Most of the numbers B2B companies publish don’t come from research they did.

That’s why B2B companies use so many of them. A number makes a problem feel real and gives readers something to repeat. But publishing a fact isn’t the same as publishing someone else’s fact.

When the evidence is yours, you can come back to it: explain it, update it, use it again when the market asks a related question. When it’s borrowed, the authority stays with whoever did the work.

This census looks at that difference. We checked where the main number on each research page came from: the company, research it paid for, or another publisher. Citing good work is fine. It just can’t be the whole research programme.

The calendar is full, a release is coming and sales wants proof. The quickest fix is an industry statistic, explained well and placed next to the company’s point of view. The page is easy to publish. But it starts with someone else’s evidence.

That leads to three problems:

• The company has little of its own to say.
• Its own research gets one launch, then goes quiet.
• Research starts to look like it needs a big survey budget.

A fresher statistic fixes none of them.

A borrowed number isn’t a bad number. A good study sets the context and shows how big a problem is, and good research credits the work that already exists.

The trouble starts when the borrowed fact becomes the centre of the page. You can add an explanation, a product point or an opinion around it. You can’t make the finding yours by putting it on a new page.

The second number shows this is more than the odd citation. At 1 company in 4, at least half the numbers come from someone else. The page may be useful, but its best evidence sends the reader somewhere else.

A flagship report is where original work is easiest to see. It gets a launch, a title and a clear reason to exist.

But the launch is one moment. The questions the report answered keep coming back: in sales calls, in follow-up articles, and whenever the news makes the research relevant again.

Not every later page needs to repeat the report. But the research should keep shaping what the company can say next.

Surveys are what most people picture when they hear “original research”: fieldwork, a new dataset, a report. That route works. It’s also the one that makes research feel expensive, rare and separate from the rest of the business.

The census shows another route in use: numbers from what the company already sees in its sales, its customers and its product.

Not every internal metric is a finding. A product count isn’t research, and a private number isn’t useful just because nobody else has it. The work is asking a question your data can answer about the world outside your company.

Borrowed numbers can explain why a subject matters. Your own give you something new to say about it.

You don’t need to run a survey, or turn every internal metric into a public claim. The standard is simpler: publish a fact that starts with your company’s work, one you can explain and use again.