All growth notes

Growth notes

Case-study pages that AI engines cite: what to include without inventing numbers

· LumosAI

Key takeaways

  • A case study needs a real problem, what you did, and what can honestly be said about the outcome.
  • Numbers only where you can back them; otherwise describe the outcome in words.
  • Anonymise the customer if needed, but keep the specifics of the work.
  • Structure it as question, answer, detail — the same shape engines quote.

A case study that AI engines cite and buyers believe states a real problem, the specific work done, and an outcome you can back up — with numbers only where you have them, and plain description where you do not. Invented percentages are the fastest way to lose both the engine’s trust and the buyer’s.

What does an honest case study contain?

Section Content
The buyer’s situation Industry, part, quantity, the constraint that made it hard
What was asked The question the buyer actually put to you
What you did Process choice, tooling, material, secondary operations, timeline
What it looked like Real photographs of the part and the process
Outcome What you can honestly say: delivered on time, tolerance held, repeat order — with numbers only if real
What you would do differently One sentence. It is disarming and credible.

What if the customer cannot be named?

Anonymise the name and keep everything else specific: “a European manufacturer of laboratory equipment needed 300 housings a year in flame-retardant ABS”. Specifics are what make it useful and quotable; the name is rarely what a reader needs.

How do you handle outcomes without numbers?

Describe them. “The first batch passed incoming inspection and the customer moved a second part to us” is a real outcome. “Reduced costs by 37%” is only real if you measured it. Our own site says exactly what we can prove: inquiries from a dozen or so a month to a stable 40–50 and 80+ in the latest month, most with drawings — and that these figures come from our own plant, not a count of customers.

Why do engines cite case studies?

Because they contain the specific, verifiable detail that answer engines look for: a real part, a real material, a real constraint, a real process. A case page structured as question, answer, detail reads like evidence, and engines treat it that way.

How many do you need?

One good one per major application. Update them as the work continues. A case study that ends with “and they are still a customer” beats ten that end with a number nobody checked.

Frequently asked questions

Can I use a customer’s logo?

Only with permission. Without it, anonymise and keep the technical detail — that is what buyers read.

Should case studies have a call to action?

Yes: what to send for a similar project — drawing, quantity, material — and a link to the relevant product page.

Is it acceptable to estimate an outcome?

No. Estimates presented as results are invented numbers. Describe the outcome in words instead.

How long should a case study be?

Five hundred to eight hundred words with photographs. Enough to be specific, short enough to be read.

See what AI sees when it looks at your website. Run the free website check — technical SEO, structured data and whether ChatGPT and Perplexity can cite you — or read how the LumosAI system runs this every day.

Explore GEO automation