Cited by AI Does Not Mean Believed: Burson x Profound's Credibility Paradox Pushes GEO From Visibility to Persuasion

On June 2, 2026, the global PR group Burson and AI marketing platform Profound released a report, The Credibility Paradox, with a number that should make everyone doing GEO (Generative Engine Optimization) stop: across 7 major AI answer platforms, for 85 companies, they fielded thousands of brand-related answers and used a tool called Decipher to produce over 55,000 "believability forecasts". The one-line conclusion: your brand appearing in ChatGPT or Google AI Overviews is a completely separate thing from whether what the AI says about you is believed.

To see why this is a turning point, look at where GEO has gone over 12–18 months. Since 2025 the whole industry has bet on "visibility": getting cited by AI, pumping citation counts, chasing share of voice. A pile of tools (Profound, Evertune, Scrunch, CiteLens) all measure "how often you're mentioned." But nobody asked the next question — when you're mentioned, do readers believe it? Burson's report lifts the conversation from "a technical visibility exercise" to "a strategic reputation problem," effectively declaring GEO's phase one (winning visibility) has peaked and phase two (winning credibility) has begun.

Peers are diverging. Pure monitoring tools (Evertune, Scrunch) measure exposure and sentiment; newcomer CiteLens just announced GA in June, tracking citations across ChatGPT/Perplexity/Gemini/AI Overviews; and PR groups like Burson cut at "credibility" and reputation capital. The whole category is moving up from "monitoring" to "strategy."

What does this mean for SMBs? The good news: you don't have to out-mention big brands; you can compete on "when you are mentioned, is that line credible." Below we unpack the report's key numbers and give SMBs a path that doesn't require enterprise subscriptions.

Event detail + full numbers

The report scored against the 8 levers of Burson's Reputation Capital framework (Innovation, Creativity, Workplace, Products, Financial Performance, Governance, Citizenship, Leadership). Three key findings: First, business decision-makers rated AI-generated answers about 10% more believable than the general population — meaning the same AI answer lands more easily with your B2B buyers than with general consumers. Second, fact-based claims (product specs, innovation, workplace culture) are more believable than subjective claims (leadership, governance); the report even names "leadership" as AI's "clearest liability" for brand reputation. Third, believability varies by audience (forecast separately for general population, opinion elites, business decision-makers). In short, GEO is no longer the binary "am I cited," but a three-dimensional "for whom, on what claim, at what believability."

Immediate actions for three readers

• Brand owners / SMB owners: first compile the "facts" you most want AI to get right onto one page — founding year, services, concrete numbers (case count, years in business, certifications). Fact-based claims are inherently more believable; writing them clearly in a structured place on your site beats a pile of adjectives.
• Marketing / SEO operators: stop chasing "mention count." Manually ask ChatGPT, Perplexity, and Gemini 5 buyer questions each with your brand name, copy down the answers, and tag each sentence: "fact or subjective, believable or not, anything wrong." This manual audit sheet catches problems earlier than any dashboard.
• Developers / agencies: add Organization, Product, and FAQPage JSON-LD to client sites to feed machine-readable facts to AI; ensure about/sameAs point to authoritative sources (official socials, government registries).

SaaS tool comparison table

What they won't tell you

First, "believability" is predicted by Burson's own Decipher tool, not actual reader thumbs-up — it's a model with its own assumptions; 55,000 figures are "forecasts," not measured clicks. Second, this report is published by a PR group, which naturally steers the conclusion toward "you need reputation management (i.e., our service)." For SMBs, rather than fretting over a credibility score, first make sure AI hasn't gotten your basic facts wrong — misinformation hurts far more than "not persuasive enough."

The "no SaaS subscription" SMB alternative

You can run a credibility audit without enterprise tools: use a Google Sheet with columns "Platform / Question / AI answer / Fact or subjective / Correct? / Action." Each month, ask 3 platforms × 5 buyer questions and paste the answers in, tagging each line. Then reinforce site facts with free Schema.org structured data. The whole cost is 30 minutes of your time per month, yet it catches 80% of "AI got you wrong" problems.

FAQ

Visibility or credibility — which should an SMB do first?

Do "accuracy" first, then credibility, and last the volume of visibility. If AI doesn't mention you at all, start with basic structured data and site facts; once you start being mentioned, audit whether it's right and believable. For small firms, "being said wrong" hurts more than "not mentioned often enough."

I can't afford tools like Profound — can I do this myself?

Yes. The core action is "regularly ask AI manually and audit the answers line by line." A spreadsheet captures 80% of the value. Tools' advantage is scale and historical trends, but for a single-brand SMB, manual audit is closer to the truth.

Why are claims like "leadership" less believable in AI answers?

Because they're subjective evaluations lacking a verifiable factual anchor. An AI-generated "this company has strong leadership" is hard for readers to accept; whereas a checkable fact like "founded 12 years ago, served 300 clients" is inherently more believable. So focus on stacking facts, not adjectives.

Does structured data really affect how AI talks about me?

It raises the odds of being cited correctly. AI answer engines pull machine-readable fact sources; `Organization`, `FAQPage`, and `Product` JSON-LD feed "the correct version of you" straight to the model, lowering the chance it makes things up or grabs stale info.

My take

The mainstream narrative says "GEO enters the credibility era, come buy reputation management." My judgment is the opposite: for 99% of SMBs, a credibility score is a worry that's too early — you haven't even achieved steady citation, so fretting over "persuasion" is a luxury. The real priority is "accuracy," because AI getting your business scope, phone number, or price wrong is far more damaging than "that line wasn't persuasive enough." Service opportunity for ScriptWalker: package a small fixed-fee "AI visibility & accuracy health check" — quarterly, audit answers across 3 AI platforms, reinforce structured data, and list and fix what AI got wrong. It needs no expensive tools, just method and discipline — exactly what an agency can standardize and deliver.

Sources

• Burson official newsroom — The Credibility Paradox (first-party)
• PRCA Global — Burson Research Highlights Gap Between Visibility and Believability (first-party)
• CiteLens GEO platform GA launch (first-party)
• MM+M — Burson study: Leadership is AI's 'clearest liability' (third-party)
• Telum Media — Study Highlight: Burson's The Credibility Paradox (third-party)