The Customer Proof Problem: How to Turn Client Wins into Case Studies That Earn Press and AI Citations

Why most case studies never get published, and the ones that do get ignored

Ask any marketing lead how many case studies they have in the pipeline and how many have shipped. The gap is the problem. A case study usually dies in one of three places: the customer's legal or comms team never approves it, the draft is so scrubbed of specifics that nobody wants to put their name on it, or it ships and sits on a /customers page that no journalist, buyer or AI assistant ever quotes.

The shared cause is that the document is written as a vendor victory lap. 'Company X leveraged our platform to unlock transformative efficiency.' The customer has nothing to gain from approving that, a journalist has nothing to quote, and a buyer learns nothing. A language model asked 'how did a mid-sized logistics firm cut invoice errors?' has no sentence it can lift.

The position of this article is simple: stop writing case studies and start documenting one measurable change that the customer owns as much as you do. Then run it through a fixed four-week process. Everything below is built around that.

Build around one before-and-after, not a narrative arc

The unit of a useful case study is a single claim with a starting number, an ending number, a timeframe and a named owner. For example: 'Invoice error rate fell from 6.1% to 1.8% in five months, measured by the finance team on monthly reconciliations.' It is not 'improved accuracy' and it is not a basket of five loosely related benefits.

Pick the one metric that passes four tests:

• The customer already tracks it. If they have to produce a new measurement, approval slows down and the number becomes arguable.
• It has a clean baseline. You need a 'before' that predates your work, ideally from the customer's own reporting.
• It is not commercially sensitive. Percentages, time saved and error rates clear legal far more easily than revenue, headcount or cost figures. If the real number is sensitive, use a relative one (a 40% reduction) and agree that up front.
• A stranger would care. A journalist covering logistics, fintech or HR will quote 'cut time-to-hire from 52 days to 31'. They will not quote 'improved stakeholder alignment'.

If a project produced several results, choose one as the headline and relegate the rest to supporting detail. Case studies that lead with four metrics give the reader and the reporter nothing to hold onto, and they give the customer's approver four numbers to worry about instead of one.

Make the customer a co-owner, not a subject

Customers refuse to approve case studies for understandable reasons: they don't want to look like they endorse a vendor, they fear competitors learning their approach, or they simply have no incentive and no time. Co-ownership addresses all three.

Concretely, that means the customer contact gets something real out of the story:

• Their name and role lead, not your logo. The story is about the operations director who fixed a problem, with your product as a tool in that. People will approve being the protagonist of a story about their own competence far more readily than being a testimonial for a supplier.
• They write or rewrite their own quote. Never hand over a pre-drafted quote to sign. Ask two questions on a call, record it, and let them edit the transcript. Quotes that sound like a person get used; quotes that sound like a press release get cut by the customer's comms team anyway.
• They see the metric before anything else. Send the one-line claim first, with the baseline and method. Get agreement on the number before you write a word of narrative. If the number is wrong or unapprovable, you've lost ten minutes, not three weeks.
• They get something portable. A shareable version for their own LinkedIn, a speaking slot, or a byline opportunity in the trade press. The benefit to them is visibility for the person, not just the company.

Agree the ground rules in writing before the work ends, not after. A clause in the contract or a line in the closing email ('we'd like to document the result of X, subject to your approval of the final text') changes the later ask from a favour to a pre-agreed step.

The 4-week process

The reason case studies drag for months is that nobody owns the calendar. Treat it as a fixed process with weekly outputs.

Week 1: Lock the claim and get the green light

Choose the single metric and write it as one sentence with baseline, result, timeframe and measurement source. Send it to your customer contact with a short note: what the story is, who it might be shown to, and what you need from them. Ask for a 30-minute call. The only deliverable this week is a yes on the number, or a corrected number. If you can't get that, don't proceed; move to a different customer rather than force it.

Week 2: Interview and draft

Run the call. Ask what the situation was before, what they tried first, what changed, and what surprised them. The 'what they tried first' question matters because it makes the story credible and is exactly what trade journalists want: the failed attempt is more interesting than the polished win. Draft the page the same week, then send it with the claim highlighted and any sensitive passages flagged, so the approver is reviewing a short list of decisions rather than reading cold.

Week 3: Approval and the press-ready package

Chase approval politely but on a schedule: a reminder at day three, a call at day six. While you wait, build the press package: a 150-word summary written from the customer's perspective, the headline number, two usable quotes, and a clean photo of the customer contact if they agree. Tell the customer exactly where you plan to pitch it, because 'a trade title covering your sector' is an easier approval than an open-ended 'the media'.

Week 4: Publish, then pitch

Publish the page first so there is a canonical, linkable source. Then pitch. Send three to five tailored emails to trade reporters and newsletter writers who cover the customer's industry, not yours. Lead with the customer's problem and number, in two or three sentences, and offer an interview with the customer contact. Offer the customer as the source, not your CEO. Follow up once after a week. Do not expect a national outlet; a respected trade publication, a sector newsletter or a podcast is a realistic and valuable target.

Why you pitch it as the customer's story

A reporter at a trade title has no interest in your product launch dressed up as a case study. They do have a standing need for concrete examples of how real companies in their sector solved real problems, ideally with a number. When you pitch 'how a regional freight company cut invoice errors by 70%', the vendor becomes a supporting detail, which is exactly where it should be for the story to run.

It also makes the pitch cheaper to execute. You are not asking a journalist to cover you; you are offering them a source and a data point. If it runs, the customer has been publicly credited, which strengthens the relationship and makes the next case study easier to get approved. If it doesn't, you still have a published page and a quotable claim.

Structure the page so a buyer, a journalist and an AI can all lift a claim

The page itself determines whether any of this compounds. Three audiences read it differently, and one structure serves all three.

• Open with the claim in plain words. The first paragraph should state who the customer is, what the problem was, and the before-and-after number. A buyer skims it, a journalist copies it, and an AI assistant can extract it as a standalone sentence.
• Name things precisely. Company type, size band, sector, location where permitted, the role of the person quoted, the timeframe and the measurement method. Vague pages ('a leading enterprise') are skipped by every reader; specific ones get quoted.
• Use descriptive headings that are questions or statements. 'What the finance team changed in the first 60 days' is liftable. 'Our approach' is not.
• Put the numbers in a short, scannable block. A three-line summary of baseline, result and timeframe, in plain text rather than an image, so it can be read and cited by systems that can't see graphics.
• Attribute every claim. 'According to Maria Alvarez, operations director' with the measurement source named. AI assistants tend to favour sources that attribute figures to a named person and method, and journalists need that to verify.
• Add a dated, accurate headline and meta description. Include the result in the title ('How [Customer] cut invoice errors from 6.1% to 1.8% in five months') so that search and answer engines can match it to a question.

Keep one case study per URL rather than collecting them in a gallery, and link each from a simple index page. A standalone page with a specific claim is what gets cited; a carousel of logos is not.

What this costs, and what to skip

The cost is mostly time, not money. Expect roughly six to ten hours of your team's time across the four weeks (the call, drafting, chasing, pitching) plus perhaps a couple of hours from the customer. There is no required agency spend and no paid placement. If you add a photographer or a designer, that is optional and modest.

What to skip: video testimonials as the primary asset (expensive, slow to approve, invisible to text-based search and AI), long PDF case studies behind a lead-capture form (journalists and AI can't easily reach or cite them), and any claim you can't tie to the customer's own data. Also skip the temptation to produce twelve mediocre studies a year. Two or three well-evidenced ones, each with a clean number and a co-owning customer, will outperform a library of vague ones in press interest, buyer trust and citations.

What to do this week

• List your five best recent customer outcomes and mark which have a number the customer already tracks.
• Pick the one with the cleanest baseline and the most friendly contact.
• Write the one-sentence claim and send it to the customer for a yes or a correction.
• Book the 30-minute interview and block the four-week calendar now, with a named owner for each week.

Customer proof only works when it is specific, owned by the customer and easy to quote. Build for that, and the same page earns trust from buyers, a hearing from trade press, and a place in the answers AI assistants give.