The First Buyer in Your Funnel Isn't Human Anymore

There's a new evaluation happening before your prospect ever lands on your website. An AI tool has already compared you to three competitors, noted that your product descriptions lack specificity, and flagged that you don't show up in structured data feeds.

And it may have recommended someone else.

That's not a future scenario. That's what's happening right now, and the research is catching up to what a lot of us in performance marketing have been watching build for over a year.

McKinsey just published findings from a consumer survey across France, Germany, and the United Kingdom that should make every growth leader pay attention. Eighty-four percent of European consumers report using AI tools in everyday life. More importantly, 38 percent are already using AI to research products and shape purchase decisions. That number will look small in two years.

North America Is Running a Similar Playbook

The European data isn't an outlier. Bain & Company surveyed more than 2,000 U.S. consumers and found that 72 percent have already used AI in some form. A Commerce and Future Commerce survey of consumers across the U.S., U.K., and Australia found that 33 percent of Gen Z and 26 percent of Millennials already prefer AI platforms for product research over traditional search.

Adobe tracked a 1,200 percent surge in retail traffic originating from generative AI over a six-month period. Retailers didn't build for that traffic. It just arrived.

The pattern is consistent on both sides of the Atlantic. AI is collapsing the top and middle of the funnel into a single, rapid evaluation layer. Research that used to take 45 minutes across a dozen browser tabs now happens in a two-minute conversation.

Where AI Is Showing Up in the Journey

The McKinsey European data is specific about where AI is being used:

• 63 percent are using it to compare brands, models, prices, and reviews

• 55 percent are using it to learn about a category or product

• 46 percent are using it for discovery and inspiration

Usage drops significantly as the journey moves toward execution. Fewer people are using AI to build carts or complete checkout, at least for now.

That pattern is the critical insight. AI is dominant upstream, in the moments where preferences form and shortlists get built. By the time a buyer clicks through to a website, much of the evaluative work is already done. For your paid search and SEO investment, that matters enormously because the question of who makes the shortlist is being answered somewhere you're probably not optimizing for.

By the time a buyer clicks through to your website, much of the evaluative work is already done.

The Trust Gap Is Real, and It Tells You Something Useful

Both the McKinsey European research and the Bain U.S. data point to the same pattern: consumers trust AI for judgment but not for action.

Bain found that while 72 percent of U.S. consumers have used AI tools, only 24 percent are comfortable letting AI complete a purchase. Just 10 percent report having actually bought something through AI; mostly small-ticket grocery and household items. The Wildfire Systems consumer survey (1,000 U.S. adults, July 2025) put the number who are "very comfortable" letting an AI agent find products, confirm prices, and place an order at 20 percent.

The top barriers from U.S. consumers:

• data privacy at 49 percent

• unwanted subscriptions at 44 percent

• overspending at 41 percent

• incorrect product selection at 38 percent

McKinsey frames this well: consumers aren't rejecting AI judgment. They're resisting unbounded authority. They're comfortable when AI helps them reason through a decision. They get nervous when AI starts operating independently, especially for recurring or high-stakes actions.

What that tells you is that the adoption curve for full autonomous purchasing is slower and more conditional than the hype would suggest. Discovery and evaluation are already AI-mediated. Execution is going to take longer. That gap matters for planning. What you need to do in the next 12 months is different from what you'll need to do in the next 36.

There's also a more bullish signal buried in the Bain data. While only 24 percent are comfortable completing purchases via AI today, 64 percent say they have used or are open to using AI to complete a purchase, and 73 percent have used or would consider using AI to research and compare products. The base of potential adoption is much larger than current behavior suggests.

This Isn't Just a Retail Problem

The McKinsey research focuses on consumer purchasing, and most of the commentary around agentic commerce follows suit; product discovery, checkout, basket assembly, replenishment. That framing is understandable, but it's leaving a large portion of mid-market businesses thinking this doesn't apply to them yet.

It does. The mechanism is the same. Only the transaction looks different.

Think about what's actually happening when a facilities manager is researching commercial HVAC contractors in their region. Or when a CFO is shortlisting outsourced accounting firms. Or when a patient is deciding between two elective surgery providers, or a company is evaluating marketing agencies, or a business owner is looking for a commercial insurance broker. None of those buyers are using AI to autonomously complete a transaction. But they are using it for exactly what the McKinsey data shows is the dominant behavior right now: comparing options, learning about a category, narrowing a consideration set before they ever pick up the phone or submit a contact form.

The purchase comes later, through a human process. But the shortlist was built by AI.

For service businesses, the discoverability problem is arguably more acute than it is for product companies. Product companies have SKUs, structured data feeds, retail listings, and review platforms built specifically for their category, the kind of structured, machine-readable information AI systems can evaluate with confidence.

A commercial contractor, a professional services firm, a healthcare provider, or a B2B agency has to be legible from a much thinner set of signals: the completeness of their Google Business Profile, the volume and specificity of their reviews, schema markup on their website, citations in industry directories or local publications, and whether their service descriptions are specific enough for an AI system to understand what they actually do and who they do it for.

If someone asks ChatGPT or Perplexity for the best commercial roofing contractors in their city, or the top-rated elective orthopaedic surgeons in their region, and your website says "we provide quality solutions for businesses of all sizes", you're not making the shortlist. Not because the AI evaluated you and found you lacking. Because it couldn't tell what you do with enough confidence to include you.

The trust gap also runs deeper in high-consideration service purchases than it does in product commerce. Nobody is going to let an AI autonomously book their LASIK surgery or sign a contract with a managed services provider. Which means the discovery and comparison phase is where nearly all of the AI influence lands for those buyers. The stakes of being absent from that phase are higher, not lower, than they are in retail.

That's true whether you're selling commercial restaurant equipment or commercial cleaning contracts. The channel is different. The mechanism is identical.

Brand Loyalty Isn't Dead. It Just Needs to Be Encoded.

Which raises a question that applies to every category: if AI is now the first evaluator, what happens to the brand equity you've spent years building?

One of the more nuanced points in the McKinsey research is what happens to brand loyalty in an AI-mediated world. Their position is that loyalty doesn't disappear it gets reconfigured.

I think that's right, and I'd push it further.

Buyers will still carry brand preferences into AI interactions. Some will explicitly tell their AI tools which brands to include or exclude. Others will have preferences inferred from purchase history, stated values, or prior interactions. What changes is that loyalty has to be legible to machines, not just emotionally resonant to humans.

Loyalty has to be legible to machines, not just emotionally resonant to humans.

A brand that wins on visual identity, placement, and advertising spend will struggle when the first evaluator isn't a person scrolling a results page. A brand that has structured its differentiation as clear, comparable, evidence-backed claims has a real shot at surviving the AI filter.

If the evaluation layer is now algorithmic rather than positional, the quality and specificity of your positioning matters more than your media budget. That's actually an opportunity for mid-market companies that have genuine differentiation but have spent years unable to match larger competitors on ad spend. The playing field doesn't level but the rules change in your favor if your positioning is specific and your signals are strong.

If the evaluation layer is now algorithmic rather than positional, the quality and specificity of your positioning matters more than your media budget.

Why the Strategic Answer Keeps Coming Back to Foundations

I want to be direct about something, because I've been making the same underlying recommendation across client conversations for over a year and it deserves a fuller explanation than "fix your foundations before you add AI." That framing is correct, but the data now shows exactly how it plays out and the numbers make the case better than any principle does.

Seer Interactive analyzed 25.1 million organic impressions across 42 organizations and found that brands cited in AI-generated responses earn 35 percent more organic clicks and 91 percent more paid clicks than non-cited competitors on the same queries. The brands that don't get cited suffer the full CTR decline; 61 percent for organic and 68 percent for paid on queries where AI overviews appear. That gap isn't theoretical. It's already showing up in pipeline data for companies that haven't noticed yet because they're still looking at ranking reports instead of citation reports.

The mechanism behind that gap is what keeps bringing the answer back to infrastructure. A separate analysis of citation probability found that pages outside the top 10 in traditional search have a 75 percent lower chance of appearing in AI-generated responses than pages ranking on page one. AI systems don't crawl the full web for every query. They pull from pre-filtered source sets already shaped by technical signals, content structure, and authority, the same signals that have always determined organic search performance. If you haven't done that work, you're not just behind in search. You're largely absent from the AI evaluation layer built on top of it.

The filter is faster, the shortlist is shorter, and there's no page two.

BrightEdge research found that sites implementing structured data and FAQ schema saw a 44 percent increase in AI search citations. Content with verifiable, specific statistics gets 22 percent more AI citations. Using direct quotations and attributed claims boosts that by 37 percent. These aren't AI-native tactics. They're the same discipline that good content strategy has always required; specificity, structure, credibility signals, freshness. What's changed is the penalty for skipping them.

The companies that keep losing ground as each new algorithmic layer is introduced are often the same companies. Not because they've made bad strategic decisions, but because they keep deferring the infrastructure work that every one of these systems depends on. AI search doesn't create that problem. It surfaces it faster and makes it considerably harder to recover from.

What Needs to Actually Change

Most mid-market strategies right now are treating AI discoverability as a content marketing initiative or something to layer on top of what already exists. That's the wrong framing. The gaps are structural, and they compete for the same budget and attention as campaigns with measurable short-term returns.

The first gap is the one nobody wants to fund.

McKinsey calls it machine legibility — rich, structured metadata, consistent taxonomy, evidence-backed claims, strong review signals, and third-party citations that give AI systems something concrete to evaluate. Most companies don't have this, not because they haven't heard about it, but because it's unglamorous infrastructure work that doesn't produce a dashboard you can show at a monthly review. It won't feel urgent until the pipeline starts showing unexplained drops in top-of-funnel volume. By then you're rebuilding from a deficit.

The second gap is measurement.

Most companies have no idea how they're appearing in AI-generated responses for their category right now. That's not an excuse anymore it's measurable today with tools like Profound, Otterly, and SE Ranking's AI overview tracking. These tell you whether you're being cited, what questions are surfacing your brand, and where you're being filtered out. If you're not tracking this, you're making budget decisions without visibility into a channel that's already influencing your buyers.

The third gap is first-party data.

If AI agents are going to carry buyer preferences, purchase history, and stated values into evaluations, the companies that hold direct customer relationships have a structural advantage over those relying on platform data from retailers, marketplaces, or paid channels. Buyers whose preferences are stored inside those platforms may become invisible to the agents acting on their behalf when those platforms aren't in the loop.

Building direct data relationships with your customers was already overdue. This is another reason it can't wait.

Execution readiness for agent-initiated transactions is the one that's further out, but worth planning for now. Bain's data shows that 64 percent of U.S. consumers have used or are open to using AI to complete a purchase. OpenAI, Stripe, Google, Mastercard, and Visa are all actively building the payment and transaction infrastructure that will make that normal. When your category tips, is your checkout architecture ready to receive transactions from autonomous agents? Do you have the authorization logic and merchant-side controls that agentic commerce requires? This planning work takes longer than you expect, and most teams haven't started it.

Finally, the sequencing question within your own category matters more than most businesses are admitting. Not everything is moving at the same pace. McKinsey's data shows AI adoption is currently highest in research-intensive categories of apparel, healthcare, consumer electronics, and travel where buyers face real complexity and benefit most from synthesis and comparison. If that's your category, the urgency is higher than you probably have it on your priority list. If you're in a category where physical validation or in-person experience still dominates the final decision, you have more runway. The important thing is to know which situation you're actually in, rather than assuming the timeline that feels most comfortable.

The Compounding Problem

Here's what makes this genuinely time-sensitive rather than just strategically important.

The brands that are machine-legible, well-cited in authoritative sources, and consistently surfaced in AI-generated responses now are building a structural advantage that compounds. AI systems are shaped by the data available to them. Brands that show up reliably in AI training data and live inference today will have a disproportionate presence in how those systems evaluate categories in the future.

This isn't like ad spend, where pulling back immediately removes your visibility. It's more like organic SEO: the signals take time to build, the results take time to appear, and the gap between companies that started early and companies that waited keeps growing.

The funnel has a new first touchpoint, and it's not your homepage. It's whatever an AI system decides to say about you when a buyer asks.

The companies building something worth saying right now are going to be very hard to catch in three years.

Sources:

McKinsey & Company European Consumer Survey (France, Germany, UK, 2025); Bain & Company / ROI Rocket U.S. Consumer AI Payments Survey (n=2,016, March 2025); Commerce + Future Commerce / Centiment Survey (n=1,000, US/UK/AU, June 2025); Wildfire Systems / Big Village U.S. Consumer Shopping Trends Report (n=1,000, July 2025); Adobe Commerce GenAI Traffic Data (via PwC); McKinsey Agentic Commerce Opportunity Report (October 2025); Morgan Stanley Agentic Commerce Outlook (2025); Seer Interactive AI Overviews Impact Study (25.1M impressions, September 2025); BrightEdge Structured Data and AI Citation Study (2025); Dataslayer / AI Overview CTR Analysis (2025); getpassionfruit.com AI Citation Probability Study (2025).