For a company about to go public, the room moved. It is a chatbox.
A new study found that more than half of newly public US companies cannot control what AI engines say about them — and the cost of that gap shows up exactly when the company can least defend itself.
The 5W IPO AI Visibility Index measured 25 of the most-watched companies entering the US public markets between Q4 2024 and Q2 2026 across the five major AI engines now driving consumer and analyst research.
The Study
Each of the 25 companies was tested against ChatGPT, Claude, Perplexity, Gemini, and Google AI Overviews across a 50-prompt set spanning five query categories: corporate identity, product and category positioning, financial narrative, leadership, and regulatory standing. Each company-engine-prompt combination produced an observed citation outcome. Prompts were issued during May 2026.
Each company received a composite Citation Visibility score across five dimensions:
Citation Frequency (40%) — mentions across prompts and engines, normalized
Cross-Engine Breadth (20%) — number of distinct engines citing the company
Query-Type Breadth (20%) — number of distinct prompt categories the company surfaces in
Extractability (15%) — quality of cited content, including schema markup, factual specificity, and structural clarity
Crawl Access (5%) — robots.txt and llms.txt audit results
Each company is then sorted into one of three tiers: companies that control their AI answer, companies that are famous-but-defenseless, and companies that are largely invisible. The composite scoring methodology is the same Citation Index framework 5W has applied across beauty, gambling, and lottery sectors.
What the Index found
Tier 1 — Control Their Answer
Eight of the 25 companies. The engines explain the business confidently and largely in the company’s own framing. Heavily skewed toward enterprise infrastructure and developer tools — categories with strong technical documentation, dense third-party validator coverage (Gartner, Forrester, G2), and consistent founder narrative across owned channels.
Tier 2 — Famous and Defenseless
Twelve of the 25. The most expensive position in the Index. The AI engines recognize the company instantly but explain it through press narrative the company cannot shape. Skewed toward consumer marketplaces, fintech, and recently restructured technology companies. High recognition. Low control.
Tier 3 — Invisible
Five of the 25. The AI engines describe the company through generic category language rather than company-specific positioning. Skewed toward B2B vertical software and specialty fintech.
Two companies. Same news cycle. Two completely different AI explanations.
One marketplace company in the cohort surfaced consistently through layoff coverage when buyers asked AI engines what the company does. The product description came in second. The company’s own framing of its business model came in fourth.
One infrastructure company in the same cohort surfaced through product architecture documentation. Layoffs that hit the same company during the same period did not appear in the AI summary at all.
Same news cycle. Two completely different AI explanations.
The variable was not company size, sector, or news exposure. The variable was the volume and structure of citation-eligible content each company had published about itself before the news cycle hit. The infrastructure company had architecture documentation, founder interviews on developer platforms, third-party analyst coverage, and technical case studies in the citation graph before the layoff news landed. The marketplace company had product launch coverage and a press release archive.
Two different inputs produced two different outputs. The engines repeated whoever had written the most.
Recognized is not controlled
Being recognized by the engines and being in control of what they say are two completely different things.
The companies that own their answer did not get there with a clever campaign. They got there because they published more about themselves than the market did. Out-wrote everyone. The machine repeats whoever wrote the most authoritative thing.
The middle of the table is where the crisis lives
The dangerous position is not invisibility. It is famous-but-defenseless.
Household names. Recognized instantly. Explained largely through the decline story the press wrote after the listing. Accurate enough. Completely uncontrolled.
That is the most expensive position there is: famous inside the engines, and defenseless about what they say.
The Quiet Period Problem
The quiet period limits what a company can say. It does not limit what the engines repeat.
This is the most critical IPO communications finding in the study. When the lock-up clock is running and corporate communications is legally constrained, AI engines continue serving answers to every analyst, reporter, recruit, and prospective investor who types the company name into a chatbox. The engines cite whatever sources they have. If the dominant source available is press narrative from the launch cycle, that becomes the answer — for months — regardless of whether the company’s actual story has moved on.
A company that has not built citation infrastructure before the quiet period locks in has no operational way to correct it during the quiet period. The infrastructure either exists or it does not. The window to build it is before the S-1 filing. Not during the roadshow. Not after the bell.
Five of the 25 companies in the study spent the post-IPO quiet period with AI engines confidently describing them in terms the company would not have chosen. Three of those five had not produced a single structured corporate identity page on their own site within the 90 days before listing.
The quiet period turns a content gap into a reputation problem. The earlier the gap is closed, the smaller the eventual reputation problem becomes.
What every IPO communications team should do now
Build the infrastructure before the crisis, not during it. That principle has not changed in the AI era. It has sharpened.
Three moves separate the IPO companies that own their answer from the ones that inherit one:
Run a baseline Citation Share audit across all five major AI engines before the S-1 filing date. Know where the company stands before designing the response. Most companies entering an IPO window do not know.
Audit AI accessibility infrastructure — robots.txt configuration, llms.txt deployment, schema markup coverage on the highest-intent corporate pages. Most newly-public companies have weak AI accessibility because the discipline did not exist when the investor relations site was built.
Build a structured content cadence around the corporate narrative — analyst-style explainers, founder profiles, product technical documentation, third-party validator citations. The companies that publish more about themselves than the market does are the ones the engines learn to cite confidently.
The closing principle
AI Communications is a mix of journalism, psychology, and engineering — and the audience is now the machine.
IPO communications used to mean controlling the room. Now it means controlling the answer before the room asks the machine.
Ronn Torossian is the founder and chairman of 5W AI Communications, the AI Communications Firm. He is the publisher of Everything-PR and the author of two best-selling editions of For Immediate Release.