The Content Marketing Institute surveyed 980 B2B marketers between June 25 and August 16, 2024, and published the result on October 9, 2024: 53% ranked case studies and customer stories among the content types that produced the best results for their teams. Only video scored higher, at 58%. A year earlier, case studies had held the top spot outright.
So marketers believe in them. The harder question is whether buyers do, and whether the AI assistants that now sit between your buyer and your website ever read them. The case study statistics below come from marketer surveys, buyer surveys and AI citation studies, each with its source and date. Read together, they say something uncomfortable. The case study as most companies produce it (a PDF behind a form, written in the vendor’s voice, published once and never repeated anywhere else) is losing value at exactly the moment proof matters most. At the end you get a scoring model for fixing that.
Marketers still rank case studies near the top
Start with the CMI data, because it is the largest recurring benchmark in the field. In the B2B Content Marketing Benchmarks, Budgets, and Trends report for 2025, 75% of B2B marketers said they had used case studies or customer stories in the previous 12 months. That put them third in usage, behind short articles and posts at 92% and video at 76%. On effectiveness, case studies came second, with 53% naming them among their best performers.
The pattern is not new. B2B Marketing’s benchmarking report of July 3, 2015, based on 112 B2B marketers, found 66% rated case studies “very effective.” Our roundup of B2B content marketing statistics covers the rest of the CMI findings.
What marketer data cannot tell you is why case studies work, or for whom. For that, you need buyer surveys.
What do buyers actually do with a case study?
Demand Gen Report’s 2022 Content Preferences Survey, which polled 174 B2B executives, gives the clearest picture of buyer behavior with this format. Forty-nine percent said they had consumed case studies in the past year. That placed case studies below webinars at 67%, e-books at 56%, research reports at 55%, blog posts at 54% and white papers at 52%.
Consumption is not the same as value, though. When the same buyers ranked the content most valuable for researching a purchase, case studies tied with webinars at 40%, behind only research and survey reports at 43%. Forty-four percent said case studies were most valuable in the middle of the buying process, the stage where a buyer has a shortlist and needs a reason to move one name up it. And 40% said they shared case studies with colleagues.
That last figure matters. A case study is one of the few assets a champion forwards to the people who sign the contract, a sales tool that travels without your rep in the room. That explains why marketers rate it so high even when buyers consume it less than webinars.
The trust gap: buyers discount your own stories
Here the numbers turn. UserEvidence’s 2025 Evidence Gap report, based on 811 B2B software buyers, sellers and marketers, found that buyers rank case studies and testimonials as only the fourth most trustworthy type of evidence. Analyst reports topped the list, with third-party reviews and peer recommendations close behind. The detail inside the ranking is useful: 64% of buyers trust named customer testimonials, and 60% trust blind but verified ones. A real name attached to a real result earns trust. An anonymous “Fortune 500 retailer” earns less.
Buyers do not distrust vendors across the board. The same report found 68% of buyers trust vendor-provided data completely or most of the time. They trust facts. They discount framing.
Gartner measured the same instinct from a different angle. Its survey of 771 B2B buyers, fielded in November and December 2022 and published June 8, 2023, found that buyers value third-party interactions 1.4 times more than digital interactions with the supplier. Gartner counted customer references and reviews among those third-party interactions. A case study on your website is a supplier interaction. The same customer saying the same thing on a review site, in a trade publication or on a reference call is a third-party one.
TrustRadius adds the review side. Its 2026 B2B Buying Disconnect report, released July 15, 2026 and based on 1,862 technology buyers surveyed in January 2026, found 74% of buyers use reviews to inform purchase decisions. The same survey found only 13% used analyst reports to make a decision, a 63% drop since 2022. Analyst reports may top the trust rankings in the UserEvidence data, yet few buyers actually use them. Reviews are the proof buyers reach for in practice, and a review is a short case study the customer wrote without your editing.
Missing proof is costing deals
The UserEvidence data also puts a number on what weak proof costs. Twenty-six percent of deals fail for reasons tied to customer evidence: unproven ROI, no live reference available, or no clear difference from a competitor. Younger buyers feel the gap more. Among buyers aged 20 to 30, 21% said a deal fell through because they could not get a live customer reference, against 8% of buyers overall.
The supply side explains the shortage. In the same report, 78% of marketers produced five or fewer customer stories in the previous six months, and 36% produced two or fewer. Forty-seven percent said access to customers was their biggest obstacle. Yet 65% said their customers can quantify the impact of the product. The results exist. They are stuck in customer success notes and QBR decks because nobody has the time or the access to turn them into published proof.
A sales team losing a quarter of its deals to proof gaps is getting roughly one new customer story every five weeks. Most teams know which accounts love them. Few ask those accounts to say so in public, which is the subject of our guide on how to write case studies customers will sign off on.
Buyers pick a favorite before they call you
Timing makes all of this worse. 6sense’s 2025 Buyer Experience Report, released in November 2025 and based on more than 4,000 B2B buyers, found that 94% of buying groups had ranked their preferred vendors before first contact with any seller. They bought from that early favorite 77% of the time. On average, buyers made first contact 61% of the way through the process, earlier than the 69% 6sense measured in prior studies, but still well past the point where the shortlist forms. The average buying cycle ran 10.1 months, down from 11.3 in 2024.
Gartner’s older work shows how little of that time vendors get. In a survey of 750 B2B buyers published August 15, 2018, Gartner found buyers spent about 17% of their buying time meeting with potential suppliers. With three suppliers in the running, each one gets about 5%. Buyers spent much of the rest researching on their own and talking to peers.
Gartner’s newest buyer data, a survey of 645 B2B buyers fielded in August and September 2025 and released May 20, 2026, found buyers used an average of seven information sources in a recent purchase. Forty-five percent used generative AI, mostly to research vendors and products, and 69% said they prefer to check AI-generated insights with a sales rep. Seventy percent want a fully digital, self-service buying experience.
Put those numbers together and the job of a case study changes. It used to be something a rep sent after the discovery call. Now it has to be findable, readable and believable during the 83% of the journey your rep never sees, which increasingly means inside an AI answer.
Do AI assistants cite case studies?
The research here is thinnest, so treat each figure as a direction. No study we found isolates case study pages as their own citation category. Several come close.
Omniscient Digital published a study on January 20, 2026 that ran 240 branded prompts through ChatGPT, Perplexity, Gemini, Google AI Mode and AI Overviews and classified 23,387 cited sources. For prompts that named a brand, 57% of citations went to a bucket the researchers called reviews and social proof: customer reviews, listicles, forums, social media and case studies. Directory and reference sites took 17%, and product pages 12%. When a buyer asks an AI assistant whether your product is any good, the assistant mostly answers with what other people said about it.
Whose pages carry that proof? A Growfusely analysis published July 28, 2026 ran 128 queries across 16 B2B software categories through Perplexity and OpenAI’s web search API and logged 1,739 citations. Vendor self-citation sat between 12% and 24% depending on the category, about one in six overall. The engines differed: vendor pages made up 24% of citations in the ChatGPT-family results against 12% in Perplexity. Review platforms such as G2, Capterra and Gartner took 7.3% overall, rising to 13% to 16% in categories like applicant tracking and HR software.
AirOps research from October 2025 points the same way from the brand side. First-party pages made up 13% to 21% of brand mentions in Claude and Perplexity answers, and 4% to 11% in GPT-5. Product pages accounted for 19.3% of those first-party mentions and homepages 7.1%.
And when TripleDart looked at 65,583 AI-referred sessions across 28 B2B SaaS companies in the first quarter of 2026, case study pages did not appear among the top 15 of 35 page types. Homepages took 31.9% of sessions and general blog posts 18.5%.
The broader citation picture backs this up. Muck Rack’s Generative Pulse report, released May 7, 2026 after analyzing more than 25 million links cited by ChatGPT, Claude and Gemini, found earned media made up 84% of citations. That figure has critics. In PR Daily on August 17, 2026, David Clare of Fire on the Hill argued the 84% bundles journalism with Wikipedia, academic papers and third-party corporate content, and that journalism alone sits nearer a quarter. Muck Rack’s own number for journalism is 27%. Either way, the conclusion for case studies holds: the version of your customer story that lives only on your domain is a minority input. The version that appears in a review, a trade article or a partner’s blog is the one the models pick up.
Buyers are using those answers. G2’s 2025 Buyer Behavior Report, based on 1,169 B2B decision makers surveyed in April 2025, found 29% now start research on tools like ChatGPT more often than on Google. Forrester’s 2026 buyer survey found 29% of buyers named AI search or analysis tools among their most meaningful interactions in a purchase. And TrustRadius found 63% of buyers used AI during their purchase journey, with 94% of those buyers fact-checking the answers at least some of the time. A buyer who fact-checks an AI claim about you lands on a source. You want that source to be a named customer saying something specific. Our piece on optimizing case studies for AI covers the on-page side; the case study statistics above say the off-page side matters more.
The Proof Portability Score
These findings point to one fix. A case study’s value now depends less on how well it is written and more on how far it travels: off your site, into reviews, into press, into the answers. The Proof Portability Score is a 10-point check for exactly that. Score each of five criteria zero, one or two, and run it on your three most-used case studies before you write a new one.
Named. Does the story carry the customer’s company name and a named person with a title? Zero if anonymous, one if the company is named but no person, two if both. UserEvidence’s finding that named testimonials earn 64% trust against 60% for blind ones is the floor; a named human is also the only version a journalist or an AI system can attribute.
Numbered. Does it state one result as a number with a baseline and a time frame, such as “cut onboarding from 21 days to 9 in one quarter”? Zero for adjectives only, one for a number without a baseline or time frame, two for all three. Buyers trust vendor data 68% of the time when it is data. A claim with no baseline is framing.
Front-loaded. Can a reader, or a model pulling a passage, get the customer, the problem and the result from the first 100 words of an open HTML page? Zero if the story sits in a gated PDF, one if it is open but the result is buried, two if the first paragraph works as a standalone quote. Most AI-referred sessions land on homepages and blogs, not case study libraries, so the story has to survive being excerpted.
Comparable. Does it say what the customer used before, or what else they evaluated? Zero if not, one for a vague reference, two for a named prior tool or approach. UserEvidence found unclear differentiation is one of the three evidence gaps behind failed deals, and comparison language is what lets an AI answer place you against a competitor.
Echoed. Does the same customer result appear on at least one domain you do not own: a review on G2 or TrustRadius, a trade press article, a podcast episode, a partner’s blog, a joint conference talk? Zero for none, one for one outside source, two for two or more. This criterion carries the most weight in practice. Gartner’s 1.4 times preference for third-party interactions, the 74% of buyers who use reviews and the citation data all say the same thing: the off-site echo is the version buyers believe and the models read.
Read the total in bands. Eight to 10 means the story is doing its job in the buyer’s independent research. Five to seven means the story is solid but stuck on your site; spend your next month on echo, not on a new asset. Below five means the case study is a sales attachment, useful after the first call and invisible before it, which the 6sense data says is too late for 94% of buying groups.
Four of the five criteria are writing fixes you can make in an afternoon. Echo takes outreach, which is why most teams skip it and why it separates the stories that get cited from the ones that get filed.
What changes in 2027
The case study statistics in this piece describe a format that marketers trust, buyers use in the middle of a purchase and AI systems mostly encounter secondhand. The production gap (five or fewer stories in six months for most teams) will not close by writing faster. It closes by getting more mileage out of each customer who says yes: one interview turned into a named page, a review request, a trade press pitch and a podcast booking. If you want to see where your customer proof already shows up in AI answers, Instant Press runs a free AEO rating that checks how assistants describe your brand for real buyer questions.
The next customer story that matters in your market will be the one a buyer reads in an AI answer, attributed to a name, before anyone from your team gets a meeting.