather than a copy problem ... the same one behind messaging that doesn't move buyers.

We work through exactly this kind of rewrite in the AI Workforce Clinic, and the first Clinic is free.