AI Is Becoming the First Step in the Buying Journey
McKinsey research shows half of consumers now seek out AI-powered search, with up to $750 billion in US revenue at stake by 2028. Here is what that means for AEO, GEO, and the emerging discipline of AX.
Jeremy Prasetyo
Co-CEO, Lanesra AI
The shift from traditional search to AI-powered discovery is no longer a future scenario. It is already changing how people research categories, compare products, evaluate brands, and make decisions.
McKinsey's AI Discovery research found that half of surveyed consumers intentionally seek out AI-powered search, and projects that as much as $750 billion in US revenue could flow through AI-powered search by 2028. It also estimates that brands unprepared for the shift could see 20% to 50% of traditional-search traffic at risk.
The short answer
Search is moving from a list of links to a decision layer. GEO helps your brand become usable evidence for that layer. AEO goes one step further: it asks whether the AI actually mentions, compares, recommends, and positions your brand when a customer is making a decision.
1. AI Search Is Already Part of the Entire Decision Journey
AI-powered search is not only used at the top of the funnel. In McKinsey's survey of 1,927 US consumers, more than 70% of AI-search users reported using it to learn broadly about categories, brands, products, or services. But usage remained high further down the funnel: 57% used it for personalized recommendations, 61% for comparing specific products or services, and 60% for summarizing reviews.
Lanesra visualization based on McKinsey's AI Discovery Survey.
This matters because the decision can increasingly form before the click. A consumer can ask what to consider, identify suitable brands, compare specific products, summarize reviews, and narrow the shortlist inside one AI conversation.
2. AI Search Is Becoming a Preferred Starting Point
Among AI-powered search users in the same survey, 44% ranked AI-powered search as their primary and preferred source of information for researching or making purchase decisions. Traditional search engines followed at 31%, retailer or brand sites at 9%, and review sites at 6%.
Source data: McKinsey's AI Discovery Survey. Figures may not sum to 100% because of rounding.
The implication is not that traditional search disappears. It is that AI search becomes an additional front door - and often an earlier one. When users finally reach a website, they may already have a shortlist, a comparison frame, and a preliminary opinion about the brand.
3. Why Strong Brands Can Still Be Invisible
Traditional SEO gives marketers significant control over their own domain. AI search is different because the answer can be assembled from a much broader evidence layer.
McKinsey's analysis found that, in many cases, a brand's own sites represented only 5% to 10% of the sources referenced by AI search. The rest may come from publishers, affiliates, user-generated content, reviews, research, marketplaces, media, and other sources. The exact mix varies by model, location, category, and question type.
Lanesra interpretation of McKinsey's reported 5-10% owned-site source share. The 90-95% range is the mathematical complement and should be read as directional, not as a universal rule for every model or query.
This is the strategic break with classic SEO: you can rank well, have strong brand awareness, and still fail to appear in an AI answer if the external evidence layer does not support you strongly enough.
THE LANESRA FRAMEWORK
4. GEO Solves the Evidence Problem. AEO Adds the Recommendation Problem.
McKinsey frames the emerging discipline primarily as GEO - generative engine optimization: improving the chance that a brand's information is discoverable, credible, structured, and usable by generative systems.
That is necessary, but from an operating perspective marketers need one more layer.
At Lanesra, we use AEO - answer engine optimization - to describe the next question: once the evidence is available, does the AI actually select your brand for the answer?
| Layer | Primary question | What to measure |
|---|---|---|
| SEO | Can search engines discover and rank us? | Rankings, traffic, crawlability, technical quality |
| GEO | Can generative systems find and trust our evidence? | Citations, source coverage, authority, factual consistency |
| AEO | Do AI answers mention, compare, and recommend us? | Mention rate, recommendation rate, position, sentiment, accuracy |
| AX | Can an AI agent evaluate and act on our offering? | Machine-readable product facts, availability, policies, transaction readiness |
The SEO -> GEO -> AEO -> AX progression is the Lanesra Framework.
This extension also reflects the direction of the underlying products. OpenAI says ChatGPT Search can retrieve current web information and include source citations, while Google describes AI Overviews and AI Mode as experiences built from multiple web sources. OpenAI · Google Search Central
5. AEO Changes What Marketers Need to Monitor
A brand can technically be present in the evidence layer and still lose the commercial answer. That is why AEO measurement needs to go beyond citations.
Visibility
How often does the brand appear for commercially relevant prompts?
Recommendation
Is the brand actively suggested, or merely mentioned in passing?
Position
Is the brand first, one of several options, or buried at the end?
Sentiment
Which strengths, weaknesses, risks, and objections does AI associate with it?
Accuracy
Are pricing, integrations, availability, and product claims represented correctly?
Sources
Which pages and third-party sources are shaping the answer?
That is the difference between asking "Are we indexed?" and asking "When the AI explains this category, does it understand us correctly and choose us when we fit?"
6. What Marketers Should Do Now
McKinsey found that only 16% of brands in a small Fortune 500 CMO survey systematically tracked AI-search performance. The exact percentage should be interpreted cautiously because that CMO sample was only about 30 respondents, but the operating gap is clear: most organizations are still building the measurement layer.
- Build a prompt map. Track the questions customers ask at awareness, consideration, and decision stages - not just keywords.
- Audit the source ecosystem. Identify which publishers, communities, marketplaces, review sites, and first-party pages shape answers in your category.
- Make owned facts unambiguous. Pricing, product specifications, integrations, availability, security, policies, and comparisons should be easy for machines and humans to verify.
- Strengthen third-party evidence. AI answers often depend on sources you do not own. Reviews, expert coverage, partner pages, benchmarks, and communities matter.
- Measure answers continuously. AI systems, retrieval behavior, source preferences, and model outputs change. One manual prompt test is not a monitoring strategy.
7. The Next Front Door May Not Be a Search Box
The progression does not stop at AI answers. McKinsey's more recent work on the future of marketing describes agentic experiences as conversational, adaptive, and increasingly capable of moving from recommendation toward action. In that world, brands need to be not only visible and trusted, but increasingly machine-consumable. McKinsey
That is where AX - Agentic Experience - enters the roadmap. GEO makes the evidence available. AEO improves the chance of being selected. AX prepares the brand, product, and underlying systems for agents that can evaluate options and eventually act.
The shift is bigger than SEO versus AI search. The customer interface itself is changing: from links, to synthesized answers, to recommendations, and eventually to agents that can act on those recommendations.
The Lanesra View
The question for marketers is no longer only "Do we rank?"
It is also: "When AI explains our category, are we present, accurate, trusted, and recommended?"
That requires the foundations of SEO, the evidence discipline of GEO, and the outcome measurement of AEO. And as agentic experiences mature, it will increasingly require AX as well.