Your Buyers Use AI to Find You. They Use Earned Media to Trust You.
Forrester's 2026 buying data reveals the two-step process every B2B purchase now follows. Most brands are only building for step one.
Forrester just published the most useful data on B2B buying behavior this year. Most coverage is treating it as an AI adoption story. It is not. It is a trust architecture story, and the implication for how you build brand authority is not what most people are drawing from it.
Here is the finding from Forrester's State of Business Buying, 2026, based on a survey of nearly 18,000 global business buyers: 94% of B2B buyers now use generative AI during their purchasing process. That number is everywhere right now, because it is striking. But it is the second number that changes what you do about it.
20% of those same buyers report feeling less confident in their decisions because AI gave them inaccurate or incomplete information.
So the picture is not "AI has taken over B2B research." The picture is more complicated. Buyers use AI to start the research process. Then they validate what they find against sources they actually trust — industry analysts, peers, subject matter experts, third-party editorial coverage. Forrester found that buyers are now more likely to engage a vendor based on input from industry experts than from AI recommendations.
The buyers using AI most extensively — and the ones who are most skeptical of it — are procurement professionals. Among procurement respondents, 28% said they felt less confident after AI-generated information, compared to 19% overall. These are the people who sign off on the deal. They are the most sophisticated users of AI in the buying process and the least likely to trust its output without verification.
The two-step process nobody is building for
The mistake most B2B brands are making is treating their AI visibility strategy as a single-layer problem. Get into the AI answer. Done.
That is only step one. And step one, by itself, does not close deals.
Here is what Forrester's data actually describes: a two-stage purchase journey. Stage one is AI-assisted discovery — buyers use generative AI tools to build an initial picture of the category, identify potential vendors, and get oriented fast. Stage two is human-assisted validation — buyers consult trusted external voices to check whether what AI told them is accurate, and to add the layer of credibility that AI cannot provide on its own.
A brand that appears in the AI answer but has thin third-party editorial presence gets through stage one and stalls in stage two. The buyer found you. But when they looked for confirmation — when they checked whether trusted industry publications had written about you, whether your name appears in the reporting that their peers and analysts already trust — you were not there.
Forrester's data shows the average B2B purchase now involves 13 internal stakeholders and nine external influencers. For purchases involving AI features, the buying group size doubles. Every one of those people is running their own version of this two-step process. Multiple people validating in parallel, across multiple external sources. The question is not just whether AI surfaces your brand. The question is whether the validation infrastructure exists when those people start checking.
Why buyers trust what they trust
The underlying logic is not complicated once you look at it directly.
AI engines and human validators are drawing from the same source pool. When a buyer asks ChatGPT who the most credible vendor in their category is, the model answers based on what it has seen in the publications it was trained on — Tier 1 editorial outlets, industry press, research reports. When that same buyer then goes to validate the AI's answer, they check the same publications.
The brands winning in stage two are the same brands that built strong AI visibility in stage one. Because they are the same infrastructure. Earned media in trusted publications is what AI engines cite when building their answers, and it is what human validators reference when checking those answers.
This is why investing in AI citation visibility without earned editorial coverage produces a fragile strategy. Technical optimization — schema markup, structured content, entity signals — can improve how AI engines process and extract information about your brand. But there is a ceiling. AI engines will not confidently cite a brand that lacks independent editorial coverage in publications they treat as authoritative. And human validators will not confirm a brand that has no third-party press history to check.
The Ahrefs analysis of 75,000 brands found that brand web mentions — the direct output of earned editorial coverage — correlate 0.664 with AI visibility. Backlinks, the currency of SEO, correlate 0.218. That ratio tells you which infrastructure actually matters. The gap between them is the strategy.
The validation squeeze
There is a compression effect happening that makes this more urgent.
As more buyers use AI for initial research, the initial discovery phase is getting faster and cheaper. AI compresses the first 60% of the research process. The final 40% — the validation and trust-building that actually determines vendor selection — has gotten harder, not easier.
Buyers are more skeptical. Buying groups are larger. The information environment is noisier. And the brands that built their strategy around paid visibility and owned content are discovering that neither shows up in the validation layer.
Forrester documents B2B website traffic declines of 10–40% over the past year. Buyers are spending less time on vendor websites, not because they are less engaged with the category, but because AI is doing the early orientation work that used to require direct site visits. What that means practically: you have fewer opportunities to earn trust through owned channels, and more pressure on the trust signals that exist outside them.
The brands navigating this well are not the ones with the most sophisticated content marketing programs. They are the ones with the strongest footprint in the editorial sources that both AI engines and human validators reference. They are in Forbes when their category comes up. They are cited by Reuters when industry reporters cover their space. Their executives are quoted in TechCrunch when the space is moving. That footprint is what closes the trust loop. The Muck Rack Generative Pulse analysis of over one million AI prompts confirmed that 82% of links cited by AI systems come from earned media — Reuters, Financial Times, Forbes, Axios. The same outlets human validators check when they go looking for proof.
Machine Relations as the operating model
The two-step buyer journey Forrester describes is the same mechanism that Machine Relations names as the defining challenge of AI-era brand strategy.
PR always worked because earned media in trusted publications was the most powerful trust signal that existed. That mechanism did not change when AI engines became the primary research interface. What changed is who is reading. The AI engines that now handle stage one of buyer research are pulling from the same editorial sources that shaped human brand perception for decades. The human validators in stage two are checking those same sources.
Machine Relations is what happens when you run this logic to its conclusion: earned media is no longer just a brand play. It is citation infrastructure — for the machines doing the initial synthesis and for the humans doing the final verification. The only editorial coverage that does both is the kind you build through real relationships with real publications over time.
That is not a vendor pitch. It is what Forrester's own data on buyer behavior now confirms. The brands that will win in this environment are the ones with the presence to pass both tests. Building for stage one only is a strategy for being discovered and not trusted.
If you want to see where your brand currently stands in that two-step process, the visibility audit at app.authoritytech.io/visibility-audit shows exactly how AI engines are representing you — and what is missing in the editorial layer that drives validation.