Does Keyword Strategy Affect Visibility in AI Search Results?
Yes, keyword strategy still affects visibility in AI search results, although the way it works has changed. Keywords now decide which questions a page answers, which page on a site owns each topic, and whether the page is relevant enough for a search system to retrieve it, once it is indexed and eligible. They no longer need to be repeated in exact-match form. Google’s guide to generative AI features on Search, last updated in July 2026, says its AI systems understand synonyms and general meaning, so “you don’t have to worry that you don’t have enough ‘long-tail’ keywords or haven’t captured every variation of how someone might seek content like yours.”
That sentence is the short version of the whole topic. Keyword research remains the way a business learns what its customers ask and in what words, while chasing every phrasing with its own paragraph, heading or page has stopped paying off. The rest of this article explains why, what keyword work still decides, the mistakes that make it backfire, and how to check whether a site actually appears in AI answers.

What Google says about keywords and AI search
Google’s position is that SEO still applies to AI Overviews and AI Mode because those features are rooted in its core Search ranking and quality systems. The guide describes retrieval-augmented generation, also called grounding, in which the AI feature uses the core ranking systems to retrieve relevant pages from the Search index and then writes an answer that links to them. A page has to be indexed and eligible to show with a snippet before it can appear as a supporting link, and Google’s separate page on AI features adds that there are no additional technical requirements and no special optimizations needed.
This is why keyword strategy has not become irrelevant. If retrieval runs through ordinary Search ranking, then a page has to be relevant to the query, or to the related queries the system generates, before it can be pulled into an answer. What has changed is the meaning of “relevant.” Google says its systems can understand the relevance of a page “even when there is no exact match between the query and the page’s primary content,” which describes a ranking system that reads the topic of a page instead of counting phrases.
Google also addresses the labels people put on this work. It says AEO and GEO are terms used for work focused on visibility in AI search, and that from Google Search’s perspective optimizing for generative AI search “is optimizing for the search experience, and thus still SEO.” Readers who want the terms side by side can see how AEO, GEO and SEO relate in a separate comparison.

How query fan-out changes keyword research
Google documents a technique called query fan-out, in which the model issues several related searches at the same time to gather more material for its answer. Its own example is a question about fixing a lawn full of weeds, which might fan out into searches about herbicides, removing weeds without chemicals and preventing weeds. A page can therefore be retrieved for a sub-question the person never typed, as long as the page answers that sub-question well.
For keyword research, this shifts the unit of work from the phrase to the question set. Take an illustrative example: a Calgary contractor targeting “basement development cost Calgary.” The fan-out behind a question like that could plausibly include whether a permit is needed, how long the work takes, what drives the price up and whether a legal secondary suite changes the scope. A keyword list built only around the head phrase and its variants would miss most of those questions, while a page that answers them clearly, with figures the contractor can stand behind, gives both ordinary Search and AI features something to use.

The raw material for keyword research has not changed: Search Console queries, People Also Ask boxes, related searches, keyword tools and the questions customers ask on the phone. The difference is what happens to the list afterwards: phrases are grouped by the question they represent, each group is assigned to one page, and the sub-questions become sections of that page instead of new pages. That approach is also consistent with Google’s guidance against publishing separate content for every variation of a query mainly to manipulate rankings.
What keyword research still decides
A keyword strategy still does three jobs that AI search depends on. It shows which topics have real demand, so effort goes to pages people look for. It decides which page owns which topic, so a site does not split one subject across several thin pages that compete with each other. And it captures the customer’s own vocabulary, which matters for the reader even when the ranking system would understand a synonym.

The vocabulary point has a Canadian twist. A Calgary homeowner may search for eavestrough repair while a national competitor’s page says gutter repair, and someone else may search furnace repair where a page says heating repair. Google says its systems can understand synonyms and general meaning, so a business usually does not need a second page for the second word, although a broader term such as heating repair can cover more than furnaces and is worth checking against the actual results first. The page should still use the word its own customers use, because that is the word a reader expects to see when they arrive and the word that makes the page feel local.
| Keyword task | Still matters for AI search | Matters less now |
|---|---|---|
| Finding topics with demand | Yes, it decides which pages are worth writing | |
| Mapping one topic to one page | Yes, it prevents thin, competing pages | |
| Answering the sub-questions behind a query | Yes, this is what fan-out retrieves | |
| Using the customer’s own words | Yes, for the reader and for clarity | |
| Exact-match phrase repetition | Google’s systems understand synonyms and meaning | |
| A separate page for each phrasing | Can breach the scaled content abuse policy | |
| Covering every long-tail variant | Google says this is not needed |

Mistakes that make keyword work backfire
The most expensive mistake is turning fan-out queries into pages. It is tempting to take every related question a tool reports and publish a page for each one, but Google’s guide warns that creating separate content for every variation, mainly to manipulate rankings or AI responses, violates its scaled content abuse policy. It also calls the approach ineffective, because a high number of pages does not make a site more relevant or more useful.

Several other tactics sold as AI search optimization are ones Google says it ignores or does not need. An llms.txt file or other AI text file is not used by Google Search, although it will neither help nor harm a site there. Breaking content into tiny chunks is not required, because Google’s systems can find the relevant part of a longer page, and Google says there is no ideal page length. Structured data is not required for AI features and there is no special schema to add, although it remains useful for rich results. Chasing brand mentions also has limits: Google says AI features can show what is said about a business across the web, but that seeking inauthentic mentions is not as helpful as it might seem, because its systems focus on quality and block spam.

What does help is the work that has always helped. Google’s guide puts the weight on unique, non-commodity content: a first-hand view, practical experience and detail that a summary of other pages could not supply. Clear headings and well-organised paragraphs help readers, which is why Google mentions them, and a page that is crawlable, indexed and works well on a phone covers the technical basics. The practical side of showing up in AI Overviews is covered in more detail in a separate guide.
ChatGPT, Perplexity and other assistants
Google’s guide covers Google Search only. Claims about how ChatGPT, Perplexity or Claude choose their sources should be treated as theories unless the provider documents them, and the providers say far more about access than about ranking. OpenAI, for example, explains that OAI-SearchBot is the crawler used to surface websites in ChatGPT’s search features, while GPTBot is a separate crawler for model training, and that each can be allowed or blocked independently in robots.txt. It adds that sites opted out of OAI-SearchBot will not be shown in ChatGPT search answers, although they can still appear as navigational links.

That makes robots.txt the first thing to check before any keyword work for these assistants. A site that blocked every AI crawler to keep its content out of training data may also have removed itself from ChatGPT’s search answers, and no amount of keyword research fixes that. Beyond access, the providers publish little about how they choose sources, so the reasonable working assumption, and it is only an assumption, is that the same fundamentals help: a clear page that answers the question in plain language, on a site that is consistent about what the business does and where it works.
How to check AI search visibility
For Google, the most reliable source is the Generative AI performance report in Search Console, which Google rolled out to all websites worldwide as of August 31, 2026. It shows impressions from AI Overviews and AI Mode grouped by page, country, date and device. It has no query dimension, so Google does not tell a site which keywords triggered its appearances in AI answers. A business can still see which pages are being used, and a Calgary business can filter the country dimension to Canada to separate local demand from impressions abroad.

Google’s page on AI features also notes that appearances in AI Overviews and AI Mode are counted inside the ordinary Performance report under the Web search type. That is where keyword-level data still lives, mixed in with classic results. Reading the two reports together gives useful context, such as which pages AI features draw on and which queries those pages rank for, but neither report links a particular AI appearance to a particular keyword.

For assistants outside Google, there is no equivalent official report for most of them, so the practical method is a dated prompt log. As a suggested starting point, pick ten to twenty questions customers actually ask, run them in the assistants that matter to the business on a fixed schedule, and record whether the business or its pages appear and what the answer says. Answers vary between runs and between users, so the log shows a direction over months rather than a precise ranking. Google also warns that no third-party tool has access to its internal ranking or AI systems, which is worth remembering when a tool promises AI ranking scores.
A sensible order of work
The order that makes sense for most businesses starts with access and ends with refinement. Check that robots.txt and any security or CDN settings allow Googlebot and the search crawlers of the assistants that matter, and that key pages are indexed. In Search Console, also confirm the site is included in Search generative AI features, because Google says a site must be included there to be eligible to appear in its generative AI features. Then review the keyword map: one page per topic, each page answering the sub-questions customers ask, in their words. Only after that is it worth rewriting individual pages for depth and first-hand detail, which is the slowest work and the most valuable.

None of this produces quick results. SEO takes months, and Google says indexing and serving are never guaranteed even when a page meets every requirement. A business that needs enquiries while organic and AI visibility build can run Google Ads on its most valuable searches in the meantime, keeping the paid and organic reports separate so each can be judged on its own.
SEO Company To-The-TOP! is a Calgary practice that has worked on keyword research and search visibility since 2007, and the same person who reviews a site does the work. If you would like a second opinion on how your keyword plan holds up for AI search, call (403) 308-5949.
Contact Calgary SEO Company To-The-TOP!
Questions about anything in this article, or about your own rankings? Talk to a Calgary SEO specialist directly.
Phone: (403) 308-5949
Address: 1509 14 Ave SW, Calgary, AB T3C 0W4
Hours:
Monday to Friday: 10:00 am – 7:00 pm
Saturday: 12:00 pm – 4:00 pm
Sunday: closed
