Case Studies That AI Engines Cite

#01Why do case studies matter for AI search?

Our view: when buyers ask engines "does [product] work for companies like mine?", engines look for evidence. A case study that clearly states the customer's sector, size and outcome gives the engine something concrete to cite. Google's AI optimization guide favours unique, first-hand content over summaries of what's available elsewhere (Google).

#02What structure works best?

#03How do you measure results honestly?

State the metric, the baseline, the period and who measured it. "Invoice processing time fell from 6 days to 2 days between January and June 2026, measured in the customer's ERP" is credible. "Saved 70% time" without context isn't. Our view: if you can't measure, describe qualitative change honestly instead of inventing numbers. The GEO research found that adding statistics improved visibility in its test setup, but only real statistics help readers (Aggarwal et al.).

#04Why include limits?

Our view: a case study that mentions a slower-than-expected rollout or a feature the customer didn't adopt reads as honest, and helps similar buyers set expectations. It also distinguishes your page from marketing copy that engines might weigh less.

#05What about permission and disclosure?

Get written approval from the customer for their name, quote, logo and figures. If the customer received discounts or other benefits for participating, disclose it. In the EU, unfair commercial practices rules prohibit misleading claims to consumers (EUR-Lex). Check rules in your markets.

#06How should case studies be published?

• As HTML pages, not only PDFs or videos
• With a descriptive title naming the customer and outcome
• Dated, with customer size at the time
• Linked from relevant use-case, industry and integration pages
• With a short text summary if there's a video

#07What's a worked example?

Fictional example: "How Müller Bau cut subcontractor invoice approvals from 6 days to 2".

• Customer: 120-person construction firm in Bavaria.
• Problem: paper invoices lost between site and office.
• Workflow: mobile approvals by site managers, synced to accounting.
• Results: measured over six months in the customer's accounting system.
• Limits: two sites still on paper due to connectivity.
• Quote: named finance director.

#08How do you measure case study impact?

• Prompt tests: "[product] customer examples in [sector]"
• Whether engines cite the case study or repeat its numbers accurately
• Sales use of the case study in deals

#09How many case studies do you need?

Our view: enough to cover your main segments and use cases, with at least one per important sector. Quality beats quantity; five detailed, measured case studies are better than thirty logos with a sentence each.

#10How do you keep them current?

Add a "results as of" date, update if the customer's situation changes significantly, and archive case studies for customers who've left, or note the date clearly. A case study for a churned customer presented as current is misleading.

#11How do you get customers to participate?

Ask at a moment of success, such as after a measured improvement or renewal. Offer to draft the text from an interview and give the customer full approval rights. Our view: make it low effort: a 30-minute interview, one review round and a clear publication date. Offer reciprocal value, such as a link to their site or a joint webinar, and disclose it if relevant.

#12What should you avoid?

• Results without baselines or dates
• Generic quotes written by your marketing team
• Hiding the customer's size, which matters to similar buyers
• PDFs as the only format

#13How should video case studies be handled?

Video interviews are persuasive, but text is easier to quote. Publish a written summary with the key figures and quotes alongside the video, and follow Google's video guidance so the video can be found: embed it on a dedicated, indexable page with a title and description (Google). Our view: include a transcript so every claim in the video is available as text.

#14How do case studies relate to reviews?

Reviews are many short opinions; case studies are one deep, verified story. Our view: engines may draw on both, so encourage case study customers to leave a review on independent platforms too, without incentives that platform rules prohibit. The two together give a more complete and credible picture than either alone.

#15How do you write the summary sentence?

Name the customer, the problem and the measured result in one line, such as "Müller Bau cut invoice approvals from 6 days to 2 in six months". Our view: this is the line most likely to be quoted, so make it accurate and self-contained.

#16Checklist

• Named customer with approval
• Problem, workflow, results, limits
• Metrics with baseline, period, method
• HTML page, dated
• Linked from use-case and industry pages

#17From MentionX data: how often brands appear by prompt typeOriginal data

Method. Read-only aggregate queries over MentionX scans from 22 August to 29 October 2026 (UTC): 2,217 engine runs (2,154 completed) across 18 anonymised projects and six engines.

Case studies are most useful when buyers ask which product suits a situation. In the MentionX sample, brands are mentioned far more often in comparison prompts than in problem prompts:

Cited URLs come from a mix of page types, with articles the largest:

Our view: a case study published as a clear article-style page, with the customer type and outcome in its first lines, competes in the most cited category and gives engines evidence for comparison and recommendation prompts.

Limits. A convenience sample of mostly B2B software projects, run through engine APIs, which can behave differently from the consumer apps. Perplexity, Copilot and DeepSeek figures come from a single day (26 October 2026), so treat them as a snapshot, not a trend.

#18Worked example: Ledgerline rewrites a customer story

Ledgerline is a fictional accounts-payable tool. Figures are illustrative.

Ledgerline's best customer story is a PDF with a headline about "transforming finance". Prompt checks show it's never cited.

The team rewrites it as a web page. The first sentence says that a fictional 40-person logistics firm cut its invoice approval time after moving to Ledgerline, and how the time was measured: the average days between invoice receipt and approval, before and after, taken from the customer's own reports over three months. The page covers the customer's setup, the problem, what they configured, the results and the limits, including that the firm also hired an extra clerk during the period.

The customer approves the text and the quote in writing. The page links to the relevant product and integration pages, and the old PDF redirects to it.

#19How do you choose which customers to feature?

Pick customers that match the buyers you want to reach, by industry, size or use case. Cover the main situations buyers ask about, such as switching from a competitor or adopting a specific integration. Our view: a small set of varied, specific stories is more useful than many similar ones about your largest customers.

#20How do you interview a customer for a case study?

Ask what they used before, what prompted the change, how they chose, what they set up and what changed afterwards. Ask how they know it changed, and for any figures they're willing to share. Ask what didn't go well. Record the interview with permission so quotes are accurate, and send the draft back for approval.

#21How do you title a case study page?

Use a title that says who the customer is and what changed, such as "How a logistics firm sped up invoice approvals". Avoid titles that only name the customer, which engines and readers can't match to a question. Keep the customer name in the first paragraph if they've agreed to be named.

#22Where should case studies link?

Link from each case study to the product, feature and integration pages it mentions, and from those pages back to relevant case studies. Group case studies by industry and use case on an index page. These links help readers find proof at the point they need it and help engines connect evidence to the right pages.

#23How do you write results that are easy to quote?

State each result in one sentence with the measure, the change and the period, such as "Approval time fell from five days to two over three months." Say how it was measured. Avoid vague claims like "massive savings". Our view: a sentence engines can quote accurately is worth more than an impressive but vague headline.

#24How do you handle results that can't be shared?

Some customers can't share figures. Describe the change in words, such as which steps were removed, and use quotes from the people involved. Say clearly that figures weren't shared, rather than implying results you can't support. Where a customer agrees, update the case study a year later with what happened next, including any setbacks. Keep the customer's approval email on file with the published version.

#25Common questions

Should case studies be gated? Our view: no. A gated PDF can't be read by most crawlers, so engines can't cite it.

Can we publish anonymous case studies? Yes, if the customer prefers. Describe the company type and size, and keep the results and method specific.

How long should a case study be? Our view: long enough to cover the problem, solution, results and limits, often 600 to 1,200 words.

#26Related

• Use-case pages that get recommended
• Original research as a GEO moat