Your Business Has Two Customers Now. You're Only Marketing to One of Them.

Here's a question most business owners haven't thought to ask: when a potential customer asks ChatGPT "who's the best plumber in Adelaide," does your business show up?

Not on Google. Not on Facebook. In the AI's answer.

Because that's where a growing number of buying decisions now start. According to the IBM-NRF Consumer Study of 18,000 consumers across 23 countries, 45% of shoppers now use AI tools during their buying journey. They're using AI to research products (41%), interpret reviews (33%), and hunt for deals (31%). And this isn't some far-off prediction. It's data from late 2025.

The implication is simple and uncomfortable. Your business now has two customers: the human who eventually picks up the phone or fills in the form, and the AI agent that helped them get there. You're spending your entire marketing budget talking to the first one. You're barely whispering to the second.

The $20.9 Billion Nobody Told You About

The scale of this shift is accelerating faster than most business owners realise.

This isn't limited to online retail. When a homeowner asks an AI assistant to "find me a pest control company with good reviews in my area," an agent evaluates businesses on their behalf. When a business owner asks their AI to "compare digital marketing agencies in Adelaide," the AI reads your website, checks your reviews, scans your citations, and makes a recommendation. All before the human sees a single result.

Byron Sharp's Missing Third Pillar

Byron Sharp's How Brands Grow identified two forces that drive brand growth: mental availability (being thought of when a need arises) and physical availability (being easy to find and buy when someone acts on that need).

For decades, this framework has explained why brands win. You need people to think of you, and you need to be there when they look. Every Google Ad you run, every SEO strategy you implement, every directory listing you claim is an investment in one of these two pillars.

But Sharp's framework was built for a world where humans do the thinking and the finding. In 2026, there's a third player in the system: the AI agent that sits between the human's need and their action.

Call it machine availability: your brand's propensity to be surfaced, understood, and recommended by AI systems when they evaluate options on a buyer's behalf.

Sharp's three pillars of physical availability (Presence, Prominence, Relevance) now have machine-readable equivalents:

If you're invisible to AI agents today, you're invisible to a growing share of tomorrow's buyers. We explored a related angle in why your website needs to be a source, not just a brochure. But being a source for search engines is table stakes. Being a source for AI shopping agents is the next level.

Cialdini's Social Proof at Machine Scale

Here's where it gets interesting. Robert Cialdini's principle of social proof has always been one of the most powerful forces in marketing. People look to others' behaviour to determine the correct action, especially when they're uncertain. That's why reviews, testimonials, and "500+ businesses trust us" messaging works.

AI agents take this principle and put it on steroids.

When a human reads reviews, they skim. They might read three or four, notice the star rating, and form a gut feeling. That's Kahneman's System 1 thinking at work: fast, emotional, heuristic-based.

When an AI agent reads reviews, it reads all of them. It weighs recency, sentiment, specificity, and volume. It cross-references what your reviews say against what your website claims. And it does this across every competitor simultaneously.

The data backs this up. Hexagon's citation analysis found that review velocity (how frequently new reviews come in) has a 0.67 Pearson correlation with whether an AI recommends a brand. That's a remarkably strong signal. Not just having reviews. Having fresh reviews, consistently.

Meanwhile, Bazaarvoice research shows that while 66% of shoppers use AI for product discovery, 92% still say reviews and real customer photos are essential before buying. Reviews are doing double duty: they convince the AI to recommend you AND convince the human to trust that recommendation.

Cialdini's authority principle follows a similar pattern. Brands cited in 5 or more independent editorial sources are 4.7x more likely to be recommended by AI assistants than brands with fewer than two third-party mentions. Third-party editorial mentions from high-authority publications are 3.2x more predictive of AI citations than the volume of content on your own website.

Think about what that means. You can publish 50 blog posts on your own website and it matters less than being mentioned in 5 industry articles you don't control. For years we've been told "content is king." For AI agents, third-party credibility is king.

The Sutherland Paradox: What Happens When the Customer Can't Be Charmed?

Rory Sutherland built his career on a beautiful insight: perception beats reality. The London Underground's greatest improvement in passenger satisfaction wasn't faster trains. It was dot-matrix display boards showing when the next train would arrive. The wait didn't change. The uncertainty did.

His principle of costly signaling explains why expensive advertising works: if a company invests heavily in promoting something, consumers rationally infer confidence in the product. "A flower is simply a weed with an advertising budget."

But here's the paradox. AI agents don't experience uncertainty the way humans do. They don't feel reassured by a beautiful website. They don't infer quality from expensive branding. They read structured data, cross-check citations, and evaluate review patterns.

Does that mean Sutherland's psychology doesn't apply?

Not exactly. Recent research from Columbia and other universities reveals something fascinating: AI agents reproduce human-like heuristics and biases, including position bias, endorsement salience, and price-quality associations. But they respond in ways that are "both more extreme and more predictable" than humans.

In other words, the psychological principles that govern human buying behaviour don't disappear when AI agents shop. They get amplified. An AI agent that's trained on millions of human purchasing decisions carries those same biases, just in a more systematic, consistent way.

Sutherland's costly signaling still works, but the signal has changed. It's no longer "we spent a lot on this TV ad, so we must be confident in our product." It's "we have 247 reviews with a 4.8 average, we're cited in 6 independent publications, and our structured data is comprehensive and accurate." The signal of investment has shifted from media spend to data infrastructure.

As Sutherland himself noted in a recent conversation, his interest now centres on "the interaction between AI and humans and the extraordinary importance in human decision-making with context." The context has changed. And the businesses that understand the new context will win.

The Trust Stack: Human Layer vs Machine Layer

This dual-customer reality creates what we'd call a trust stack. You need trust signals that work on both layers simultaneously.

This is why your marketing dashboard might be misleading you. The metrics most businesses track (impressions, clicks, conversions) measure human behaviour. They tell you nothing about whether AI agents can find you, understand you, or recommend you.

The 78% Concentration Problem (And Why SMEs Should Care)

Remember that Hexagon stat: the top 2% of brands capture 78% of AI recommendations. That sounds terrifying for small businesses. But there's a critical nuance.

Those top brands aren't winning because they're big. They're winning because they've built what Hexagon calls minimum viable authority: a combination of at least six of eight measurable signals. Many of those signals are achievable for any business willing to invest in the basics.

Here's the SME advantage. Most large brands have messy data infrastructure. They have outdated directory listings, inconsistent business information across platforms, and review response rates below 20%. A small business that systematically gets these fundamentals right can outperform a much larger competitor in AI recommendations.

This mirrors Sharp's observation about physical availability: the businesses that make themselves easiest to buy from grow fastest. The definition of "easy to buy from" now includes "easy for an AI to recommend."

What This Means for Your Business

You don't need to overhaul your marketing. You need to add a machine layer to what you're already doing. Here are the highest-impact actions:

None of this replaces your Google Ads, your Facebook campaigns, or your landing page optimisation. Those still reach the human customer. But they don't reach the AI agent that increasingly shapes what the human sees before they ever click your ad.

The businesses that build for both audiences will compound their advantage. The ones that keep marketing to humans alone will gradually wonder why their phone stopped ringing, even though their ads still look great.

You have two customers now. Make sure both of them can find you.

Further Reading

• IBM-NRF Study: AI Shapes Consumer Decisions Before Shopping Begins - 18,000-consumer global study on AI's role in the buying journey
• Why AI Search Engines Recommend Some Brands Over Others (Hexagon) - Data analysis of 100,000 AI citations revealing the authority signals that drive recommendations
• What Is Your AI Agent Buying? (Columbia/ACM Research) - Academic research on AI agent purchasing biases and behaviour patterns
• How Agentic AI Will Reshape Shopping in 2026 (eMarketer) - Market projections for AI-driven commerce
• Rethinking Decision-Making: Rory Sutherland on AI and Human Behaviour - Sutherland's evolving perspective on AI's impact on behavioural economics