Your Buyers Dropped Google. They're Asking ChatGPT. The Answer Was Written 90 Days Ago.
47% of enterprise tech buyers now start vendor research with AI, not Google. The credibility AI surfaces was built from the last 90 days of earned media — or it wasn't built at all.
The vendor shortlist is being built before your sales team gets involved.
Last Tuesday, Treble — a B2B tech PR agency — released its Press-to-Pipeline Activation in the Age of AI report. They surveyed 300 senior B2B buyers — CIOs, CISOs, CTOs, VPs, Directors — across retail, finance, healthcare, manufacturing, and telecom. The finding: 47% of enterprise technology buyers now start vendor research with AI assistants like ChatGPT and Google Gemini. That number sits ahead of Google Search (43%), vendor websites (42%), and trade publications (40%).
AI is the #1 research starting point in enterprise buying. Not a trend — a fact, with a citation, and a sample of people who sign the purchase orders.
There's a follow-on number that most teams are still missing: 93% of buyers use AI to summarize or compare vendors once they're in evaluation mode. They open with AI. They close with AI. The entire discovery-to-shortlist arc runs through a machine with one job: surface the vendor with the most credible third-party consensus.
Not the best product. The most credible-looking one, according to sources the machine has already aggregated.
The industry got the diagnosis right and the treatment wrong
The reflex response to AI search domination has been GEO — generative engine optimization. Track share of model. Implement structured data. Map content to AI answer patterns. These tactics are real. They matter.
But they solve the wrong problem.
You cannot GEO-optimize your way onto a shortlist being built from sources that don't exist yet. GEO assumes you have a body of third-party content the AI can surface. If that content doesn't exist — if your brand has been quiet, self-promotional, or only publishing vendor-owned assets — the AI doesn't lower your rank. It removes you from the conversation entirely.
High-maturity brands now spend nearly twice as much on GEO as their peers, according to a February 2026 analysis in Martech. That spend matters. But GEO amplifies a signal that has to already exist. It can't manufacture one from nothing.
The Treble report explains the mechanic directly: "AI does not generate credibility; it aggregates it, drawing from earned media, analyst reports and third-party validation. Companies without consistent, recent coverage risk becoming invisible in AI-generated comparisons before a salesperson ever enters the conversation."
This is not a search optimization problem. It's a trust aggregation problem. The solutions are completely different.
The number that defines the new operating window: 90
Ninety-two percent of enterprise buyers say coverage published within the past 90 days is important to vendor credibility. Not the Forbes placement from 2023. Not the thought leadership series you ran pre-product launch. The last 90 days. Rolling. Always.
This means AI-driven vendor perception isn't a mountain you climb once. It's a treadmill running at a fixed pace. The moment you step off — halt earned media for a quarter, go dark on third-party placements, stop generating analyst interactions — your signal degrades.
A vendor that was well-covered in Q3 and silent in Q4 is not just less visible. According to 92% of buyers, they're less credible. The absence reads as a signal itself.
KnewSearch's 2026 AI Search Visibility Benchmark reinforced this from the platform side. After analyzing over 52,000 AI queries across ChatGPT, Perplexity, Gemini, Claude, and Google AI Overviews, they found that the top three brands in any given category capture 67% of all AI mentions. The math is unforgiving: if you're not among the top three by credibility signal, you're splitting the remaining third of AI mentions with every other brand in your space.
The concentration problem runs deeper than simple ranking. Research published by Search Engine Land in February 2026 found that ChatGPT brand recommendations are highly probabilistic — running the same 12 prompts 100 times produces materially different brand recommendations each time. The implication: brands with thin third-party signal don't just rank lower, they disappear from AI responses entirely when the probabilistic dice don't land their way. Density of earned signal is what creates consistency. Sparse coverage produces inconsistent, unreliable AI mentions.
The top three win because they've built coverage density. Everyone else is hoping the dice roll in their direction.
Your pipeline is already pre-filtered
There's an implication here that should land hard.
Your SDRs are working leads that already have a mental model of your brand — or don't. That model was built by a machine, using content that either exists or doesn't. By the time your sales team touches a prospect, the evaluation has already started. Often, the shortlist is already formed.
You're selling in round two of a game where round one happened without you.
Research from Recon Analytics published in February 2026 found that even enterprise distribution advantages don't protect against this. Microsoft Copilot lost 7.3 percentage points of paid subscriber share in seven months — despite deep Office 365 integration — while Google Gemini gained ground. The finding: "Distribution advantages do not lock in market position. The platform that delivers the most reliable results wins, regardless of how it was deployed." The same principle applies to vendor shortlisting. Being embedded in the buyer's stack doesn't put you in their AI-generated comparison.
The reason traditional PR metrics failed to predict this is clear: coverage was measured by volume and placement, not by what that coverage fed downstream. An earned media hit was a brand asset. That framing is now operationally wrong.
Coverage is now training data. Every press release, media mention, analyst report, and third-party byline feeds the AI systems buyers consult first. As Matt Grant, EVP at Treble, said: "PR has always been about shaping perception. What's changed is that the perception is now shaped by an algorithm that draws on earned media."
The algorithm isn't reading your website copy. It's reading what others say about you — and weighting the recency of that signal hard.
Related Reading
- AI Visibility for SaaS Companies: The 2026 Earned Media Playbook
- AI Visibility for Fashion: The 2026 Earned Media Playbook
The operating model this demands
Earned media has always shaped AI search results — but the mechanism is now the primary B2B discovery engine, not a secondary awareness play. That changes what it means to run it.
Traditional PR was episodic. You had a launch, pitched it, got coverage, moved on. The goal was awareness. The cadence was campaign-based.
What this data demands is continuous. The AI systems buyers are using to build shortlists are always-on — pulling, aggregating, and reranking based on the freshest third-party signal available. Sixty to ninety days of earned media silence isn't a tactical gap. It's a pipeline event. The buyer who opened ChatGPT this week to compare vendors saw a shortlist that may not have included you — and that moment is already gone.
This is why Machine Relations is the right frame for the era we're in. Not a PR campaign with a Machine Relations wrapper. An operating model built on the premise that machines — LLMs, AI agents, recommendation engines — are now primary stakeholders in how buyers perceive your brand. The 90-day window isn't a GEO opportunity. It's the new definition of market presence.
If you want to know where your brand stands today in AI-generated buyer research — what the machine is already saying, what's missing, and what earned media gaps are costing you in shortlist inclusion — start with the free visibility audit.
The buyers have moved. The research habit has shifted. The only variable is whether the media record feeding AI reads like a credible vendor — or a company that went quiet when it mattered most.