Why AI systems cite some SMB-to-enterprise stories and ignore others

Customer stories from companies that scaled from SMB to enterprise are some of the most commercially useful content a B2B marketer can produce. They demonstrate product-market fit at multiple price points, de-risk the buying decision for mid-market prospects, and give sales something concrete to use in committee conversations. So why do AI systems almost never cite them?

Not because they're bad content. Because of how they're built.

The structural problem with how most case studies are written

Most SMB-to-enterprise stories follow the same template: customer had a problem, found the product, grew with it, numbers improved. The company is named, a quote is attributed, a metric is cited. It reads well. It converts reasonably well. And it is almost completely invisible to AI systems reasoning about a category.

The reason is that AI systems don't retrieve these stories the way a search engine does. They absorb claims that appear consistently across multiple independent sources and associate them with a company. A single case study on a vendor's own website, however well-written, is a first-party assertion. There is no independent corroboration. The model has no basis to treat it as a signal about what the company actually does, rather than what the company says it does.

This is the structural gap. The content exists. The citation mechanism doesn't.

When I put together the 10-point plan for scaling from SMBs to the enterprise, the thing I kept coming back to was that the jump from SMB to enterprise isn't just a product problem or a sales problem. It's a credibility problem. Buyers in enterprise procurement committees need to believe the product was genuinely used and trusted by companies they recognise, in situations they can map to their own. AI systems have the same requirement, and they're looking for the same signals: third-party corroboration, analyst coverage, press coverage, community discussion. The vendor's own case study page satisfies none of those.

What gets cited, and why

The SMB-to-enterprise stories that do get cited by AI systems share a pattern. They appear in multiple places, not just on the vendor's site. They get discussed in analyst reports, referenced in industry newsletters, picked up in comparison posts on third-party review platforms. The metric cited in the original story gets repeated by a journalist or a community member. The company name and the vendor name appear together in contexts the vendor didn't control.

That's not a content strategy. It's a distribution and corroboration strategy. The content is the raw material, but the citation happens downstream.

This is the same category error I've written about before: treating AI visibility as an SEO problem misses that there's no page two when a model synthesises the answer instead of linking to you. If a claim about your company appears in one place, attributed to you, the model's prior on it stays weak. If the same claim appears across ten independent sources, the model treats it as established fact about your category. Nowhere is that more visible than in how customer stories either propagate through third-party coverage or stay trapped on the vendor's own domain.

There's a useful framing from Brin and Page's original PageRank paper that maps onto this: citation weight flows from the independence of the sources, not the quality of the content. The mechanism is different, but the same principle holds for what a language model treats as established.

What this means practically

If you're producing SMB-to-enterprise case studies and they're not appearing when buyers ask AI systems for examples of companies that scaled with your product, the fix is not to rewrite the stories. It's to create conditions for the stories to be discussed outside your own properties.

That means: pitching the story to analysts who cover your category. Getting the customer to publish their own version, in their own words, on their own blog. Submitting it to industry newsletters. Making it easy for community members to reference the specific metric in their own comparisons.

The case study is the beginning of the process, not the end. What AI systems cite is what the broader ecosystem decided was worth repeating. Your job is to make that repetition happen.