The question underneath the question
A business already paying for SEO is not asking what AI SEO is. It is asking whether the money is about to be wasted. The short answer is no, and on Google it is not an opinion.
DOCUMENTED Google’s site-owner documentation for AI Overviews and AI Mode says: “The best practices for SEO remain relevant for AI features in Google Search. There are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary.” The eligibility gate is the ordinary one: indexed, and eligible to show with a snippet. Google’s May 2025 guidance adds that “the underpinnings of what Google has long advised carries across to these new experiences”.
TESTED What has changed is where the informational load lands. Pew’s US browsing panel found 18% of Google searches in March 2025 produced an AI summary, and that users who saw one clicked a traditional result in 8% of visits against 15% without. Ahrefs estimated a 34.5% lower click-through rate for the top-ranking page on informational keywords with an AI Overview. The need to be visible did not move; the format of visibility did.
What carries over unchanged
DOCUMENTED Google’s list of fundamentals for AI features is a list of SEO fundamentals: crawling allowed in robots.txt and by any CDN or hosting layer, content findable through internal links, good page experience, important content in text form, images and video where they help, structured data that matches the visible text, and current Merchant Center and Business Profile information. Index health is a strict prerequisite, because a page that is not indexed is not a candidate.
INFERRED Technical health arguably carries more weight than it did, because retrieval agents fetch and read a page at answer time and a page whose content only exists after JavaScript runs may contribute nothing to that fetch. No platform publishes its rendering capability for these agents, so this is stated as inference.
What is genuinely new
Access is per agent
DOCUMENTED Traditional SEO managed Googlebot and Bingbot. AI search adds agents that platforms document and control separately: OAI-SearchBot and ChatGPT-User alongside GPTBot at OpenAI; PerplexityBot and Perplexity-User; Claude-SearchBot, Claude-User and ClaudeBot; and Google-Extended for Gemini training and grounding. Blocking all of them to stop scraping also removes a site from the answers its buyers read. Each is an independent decision. The crawler access guide covers checking them.
The unit is the passage
INFERRED Optimising for a ranking position and optimising for retrieval into a composed answer are different jobs. The second rewards a section that answers its question in its first sentences and stands on its own when lifted. Older content fails here more than anywhere else, because a decade of content marketing produced long preambles in front of the answer. Answer Engine Optimisation covers the passage rules.
The business must resolve as one entity
INFERRED Keyword SEO rarely required a business to settle on one canonical name, one address form and one set of service names across every public surface. A composed answer has to name something specific, so ambiguity now has a direct cost. This is the substance of entity SEO.
Prompt coverage replaces keyword lists
Keyword research maps what people type into a box. Prompt research maps what people ask an assistant, which is longer, comparative and full of constraints. “AI SEO agency Perth” and “who should I talk to about my business showing up wrong in ChatGPT” express the same intent in incompatible formats. DOCUMENTED Google documents that both its generative surfaces may fan out one question into multiple related searches, which is why coverage of the surrounding questions matters more than one page for one phrase.
Corroboration replaces link authority as the off-site job
INFERRED Link building was about authority flowing through a graph of websites. Corroboration is about independent sources agreeing on facts, because an answer composed from several retrieved sources reflects what they agree on. Links still matter for ranking; agreement is the new thing to manage.
Measurement is by sampling
INFERRED There is no rank position inside a composed answer, and the same prompt returns different answers across runs. Visibility is therefore measured by running fixed prompts repeatedly and reporting frequencies, which is the basis of this publication’s methodology. DOCUMENTED On Google, AI feature traffic is folded into the Search Console Performance report under Web search, not reported separately.
Side by side
| Area | Traditional SEO | AI SEO |
|---|---|---|
| Primary target | A ranking position | Being a source a composed answer draws on and cites |
| Unit of optimisation | The page | The passage |
| Query research | Keyword volume and difficulty | Prompt coverage across the decision journey |
| Off-site work | Links and authority | Corroboration and factual agreement |
| Identity work | Optional, mostly local SEO | Foundational for every business |
| Access | Googlebot, Bingbot | Plus each platform’s search and user agents, and Google-Extended |
| Measurement | Rank tracking, clicks, impressions | Repeated-run frequencies, citation context, referrals |
| What is guaranteed | Nothing | Nothing, and more visibly so |
Three myths
“SEO is dead.” DOCUMENTED The mechanism feeding Google’s AI features is the Search index. Google says so, and says the existing practices remain relevant. A site that is not technically sound and indexed is not a candidate for an overview.
“AI SEO is a completely new discipline.” Most of it is technical SEO and content quality done properly. The genuinely new parts are the ones above: per-agent access, passage-level structure, entity resolution, prompt coverage, corroboration and sampled measurement.
“You need a specialist AI SEO tool.” Tools help sample assistant responses at volume. None substitutes for verifying crawler access, reading the rendered HTML, or fixing a registry record.
What changes for an existing Perth programme
INFERRED In most cases the change is a re-sequencing of what is already spent, not a new line item. A programme producing keyword-led articles every month has the budget; what changes is what it buys. A practical order: verify access agent by agent and read the rendered HTML; run a brand-fact inventory across every public surface and fix disagreements in order of source authority; restructure the highest-value pages so the answer sits at the top; redirect new content toward the comparison, cost and suitability questions Perth buyers ask; and set a repeated-run baseline so there is something to compare against in six months. Nothing in that order requires stopping technical work, and cutting it removes the input that feeds the surface being targeted.
Whether Perth businesses are being named in these answers today, and by which sources, is the observation programme’s question. This page does not answer it.
What is documented and what is inferred
- Documented: Google’s statement that SEO best practices carry over; the fundamentals list; per-agent access controls; fan-out; Search Console reporting.
- Tested: the Pew and Ahrefs click figures, both US.
- Inferred: the passage as unit; entity resolution as a new cost; corroboration as the off-site job; sampled measurement; the re-sequencing advice.
- Not established: any Perth-specific frequency.