Why 60% of UK Engineering and Manufacturing Websites Have No Case Studies to Show

Sixty of the hundred UK engineering and manufacturing websites Nebula audited in August had nothing on the page you could point to and call a case study. The check itself was generous: it only looked for the phrase "case studies" appearing anywhere in the public text, not whether the work behind it was any good. Even against that low bar, fewer than half the sample cleared it.

That's not a capability problem. Most of the businesses in the sample will have solved something genuinely difficult in the last twelve months: a tolerance nobody else could hold, a lead time cut in half, a part redesigned after a supplier failure elsewhere. The knowledge exists. It just tends to live with the engineers, the project managers and whoever answered the phone on the day, rather than on a page a prospective customer can read before they ever pick up the phone.

There's a reasonable objection to raise here, and it's worth taking seriously rather than skipping past. TrustRadius's 2026 buying research, based on 1,862 technology buyers surveyed in January, found reviews and prior experience carry more weight than vendor-written content, and that only 13% of buyers now bother with analyst reports at all, down 63% since 2022. Read quickly, that looks like an argument against spending time on case studies: if buyers trust strangers' reviews over a company's own account of its work, why write one?

The answer is that software buyers have somewhere to go for the stranger's opinion. G2, TrustRadius itself, Capterra: a buyer evaluating CRM software can read a hundred unfiltered reviews before ever speaking to a salesperson. A buyer evaluating a subcontract manufacturer for a five-axis machining job has no equivalent. There is no review site for jig and fixture design, and precious few for specialist fabrication or test-rig engineering. In that gap, a case study is doing a job that in software would be handled by someone else's honest write-up. It carries more weight here than the TrustRadius data would suggest for SaaS, not less.

That's consistent with the wider B2B picture too. Content Marketing Institute's 2025 benchmarking study, based on 980 B2B marketers surveyed between June and August 2024, rated case studies and customer stories as the second most effective content format overall, behind only video. Fifty-three per cent of respondents said case studies deliver strong results for them, which is a high number for a format that, per Nebula's own audit, forty out of a hundred engineering websites appear not to have at all.

The audit also found something worth acting on directly: only twenty-six of the hundred sites paired sector or industry language with case-study evidence. That combination matters more than either signal alone. A page that says "we work in aerospace" answers one question. A page that adds "and here's a tolerance we held on a validated production part" answers a different, more useful one. The first tells a buyer the company might be relevant. The second gives them a reason to believe it.

None of this requires publishing a named client's confidential production data. Most engineering NDAs allow a well-written anonymised account: the technical problem, the constraint that made it hard, the approach taken, and the measurable result. "A precision component moved from prototype to repeat production while holding a tolerance the original supplier couldn't meet" tells a buyer more than a page of equipment lists, and it does so without naming anyone.

For a technical SME with no marketing team, the practical next step is small. Pick three or four completed jobs that would make sense anonymised. Write down the problem, the constraint, the approach and the result in the same plain language an engineer would use to explain it to a colleague, not the language a brochure would use. Put at least one of them somewhere a prospective buyer can find it before they ever ring up. That's the gap the audit found, and it's a smaller job to close than most businesses assume.