Every time the SEO industry undergoes a major shift, the same thing happens: the people who understood the old model add a new word to their pitch and keep doing what they were doing. After Google's Panda update it was "quality content." After RankBrain it was "semantic search." After BERT it was "conversational keywords." And now, after the most significant shift in how people find information since the invention of the search engine, it's "AI SEO." Most of them have no idea what that actually means.
How to spot the repackage in five minutes
You don't need to interrogate an SEO consultant for an hour to figure out whether they understand AI search. Five questions reveal it quickly. If the answers still centre on backlinks, keyword density, and domain authority as the primary levers, you're talking to someone who learned SEO between 2015 and 2020 and hasn't seriously updated their model since.
"Ask them: what's your approach to llms.txt? How do you structure schema for AI citation? What's your entity consolidation process? How do you audit for answer-engine readiness? And: can you show me a site where you've improved AI citation frequency? If they hesitate on any of the first four, the fifth question is irrelevant."
What 2018 SEO looks like in 2026
To be fair to the people still selling it: 2018 SEO still works, partially, for traditional Google search. Backlinks still matter. Page speed still matters. Keyword research still matters. These things haven't become irrelevant - they've just stopped being sufficient, and they have almost nothing to do with AI visibility.
The problem isn't that the old playbook is wrong. It's that it's being sold as "AI SEO" when it doesn't touch the signals that actually determine whether an AI engine cites you. A consultant who's optimising your anchor text profile and building out a backlink campaign is not doing AI SEO. They're doing 2018 SEO and calling it something else.
The backlink myth in the AI era
Backlinks remain a signal for Google's ranking algorithm. They are essentially irrelevant to whether an AI engine cites your content. AI engines don't evaluate your link graph when deciding whether to surface your answer to a query - they evaluate the quality, structure, specificity, and trustworthiness of your content itself.
A site with 10 high-quality, well-structured pages with comprehensive schema, an llms.txt, and direct answer-led content will be cited more frequently than a site with 500 pages and a strong backlink profile but no AEO (answer engine optimisation) foundation. Every time. The "build links to rank" model was already weakening for Google - for AI search it's close to irrelevant.
Keyword density is completely dead
Exact-match keyword density hasn't been a meaningful Google ranking signal since Panda in 2011. For AI engines it's worse than irrelevant - content that's been written to hit keyword density targets reads unnaturally, and AI engines trained on massive corpora of human writing are very good at detecting and deprioritising content that reads like it was written for a crawler rather than a person.
If the SEO report you're receiving includes a section on keyword density or exact-match keyword frequency, you are being sold something that was already outdated before most of your current team joined the company.
Domain authority as a proxy for everything
Domain authority (DA) is a Moz-invented metric that attempts to predict how well a site will rank in Google. It's a useful proxy - imperfect but directionally reasonable for traditional SEO. It has no direct relationship to AI citation frequency.
Agencies that lead with DA improvement as a primary goal are optimising for a proxy metric for a platform (Google) that is itself losing share to AI tools. It's not wrong to care about DA. It's wrong to sell it as the primary lever for "AI SEO" performance.
What real AI SEO actually looks like
The terminology is still settling. Some people call it AEO (answer engine optimisation). Some call it GEO (generative engine optimisation). Some just call it "AI visibility." The label matters less than the work - and the work is meaningfully different from traditional SEO in ways that require different skills, different tooling, and a different mental model.
Structured data mastery
A real AI SEO practitioner can build and validate comprehensive JSON-LD schema markup from scratch. Not install a plugin and hope for the best - actually write the schema, understand the types and properties, know which schema types are most relevant for which pages, and validate that it's error-free and complete. This is a technical skill that requires genuine understanding of the schema.org vocabulary and how AI engines interpret it.
If the person you're talking to can't describe the difference between a SoftwareApplication schema and a Product schema, or doesn't know what sameAs does in an Organisation schema, they are not a schema practitioner.
Content architecture for AI
Structuring content for AI citation is a different discipline from structuring content for Google rankings. It requires understanding how AI engines parse semantic hierarchy, how to write question-led headings that map to the actual queries being asked, how to lead with the answer rather than building toward it, and how to create FAQ content that will be marked up with FAQPage schema and surfaced in AI responses.
This isn't about writing longer content or more comprehensive content - it's about structuring content so that an AI engine can extract a specific, credible answer to a specific question from your page without ambiguity.
Entity and knowledge graph work
AI engines build entity models. They construct a picture of what your company is, what it does, who it serves, and how it relates to other entities in your space. Influencing that picture requires consistent, corroborated signals across every platform where your company appears - not just your website.
A practitioner doing real AI SEO will audit your entity signals across your site, your LinkedIn, your Crunchbase or CrunchBase profile, your G2 or Capterra listings, any media mentions, and any partner or directory listings. They will identify inconsistencies and systematically correct them. This work is invisible in traditional SEO reporting. It shows up in AI citation frequency over time.
Red flags to watch for when evaluating an SEO hire or agency
Here are the specific things that should make you walk away - or at minimum ask a harder follow-up question.
- "We'll build high-quality backlinks to improve your AI ranking": backlinks don't directly affect AI citation. This is traditional SEO being sold as something it isn't.
- "We target a keyword density of X% on all pages": keyword density is a 2010 concept. Anyone still using it as a metric doesn't understand how modern search works.
- "Our proprietary AI SEO tool tracks your AI rankings": AI citation tracking is genuinely hard. If they have a "proprietary tool," ask exactly what it measures and how. Vague answers mean the tool is a dashboard dressed up as insight.
- "We guarantee page 1 rankings": no legitimate SEO practitioner makes ranking guarantees. AI citation is even less predictable than traditional rankings. Anyone offering guarantees is either lying or selling something that isn't SEO.
- They can't explain what llms.txt is: this is a basic AEO concept from 2024. If they've never heard of it or can't explain what it does, they are not operating in the AI search space.
What to ask instead - and what good answers look like
The questions that actually differentiate a genuine AI SEO practitioner from someone who updated their pitch deck:
- "Can you show me your schema audit process?" Good answer: a structured walkthrough of how they identify schema gaps, which validator tools they use, and how they prioritise schema types per page. Bad answer: "we install Yoast and make sure it's configured correctly."
- "What does your llms.txt implementation look like?" Good answer: a description of the file structure, what they include, and how they tailor it per client. Bad answer: a blank stare, "what's that?", or "oh yes we do that" with no substance.
- "How do you measure AI citation frequency?" Good answer: a realistic description of manually checking AI tool responses, using Perplexity and ChatGPT queries in the target category, and tracking trends over time. Bad answer: "our platform tracks it automatically" with no explanation of methodology.
- "What's your entity consolidation process?" Good answer: a description of auditing company descriptions across platforms, identifying inconsistencies, and systematically updating them. Bad answer: any answer that involves primarily on-site work only.
The transition happening right now
The SEO industry is going through what it always goes through when the underlying platform changes: a shakeout. The practitioners who genuinely understand the new model are producing results. The ones who rebranded without updating their skills are producing reports full of metrics that feel familiar but no longer connect to what their clients actually care about - being found by buyers who are using AI to research their category.
This shakeout takes 18–36 months to fully play out. Right now, you can still hire someone who will confidently show you a backlink report and call it AI SEO. In two years, that will be much harder to get away with because the clients will know enough to call it out.
The advantage of understanding this now is that you can move faster than your competitors, hire or partner with people who actually know what they're doing, and build an AI visibility foundation while most of your category is still paying for the 2018 playbook.
Adding "AI" to an SEO pitch doesn't make it AI SEO.
Ask about schema, ask about llms.txt, ask about entity consolidation, ask about AI citation measurement. If the answers are vague or redirect to backlinks and keyword rankings, you're being sold something that won't move the needle on the metric that's actually changing: whether AI tools recommend you when a buyer in your category asks for help.
How to tell the difference. Ask anyone pitching you to show their own numbers, on their own domain, with dates. If they cannot, you are buying 2018 advice at 2026 prices. I have written up the honest version of that comparison: specialist vs agency.
A quick tell. Run their own website through the Waffle-o-Meter. An agency selling clarity whose homepage scores 70 for buzzword density has told you what you need to know. Then check their AI basics too: when I ran 80 NZ and AU agency websites through my AI Visibility Check, 36 of the 41 firms selling SEO or AI search were missing at least one basic on their own homepage.
Where this fits: This is part of AEO foundations, the work I do first for New Zealand and Australian B2B teams. The whole method, step by step, is in the AI Search Playbook.
