The conversation goes the same way most times it comes up. An agency owner has a real result, they have permission to talk about it, and they still won’t use it, because they’ve already mentioned that client in two posts and a case study and they’re starting to feel the goodwill thinning. Nobody has told them to stop. They can just feel it coming.
That worry is reasonable, and it’s the constraint I run into far more often than a signed NDA. Permission isn’t a switch you flip once, it’s a relationship you can overdraw, and an agency that leans too hard on one client’s name is spending something it can’t get back.
So the practical question is what you publish when your best material is off the table.
The most convincing proof most agencies have is describing the buyer’s problem so precisely that they conclude you’ve solved it before, which costs nothing and needs nobody’s permission.
Quick Take
- The constraint is usually social rather than legal. You have permission, and you’re worried about wearing it out, which is a good instinct and a real limit.
- Describing the problem in detail reads as proof, because a stranger can’t fake the texture of a job they’ve never done.
- Anonymous case studies work as long as everything except the identity stays specific. Strip the specifics too and the whole thing reads as invented.
- If your numbers are real and modest, use the ones you can stand behind and stop making proof the lever you’re pulling.
What this covers
- Why you can’t use the proof you already have
- The proof that needs no permission at all
- How to write an anonymous case study that doesn’t read as made up
- Whether a constraint can work as proof
- What to do when your results are real but small
- How to get permission when you do want the name
Why can’t you use the proof you already have?
Three reasons, and only one of them is a contract.
The common one is over-mention. You named a client, it went well, so you named them again, and now you’re four mentions deep with the same logo and you can feel the account manager getting twitchy. Even a client who said yes enthusiastically has a threshold, and crossing it costs you more than the post was worth.
The second is attribution. You know the work moved the number and you can’t cleanly prove your piece of it, because they also hired a salesperson that quarter and their category got hot. Publishing a claim you’d lose an argument about is worse than publishing nothing.
The third is the actual NDA, which exists and is the rarest of the three in my experience. Most owners describing themselves as blocked by legal are blocked by the first two.
That matters because the fixes are different. A contract problem needs a lawyer or a workaround. A social problem needs a wider proof surface, so you stop drawing from the same well.
What counts as proof when you can’t name the client?
Being able to talk about the problem with real precision. It’s the strongest non-numeric proof I’ve seen agencies build, and buyers respond to it more reliably than they respond to numbers.
The reason it works is that specificity is expensive to fake. If you can describe what week three of the engagement feels like, which internal person usually blocks the project and why, the two things the buyer will be tempted to do that always backfire, and what the problem costs per month in a form they recognize, you’ve demonstrated experience without claiming any. They’re doing the inference themselves, and the conclusion they reach is that you’ve clearly been here before, because nobody knows this much about a problem they’ve only read about.
A buyer reading a page like that is checking a different thing than you’d expect. They’re not verifying your result. They’re deciding whether you understand their situation well enough to be worth an hour, and detail answers that question faster than a percentage does.
It’s also the only proof asset with no dependencies. You don’t need permission, you don’t need a client to reply to your email, and you don’t need the numbers to have cooperated. You need to have paid attention to the work you already did.
How do you write an anonymous case study that doesn’t read as invented?
Keep every specific except the identity, and give it a structure the reader can walk.
The version that works is problem, then outcome, then solution, in that order, which is the same order that works in a proposal. Open with the problem in enough detail that the right reader recognizes their own situation, state the outcome plainly, and then explain how you got there. That order matters because it lets someone see themselves in the story before they’re asked to care about your process, and a story a buyer can stand inside does work that a list of facts never will.
The failure mode is losing the specifics along with the name. When a case study reads as “a mid-sized client in the healthcare space saw significant improvement,” the anonymity has eaten the substance, and it lands as something you made up. Anonymize the company and keep everything else: the actual starting condition, the thing that was breaking, the sequence you ran, the timeline, the number if you have one.
The other failure is stacking facts without the structure. If it feels like you’re throwing evidence at someone hoping volume wins, it loses value even when every fact is true, because the reader can tell they’re being sold rather than shown.
Can a constraint work as proof?
Yes, if you explain why it’s good for the buyer rather than stating it as a policy.
At my last agency we capped how many clients we’d take at one time. On its own that’s a scheduling rule and nobody cares. What made it work was connecting it to the thing the buyer was already worried about, which was whether the senior people in the pitch would be anywhere near the actual work in month three. The cap answered that question structurally, in a way a promise couldn’t, because it’s checkable and it costs us something.
That’s the test for any constraint you want to use as proof. It has to cost you something visible, and the buyer has to understand why that cost benefits them. A limit nobody understands is trivia, and a limit tied to a fear they already have does the job a case study would have done.
Same logic applies to who you decline. Naming the work you won’t take is expensive to fake and reads as a person with standards rather than a vendor with availability.
What if your results are real but unimpressive?
Then use the ones you can stand behind, and if there aren’t enough of those, stop making proof your lever.
That second half is the part owners resist. Proof is one way to earn a buyer’s confidence and it isn’t the only one, so if your numbers are a 12% lift rather than a 300% one, you can lead with your point of view, your delivery model, your risk reversal, or how precisely you understand the problem. Forcing a thin result into a hero slot draws attention to exactly the thing you’d rather they didn’t examine.
When you do use a modest number, aim it at the reader who’s three steps behind your current clients rather than at your peers. A first inbound lead in week five sounds like nothing to another agency owner and lands hard on someone who hasn’t had one in a year. The small true win converts better than the inflated one, and it has the advantage of being defensible when somebody asks.
The most credible proof I know of is the story where something didn’t work and you adjusted. Almost nobody publishes those, which is exactly why they land.
How do you get permission when you do want the name?
Ask for a testimonial and write it for them.
Draft the thing you’d want them to say, send it over, and tell them to change anything that isn’t true or doesn’t sound like them. Most clients are happy to endorse you and unhappy to write, so the request that fails is the one that hands them a blank page and a deadline. Handing them something to approve turns a writing task into a thirty-second decision, and it’s the fastest way I’ve seen to get this over the line.
Two things make it land better. Ask when the result is fresh rather than at renewal, because the feeling fades and the calendar makes it awkward. And ask for something narrower than a general endorsement, since “would you confirm this specific outcome” is easier to say yes to than “would you say something nice about us.”
The short version
Before you go hunting for another client to name, write the problem page. One piece that describes the problem you solve in enough detail that someone living inside it recognizes their own week, including the parts that make you look like you’ve been burned before. That asset needs no permission, ages well, and does more work than a fourth mention of the same logo.
Then go back to the case study you can’t use and see whether it survives losing the name. If it holds up with every specific intact and only the identity removed, publish it. If it collapses into generalities, the name was carrying it, and the story needed more detail than you gave it.
If you’re sitting on results you can’t publish and a pipeline that needs something, that’s the kind of problem worth working through with other agency owners in the Dynamic Agency Community.
Frequently Asked Questions
Can I publish a case study without the client’s permission if I don’t name them?
Usually yes when the contract is silent and the details don’t identify them, though “doesn’t name them” and “unidentifiable” aren’t the same test. If the industry, the size, and the timeline together point at one company, you’ve named them without typing it. Check your MSA first, since some have a clause covering the work itself rather than just the name.
What’s the best proof for an agency with no results yet?
Your read on the problem, delivered in detail. A new agency that can describe the buyer’s situation better than the incumbents can is more convincing than one with two thin case studies, and it’s the one asset you can build before you have any clients at all.
How many times can I mention the same client?
There’s no rule, and the honest answer is that you’ll feel it before they say anything. If you’re hesitating before the next mention, you’re already at the edge. The fix is a wider set of proof rather than a better argument for one more post.
Do buyers care about numbers or stories more?
They care about recognizing themselves. A number without a situation is trivia, and a situation they’ve personally lived through pulls them all the way in, which is why the problem comes before the outcome in every version of this that works.
Should I use logos if I can’t use full case studies?
A logo wall is weak proof and it’s cheap to run, so it’s fine as a supporting element and it won’t carry a page. It also has the same over-mention risk, since clients notice their logo being used in ways they didn’t picture when they agreed.