Win AI search for every client

Pete Everitt cuts through the noise with the most important distinction: AI is not a search engine, it is a character and word predictor that generates what it thinks the best answer looks like, rather than retrieving the best answer. That is a fundamentally different problem needing a fundamentally different solution, and the root cause of bad AI SEO advice is a categorisation error, AI search was handed to SEO practitioners and web developers because it looks like search, but it operates on different logic.

He names the two camps that dominate the discourse and dismisses both: the AI is killing SEO camp, usually selling a course, and the nothing has changed camp, usually protecting a retainer. Everything else in this playbook follows from getting this right, because if you optimise for retrieval on a system that predicts, you are solving the wrong problem entirely.

Eugene Levin argues the job of SEO has expanded to include qualitative AI monitoring: it is no longer enough to know you rank, you have to track whether AI mentions your client and whether those mentions are accurate and favourable. Reputation management and SEO have effectively merged into one discipline, because brand visibility across the whole web ecosystem, third-party sites, reviews, Reddit, YouTube, LinkedIn, now directly influences AI recommendation outcomes.

This is a monitoring service you can offer today that most agencies are not even measuring, which makes it an easy wedge into a new retainer. What machines say about your client is now part of your remit, and it changes based on signals far outside their own website. Start by auditing the current AI answer for each client's key queries, because you cannot improve what you have not looked at.

Levin points out that the shift from clicks to citations breaks attribution, so agentic search is closer to brand marketing than performance marketing, and you should frame it that way to clients and their CFOs from the outset. Set the wrong expectation and a client will judge a brand play by performance metrics it was never meant to hit, then churn when the dashboard looks unfamiliar.

Getting the framing right protects the engagement and the fee. This is also why honest expectation management, which Levin calls a defining trait of thriving agencies, matters as much here as the tactics. Sell measurable brand authority, not trackable clicks, and the relationship survives the transition.

Levin stresses that agentic search optimisation, also called GEO, is a genuinely new service category that requires a market-education approach, not just demand capture, because clients do not yet know to ask for it. Agencies that invest in educating the market now will be positioned as the leaders when AI search becomes the norm, exactly as the earliest SEO educators became the trusted names of the last era.

That means content, talks, and audits that teach clients why this matters, not just a line on a proposal. Market education is slow, but it compounds into authority and inbound demand that price-shoppers never touch. Being the agency that explained the shift is worth more than being the tenth to offer the service.

Shawn Davis's research shows 68% of local searches now surface AI overviews, so the majority of local queries are already shaped by AI-generated content. Discovery has moved from a rankings model, a list of links, to a recommendations model where AI selects and presents a shortlist with reasons, which changes what a business needs to be visible for.

Local businesses are disproportionately exposed because they rely on last-minute, intent-driven searches, exactly the queries where AI overviews now appear most consistently. Davis is pointed that most existing advice is either too vague or too enterprise-biased, hire a PR agency is useless to a plumber or a coffee shop. Understanding this model is what lets you explain to a client why their old ranking report no longer tells the whole story.

Davis found the business case in the data: sites that received more AI crawler visits showed 3. 2 times more human traffic, 2.

7 times more form submissions, and 2. 5 times more click-to-call events than the median.

Yet most SMBs have completed fewer than two AI search optimisations, so the opportunity is enormous and largely untapped, and the tactics that work do not require enterprise budgets. The traditional fundamentals, NAP consistency and a strong Google Business Profile, are necessary but no longer sufficient for AI-era visibility, they are table stakes on top of which the new work sits.

Getting crawled well by AI is now a measurable driver of real business outcomes, not a vanity metric. This is the number you take into every sales conversation.

Everitt is clear that building a schema graph is the most important technical foundation, and it is not a checkbox exercise: it must be applied consistently across an entire site, not just key pages, and maintained over time. He recommends the EEAT to AEO to GEO progression as the most reliable framework available, EEAT builds credibility, AEO proves your content structure is machine-readable, and GEO gets you into the language model's training data itself.

This is the unglamorous structural work that actually moves the needle, and it is exactly what the snake-oil sellers skip because it is real work rather than a quick trick. Applied site-wide and maintained, a schema graph is what makes a site legible to machines that predict rather than retrieve. Start with your own site before you sell it, so you understand the true scope.

Everitt argues author pages need to become a standard part of website builds, not a retrofit: proper author schema, linked bylines, and a feed of attributed content keep author entities active and credible to both AI models and Google. This is how you signal genuine expertise in a form machines can parse, and it compounds over time as an author's body of attributed work grows.

Retrofitting it later is far harder than building it in from the start, so bake it into your standard site template now. Author identity is the connective tissue between the human expertise you produce and the machine credibility that gets it cited. An anonymous site has no expert to trust; a well-attributed one does.

A striking finding from Everitt is that human-authored, expert-driven content still outperforms AI-generated content in LLM search outputs, the models appear to detect their own generated content, and real human expertise with concrete examples carries more authority. Davis reinforces that content quality and relevance is one of the largest scaling factors for AI visibility, and that framing content around outcomes for users rather than self-promotion is the right direction.

So the instinct to mass-produce AI content to win AI search is backwards and actively counterproductive. Depth, expertise, and genuine usefulness are what get a brand cited, which is good news because it is defensible, a competitor cannot copy real expertise the way they can copy a keyword list. Invest in expert content as an asset, not a commodity to churn out.

Levin notes that website architecture now has to serve two distinct audiences simultaneously, human users and AI agents, and technical SEO audits must account for how AI crawlers navigate and interpret site structure. That means your audit checklist grows: it is no longer only about human usability and Googlebot, but about whether an AI crawler can cleanly parse and understand the site.

Treating the AI agent as a first-class visitor is a concrete, sellable upgrade to a service you may already offer. The sites that win recommendations are the ones built to be understood by the systems doing the recommending. Fold agent-readability into every technical audit as standard.

Krys Lambiase's data reframes acquisition: 85% of SMBs find an agency through a referral, and 90% of web pages get zero organic traffic, so being consistently present and helpful in communities is what makes your name surface when the moment comes. There is a second payoff specific to AI search, being present on Reddit and YouTube means your answers get scraped by AI and surface in AI results, making you a cited source with no additional effort.

Use the free site audit as a collaborative exploration rather than a diagnostic indictment, it builds trust without a hard sell and creates the natural opening for a roadmap conversation, and lean on before-and-after case studies because prospects look for themselves in the before picture. Nurturing leads early generates 50% more sales-ready leads and shortens the eventual sales cycle. Presence compounds into both referrals and citations.

Everitt warns that the big players control the rules and change them without much notice, Google's removal of the num=100 parameter in September instantly erased context for anything outside position 10, so this is a service that requires ongoing attention, not a one-off fix. That volatility is precisely why it should be a retainer.

Sam Sarsten offers the reassuring bridge: local SEO skills transfer directly to AI search optimisation, because the signals that rank a business in the map pack, prominence, relevance, proximity, reviews, and authority, are the same ones AI tools surface, so you can productise this as an extension of services you may already offer. Package it as monitoring, content, and schema maintenance on a monthly basis, and the agencies that win AI search will be the ones who turned it into a recurring line while everyone else was still debating it.