A case study page isn't there to tell a nice story. Its job is to hand the reader something they can carry into their own meeting and repeat, close to word for word, and have it still hold up once someone else starts asking questions. If a visitor can't make your case for you without you in the room, the page has failed. It doesn't matter how good it looks.
This matters more in B2B than most sites act on. Research into B2B buying behaviour finds that business buyers read case studies far more than consumer shoppers do, largely because the person reading your page is rarely the person who signs off on the spend. One of the clearest examples of doing this well is almost rude about it: a corporate learning platform titled a case study around a client reaching two million external learners in under three years, and put that number straight in the headline, not the client's logo. Name the client but withhold the result, and you're asking the reader to do your selling for you.
The structure that works is problem, then solution, then result, in that order, with the result stated before the detail, not after it. Most case studies get this backwards: open with the solution, what we built, what we changed, and bury the number three paragraphs down, right where the skimming reader gives up. A page that leads with the outcome, restates it in the first line of actual text, and only then explains how it got there, respects something true about how these pages get read. Most readers decide in the first ten seconds whether to keep going.
Two readers sit on the other end of this page, and most case studies only write for one of them. The first is whoever found you, comparing a shortlist, early in the process, happy to read the whole thing start to finish. The second never visits the site at all. They get the page forwarded, or printed, or boiled down to one line in a Slack message, and they decide based on whatever survived that trip. A case study has to stand alone. For a large share of its actual readers, that's the only way it will ever get read.
What a case study has to prove is different from what a testimonial proves, and mixing the two up is the most common mistake on these pages. A testimonial is a feeling: someone enjoyed working with you. A case study is a rubric, specific enough that the reader can take the same checklist and run it against a competitor's site five minutes later. If your case study can't be turned into a checklist, it hasn't actually said anything, no matter how warm the quote underneath it is.
Most teams already have the raw material for one genuinely good case study sitting in an old project folder, and have published nothing, because writing the number down plainly means admitting what it actually was. That's the real barrier. Not design. Start with the one engagement that has a real, defensible number attached to it, and write that page properly, number in the headline, before touching anything else in the case studies section. A studio that has done this for other clients can usually tell you, in an afternoon, which of your projects is that one.