Only 1 in 10 Published Case Studies Has a Number In It

The direct answer: 10.7% of published B2B case studies (36 of the 335 we could actually open) contain no number anywhere: no percentage, no dollar figure, no time saved, no before-and-after. Put the other way, about 9 in 10 published case studies do include a number, which makes a quantified result the single most common element we measured, slightly ahead of naming the customer (84.8%) and including a direct quote (83.6%). We found this by crawling 1,000 B2B software companies on 2026-08-21 and opening every case study we could find.

This is first-party data from The Customer Proof Index 2026, our own crawl of 1,000 B2B software companies. Every figure in this post traces back to that report and its methodology. We are not repeating anyone else's survey number here, and we are not extrapolating past our own sample.

How we measured whether a case study "has a number"

Of the 954 B2B software companies we could reach, 582 (61.0%) have at least one written case study. That existence question is its own topic, covered in 4 in 10 B2B Software Companies Have No Written Case Study. This post is about a narrower and, we think, more useful question: among the companies that do have a case study, what is actually in it?

Of those 582, we could open at least one individual story successfully for 335 of them (239 had only an index page we could not resolve into individual stories, and 8 more had every story fetch fail). Those 335 companies, and the 1,348 individual stories we sampled across them, are the base for every figure in this post. For each story, we checked four things: does it contain a quantified result (a percentage, a currency figure, a multiplier, or a time unit), does it name the customer through a strict signal (not just a guess from the headline), does it include a direct quote of 8 or more words, and how many words does it run.

The median matters on its own: 858 words is long enough to hold a real narrative (a challenge, what changed, and a result), not a two-sentence blurb. Most companies that clear the bar of publishing a case study at all are not publishing something thin; the gap is specifically in whether that longer piece contains a verifiable number.

The surprising part: a number is more common than a name or a quote

We expected naming and quoting to be the easier bars to clear, since both are things a marketer can largely secure alone (a 10-minute conversation, a screenshot of a logo), while a quantified result requires the customer to disclose something more specific: an internal metric, a before-and-after they may not have calculated until asked, a figure their own manager might want reviewed before it goes on someone else's website.

The data says the opposite. 89.3% of case studies include a number, ahead of both the 84.8% that name the customer through a strict signal and the 83.6% that include a direct quote. Our read: once a company invests in writing a full case study at all, median length 858 words, they tend to build it around a number, because the number is often the reason the story felt worth telling in the first place. The 10.7% with no number are not cases where getting one proved unusually hard; they are cases where the team wrote the story anyway, without one.

That reframes the diagnosis. If your own case study has no number in it, you are not failing at something most competitors find easy. You are in a smaller group than "no name" or "no quote" would put you in, which is exactly why it stands out.

Which categories publish the most rigorous case studies?

We asked the same "no number" question inside each of the 7 category groups from our companion post on case study existence, restricted this time to companies that actually have a story to open.

This produces a genuinely different ranking than the existence table in our companion post. IT, dev and security software is the worst category for having a case study at all, but it is not the worst for rigor once it does publish one; Sales & CRM is. A company can be behind on one dimension and ahead on the other, which is exactly why we measure them separately instead of collapsing everything into one "proof score."

A case study without a number is not worthless

Nothing in this data says a case study needs a number to be worth publishing. It says that most published case studies do have one, so a reader (human or AI) has learned to expect it, and a story that never states a number reads as unusually vague against that expectation, even if the story itself is strong. If you are one of the 36 companies in our sample with no number anywhere, or you are drafting a case study right now and the customer will not give you one, we wrote a full guide to that exact situation: How to Write a Case Study When the Customer Won't Give You a Number. The short version is that a specific, honest qualitative before-and-after in the customer's own words still outperforms a vague quantified claim you cannot back up.

Fresh proof is rarer than complete proof

A number, a name, and a quote describe whether a single case study is well-built. Freshness describes whether your whole library is still current, and it is the dimension most companies fail hardest. Among the 335 story-bearing companies, only 120 carry a genuine, machine-readable publish date at all (a JSON-LD field, an article meta tag, or a time element; we deliberately exclude server and sitemap timestamps from this claim because they track redeploys, not edits). Among those 120, the median newest story is about 9.9 months old, and 44.2% have not published a new one in over a year.

That means a large share of B2B case study libraries are effectively frozen: whatever got published first is still the newest thing a buyer finds, sometimes a year or more later. A single strong, quantified, well-dated case study published this quarter is doing more work than three older ones with no date at all, simply because a buyer (or an AI answer engine) has no way to tell the old ones apart from something written three years ago.

We were strict about what counts as a genuine date on purpose. Two of the ten date sources we detected, a server's last-modified header and a sitemap timestamp, together accounted for most of the raw date hits in our crawl, and using them would have made the whole dataset look far fresher than it is: 28% of dates from every source combined land inside the last week, which reads as a redeploy schedule, not a wave of new case studies. A number that flatters the finding is not more useful than an honest one; it is just wrong in a direction nobody notices until they check.

What to do about it

If your case study has no number in it, go back to the interview notes or the call recording before you assume there is not one. Most "no number" stories are missing a number the customer mentioned in passing, not a number that never existed. Our guide to case studies without metrics covers exactly how to handle it when the customer genuinely will not or cannot give you a figure.

If your case study is unnamed or thin on quoting, these are the two elements that are entirely within your control, with no customer disclosure required beyond confirming you can use their name. An "About [Customer]" section and one clean, direct quote of a sentence or two closes that gap quickly.

If your newest case study is over a year old, you are one of a large share of companies with a stale library, based on the 44.2% figure above. Republishing is not the fix; a new customer conversation is. Even one fresh, dated story changes what a buyer sees first.

If you do not know whether your case study has a genuine publish date at all, check your page source for a JSON-LD datePublished field or an article:published_time meta tag. Most case study templates never add one, which is part of why only 120 of 335 companies in our sample have a genuine date to measure.