Your buyer already shortlisted you — or didn't — before the first call

Five hours in AI search. One hour with sales. That's the ratio Lara McCulloch at Start Some Shift published on May 26 — and if you run pipeline marketing for a B2B company, that number should stop you mid-scroll.

The analysis found that for every hour a B2B buyer spends with a vendor's sales team, they've already spent approximately five hours researching independently — and the majority of that research now happens inside ChatGPT, Perplexity, and Google's AI Mode, not on Google itself.

The shortlist is built before sales picks up the phone. By the time someone fills out your contact form, the decision to engage has already been made inside an AI conversation no one at your company ever sees.

What the data actually says

This isn't one data point. It's a pattern corroborated across multiple sources this spring.

G2 surveyed 1,076 B2B software buyers in March 2026 for their Answer Economy report. 51% now start their purchasing process in an AI chatbot rather than a search engine. 71% say they rely on AI chatbots for software research — up from roughly 60% just seven months ago. And 69% said they chose a different vendor than they initially planned based on AI chatbot guidance. One in three bought from a vendor they'd never heard of before.

Tim Sanders, Chief Innovation Officer at G2, framed it plainly: the Yellow Pages compressed the market into the big book, Google compressed it into the first page of results, and now AI chatbots are compressing it into a single answer.

The implication for brand-new SaaS companies is actually good news. Unknown challengers are being lifted into deals they would never have reached through traditional channels. The uncomfortable flip side: category leaders are being quietly excluded from consideration sets they used to own by default.

Where the budget is pointing

Here's the structural problem. Most B2B marketing programmes are designed for the sales-led portion of the journey — the demo, the proposal, the closing call. That is now the final fraction of a process that has already been mostly decided in research the vendor cannot observe.

Start Some Shift's McCulloch put it directly: "Marketers are still optimising for the conversation they can see. And losing the one they can't."

I run marketing ops at a Copenhagen SaaS and I feel this every week. Our HubSpot attribution tells me a lead came from a LinkedIn ad or a Google campaign. It doesn't tell me the buyer had already shortlisted us (or removed us) three weeks earlier inside a ChatGPT thread I'll never read.

That invisible research phase is now the most consequential stage of your funnel. And it's the one most marketing teams have zero budget pointed at.

The contrarian read

The obvious response is to chase AI visibility — structured content, schema markup, get on G2, optimise for Perplexity citations. All worth doing.

But here's what I'd push back on: most of the "optimise for AI search" advice treats this like an SEO problem with a new algorithm. It isn't.

G2's own data shows buyers aren't just asking AI tools where to look. They're asking AI to synthesise everything and return a shortlist. That means your messaging clarity, your category positioning, the specificity of your customer outcomes — all of that is now what AI extracts and presents. Vague superlatives get dropped. Specific claims with verifiable outcomes get cited.

A case study behind a gated form is something Gemini will never surface. A beautifully designed PDF with text embedded in images is invisible to Copilot. The content decisions you made for lead capture are actively working against you in the research phase that matters most.

The fix isn't a new SEO checklist. It's deciding whether your company actually has a specific, defensible point of view — and then making that point of view machine-readable.

What to actually do

Start with a fifteen-minute test. Open ChatGPT, Perplexity, and Google AI Mode. Type the prompt your best-fit buyer would use. See who shows up. See what those tools say about the vendors that do. See if your company appears at all — and if it does, check whether the AI's description of you is actually accurate.

Most B2B brands have zero visibility into whether these tools are recommending them, ignoring them, or getting their product wrong. Your traditional analytics dashboard doesn't capture any of this. That's the measurement gap worth solving before the content gap.

The second step is structural. Fast Company's analysis of the same trend puts it well: buyers who arrive at your website after an AI-assisted shortlisting process have already formed an impression. What they find needs to confirm it, not explain your product from scratch. Most B2B websites are still built to explain. They need to be built to validate.

Open your chatbot of choice right now and type in the prompt your ideal buyer would use. What comes back will tell you more about where your brand actually stands in 2026 than anything in your current dashboard.

Is your company showing up in those answers — and is what the AI says about you accurate?