Most anonymous case studies die in the redaction, not the refusal.
You've got a real pilot. 12 weeks, a health system or a group practice, a metric the clinical team actually agreed to track, and a lead clinician who said something genuinely good about your product at the wrap-up call. Then the comms team comes back and says the organization can't be named.
So the write-up gets sanded down to 400 words about "a leading healthcare provider" that could describe any of 300 organizations, with a benefit statement in place of a number and a quote attributed to "a customer."
That version is worse than publishing nothing, because it tells your reader you have something to hide.
Here's what your reader is actually checking when she opens a case study: whether you've done this before, somewhere close enough to her own setting that your experience transfers to her. A recognizable logo answers that question in half a second. Every other detail in the piece answers it too, if you leave the details in.
Work out who's actually blocking it
Founders tend to hear "we can't be named" as a single, final no. It's usually one of 4 different objections, and they have different workarounds.
Endorsement. Public bodies and most non-profit health systems have a rule against appearing to endorse a commercial supplier. This one is real and it's rarely negotiable. But it's a rule about endorsement, and a factual description of work that happened is a different thing from an endorsement. Ask whether a neutral, factual account with no promotional language would clear.
Procurement timing. If a full contract is still going through a competitive process, naming the relationship publicly creates a paper trail that makes the procurement team's life harder. This is usually a "not yet" rather than a "never." Ask when it becomes a yes and calendar it.
Patient data. Anything that could reidentify a patient, a cohort, or a specific clinic on a specific date. Non-negotiable, and you should be the one raising it first. Founders who bring this up before the client does buy themselves a lot of credibility.
Approval fatigue. The person who'd have to sign it off has 40 other things in front of her and no reason to spend political capital on your marketing. This is the most common blocker and the easiest one to fix. Send a finished, factual draft that takes 4 minutes to read and needs one word back.
Ask which of those it is. The answer changes what you write next.
Describe the organization by its shape
When the name goes, replace it with the variables a buyer would use to judge whether the setting matches hers.
Compare these two openings.
"We worked with a leading UK healthcare organization to improve patient engagement."
"We worked with a 14-site optometry group running Optix across all sites, with 4 sites offering myopia management and no central clinical lead coordinating between them."
The second one identifies no one and tells you far more. A reader who runs a similar group recognizes her own operation in it. A reader who runs 3 independent practices knows immediately this isn't her, which is also useful, because you'd rather she self-select out than book a call and churn.
The variables worth keeping are the ones that change how the work goes: number of sites, patient volume, the systems already in place, who owns the workflow, and whether there's a clinical champion internally. Most of that survives any anonymity requirement, because none of it identifies the organization on its own.
Say what was broken in operational terms
"Low engagement" is a category, not a problem. "Recall letters went out on paper, roughly 30% of them came back undeliverable, and the front desk was rebooking by phone between patients" is a problem, and it's one a practice manager reading it will feel in her chest.
Operational detail is also the safest thing to publish. It describes a workflow, not an institution.
The named human inside the unnamed organization
The institution and the individual are two separate permissions, and they don't always move together.
Sometimes the organization can't be named but a clinician will speak in a personal capacity, with her own name and credentials and a line clarifying she's speaking for herself. That's a strong outcome. A named ophthalmologist with a verifiable professional record carries real weight, and readers can look her up.
Sometimes it goes the other way. The clinician doesn't want her name attached but the organization is fine being described by type. Then you want the most specific title the client will allow: "clinical lead for ophthalmology at the trust" tells a reader who was in the room. "A customer" tells her nothing and quietly suggests you made the quote up.
Ask for the title even when you can't get the name. It's a small ask and it's the difference between a quote that reads as evidence and one that reads as filler.
Ask for permission at kickoff, not at the end
The reason permission conversations go badly is that they happen after the fact, when the client's incentive to help you has already been spent.
Move it to the kickoff call, where it's just one more thing being agreed alongside scope and timelines. You want 4 things written down before the work starts:
- What gets measured. One metric both sides accept, with a baseline captured before you touch anything.
- Who writes it up. You draft, they review. Never the reverse, or it won't happen.
- What level of attribution is available. Full name, organization type only, named individual, or a titled but anonymous quote. Get the ceiling agreed early.
- Who signs off. A named person, ideally someone in the room on that call, so you're not hunting for an approver in month 6.
Founders who do this get a publishable case study. Founders who don't get a good result they're allowed to mention verbally on sales calls and nowhere else.
Worth saying plainly: agreeing this at kickoff costs you nothing with a client who's already excited about the pilot. Asking 6 months later, after a rocky implementation and a change of contact, costs you the whole thing.
Make the approval email easy to say yes to
When you do send the draft over, the shape of the request matters more than the writing.
Send the finished piece as an attachment, with the redactions already applied, and say in the first line exactly what you're asking for. Something close to: "This is written up with your organization unnamed and described only by size and setting. I need a yes or a redline by Friday, and if I don't hear back I'll assume it's a no and shelve it."
Two things happen when you write it that way. The reviewer can see you've already done the anonymizing work, so she isn't being asked to protect her employer from you. And the shelf-it default removes the anxiety that makes people leave the email unopened for 3 weeks.
Give her the specific lines you'd like to change if she pushes back, ranked. Most reviewers will grant the top one or two if the rest of the document is clearly safe.
When even the numbers are blocked
Occasionally you'll get a client who'll allow a general description and nothing quantified at all. You can still publish, and the thing to publish is the method.
Write up how you approached the problem. What you looked at first. The assumption you had going in that turned out to be wrong. The decision point where you chose one path over another, and why. The thing you'd do differently on the next deployment.
A buyer reading that learns something more durable than a percentage. She learns how you think, which is what she's really trying to work out before she puts her name on a recommendation to her board.
This kind of piece has a second use. It's readable by someone who's never heard of your client anyway, and most of your readers haven't. The logo mainly reassures the small subset who recognize it.
Where the anonymous case study earns its keep
Anonymous proof does its best work mid-funnel, in the gap between the first call and the internal decision.
It's what you send after a discovery call so your champion has something concrete to forward. It goes in the pack that sits alongside the security review. It answers the question a skeptical clinical director asks in a meeting you're not in, which is usually some version of "has anyone actually run this in a setting like ours."
Investors read them too, and they read them for something else. An investor scanning a data room is checking whether the founder has a repeatable account of how the product gets adopted inside a real clinical workflow, because that's what determines whether the next 10 deployments go faster than the first one.
Put your named case studies on the homepage when you have them. Put the anonymous ones where a buyer who's already interested will find them, and write them for her.
The version that works
Take the piece you were going to publish and strip out every phrase that survives only because it's vague. "Leading provider." "Significant improvement." "Better workflows." Anything you could paste into a competitor's case study without editing.
Then check what's left. If there are still 5 concrete facts a reader in that market couldn't have guessed, you have a case study. If there aren't, the anonymity wasn't the problem, and no amount of naming the client would have saved it.
The clients who say no to their logo will almost always say yes to the details. Those are the parts that were doing the work.