What to do when your clients cannot be named
Confidentiality breaks the standard case study playbook. What credible proof looks like without logos.
In regulated and enterprise categories the standard proof playbook collapses. No logos, no named quotes, often no permission to describe the work in any detail at all.
The usual response is to give up and fill the website with claims instead. That is the worst available option, because claims without proof read as louder claims.
Anonymised specificity beats named vagueness
"A mid-sized insurer cut quote turnaround from three days to four hours" is more persuasive than a named client saying "great to work with, highly professional".
Specificity is what makes proof feel real. The name is secondary, and readers understand confidentiality better than marketers assume.
Four types of proof that survive an NDA
Shape, not identity. Sector, size band, geography, the before and the after. Enough for the reader to recognise themselves.
Your own working artifacts. A redacted framework, a screenshot with the names removed, the actual template. These prove competence directly rather than by testimony.
Third party credibility. Speaking slots, trade press bylines, industry bodies, awards. All of these are someone else vouching for you without needing a client to.
Volume statements. "Fourteen engagements in this sector since 2022" is true, checkable by anyone who asks, and names nobody.
Ask anyway, but ask narrowly
Most people who say no to a case study are saying no to the effort and the legal review, not to the idea.
Draft the quote yourself, keep it to two sentences, offer the anonymised version as the default and the named version as an upgrade. Acceptance rates roughly double when somebody only has to approve rather than write.
What not to do
Do not invent composites and present them as real. The market you are in is smaller than you think and somebody will ask which client that was.
If you must illustrate rather than prove, label it as illustrative. Honesty is cheaper than the alternative.
Readers forgive anonymity. They do not forgive vagueness.
The test for any piece of proof is whether a sceptical reader could check it if they wanted to. Anonymous and checkable is fine. Named and uncheckable is not.