AI Search Strategy: The Four Pillars to Plan Around

Most business owners I speak to treat AI search as a side project. Someone on the team checks ChatGPT once a month, takes a screenshot, and if the brand shows up, everyone relaxes. If it doesn’t, there is a short panic and then nothing changes. The common belief is that AI search is something you “optimize for” with a few tweaks, like adding a new tag to a page.

That belief is the reason most plans go nowhere. AI answers are built from many sources, they change from one day to the next, and a large part of what they say about you comes from websites you don’t own. You can’t fix that with a tweak. You need a strategy, with goals, owners and a budget, just like you have for SEO or paid ads.

In this guide I will give you a simple planning framework built on four pillars: content AI can find and quote, mentions on other websites, technical access for AI crawlers, and measurement. I lean on a recent Ahrefs piece on AI search strategy, Google’s and OpenAI’s own documentation, and several large studies. Then I show you how to set goals, split a budget and review the plan every quarter.

Why AI search deserves its own line in your marketing plan

Here is the short version. AI search is not a separate internet, but it behaves differently enough from a classic results page that it needs its own targets and its own reporting.

Google itself says there is no special trick for its AI features. In its documentation on AI features, it states plainly:

“There are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary.”

— Google Search Central, AI features and your website

So why a separate strategy line? Because three things are different:

• The answer is volatile. Ahrefs tracked more than 43,000 keywords and found that AI Overviews have a 70% chance of changing between consecutive observations, and only about 54.5% of cited URLs carry over from one response to the next (November 2025). A single screenshot tells you almost nothing.
• Other websites speak for you. In the Ahrefs strategy article, Mateusz Makosiewicz reports that up to 89% of Ahrefs’ own appearances in AI answers came from other sites mentioning the brand, not from ahrefs.com (Brand Radar data, July 2026).
• The traffic maths is different. AI answers often satisfy the question without a click. That doesn’t make them worthless. It means you measure mentions, citations and the quality of the clicks you do get, not just raw sessions.

In brief: your SEO work is the foundation, but you need a plan that also covers your reputation on the wider web and a way to measure something that moves every day.

The four pillars at a glance

Before we go pillar by pillar, here is the whole framework on one page. I suggest you print this table and use it as the agenda for your planning meeting.

Pillar	The question it answers	Main owner	Typical outputs
1. Findable, quotable content	When an AI system looks for an answer, is there a page of ours worth using?	Content / SEO lead	Product facts pages, original data, comparison pages, refreshed guides
2. Off-site mentions	What do other trusted websites say about us, and do they put us in the right category?	PR / partnerships	Reviews, list inclusions, podcast and video mentions, community answers
3. Technical access	Can AI crawlers and agents actually reach and read our pages?	Developer / technical SEO	robots.txt rules, CDN bot settings, fast server responses, log checks
4. Measurement	Are we becoming more visible on average, and does it bring business?	Analyst / marketing manager	Prompt tracking, share of voice, AI referral reports, quarterly review

None of these is new on its own. What is new is treating them as one plan with shared goals, rather than four teams doing their own thing.

Pillar 1: pages AI systems can find, trust and quote

When ChatGPT, Gemini or Google’s AI Mode answers a question, it usually runs searches in the background and pulls short passages from the pages it finds. Google calls this query fan-out, which it describes as “issuing multiple related searches across subtopics and data sources.” If you want the mechanics, I explain the whole pipeline in my guide on how retrieval-augmented generation works, and the fan-out step in detail in ChatGPT query fan-out.

For strategy purposes, you only need to remember one thing: AI systems quote pages that already rank and that contain something worth quoting. So pillar 1 has three parts.

Your “source of truth” pages

AI answers often repeat outdated facts: old prices, a product you retired, a service area you no longer cover. The Ahrefs article makes the point that some pages are valuable even if they never rank, because they give AI systems a dependable reference. Think of a clean pricing explanation, a features page, an “about” page with founding year and locations, and a page that lists integrations or service areas.

Keep those facts the same everywhere: your website, your Google Business Profile, LinkedIn, review sites and directories. Our guide to fixing wrong information about your brand in AI answers walks through that clean-up.

Content that survives being summarized

Google’s own guide to generative AI features, updated in July 2026, calls non-commodity content the most influential factor over the long run. In plain words: if your page only repeats what every other page says, the AI can summarize the topic without you. If your page has original data, a real test, a calculator or a first-hand opinion, the AI has a reason to cite you.

Ahrefs puts it neatly when talking about free tools:

“AI can describe a backlink checker, but it can’t be one.”

— Mateusz Makosiewicz, Ahrefs, AI Search Strategy

The same goes for a mortgage calculator, a pricing estimator or a survey of your own customers.

Freshness that is real, not cosmetic

Ahrefs analyzed 17 million AI citations and found that AI-cited content is 25.7% fresher than organic Google results (December 2025), with ChatGPT showing the strongest preference for newer pages. That is a reason to plan real updates into your calendar. It is not a reason to change the year in your titles and call it a refresh, which is exactly the habit Google appears to be cracking down on (more on that in my post about self-ranking “best” lists).

Here is what I suggest you put in the pillar 1 plan for a quarter:

• Audit and fix 5–10 “source of truth” pages (pricing, features, locations, about, policies).
• Publish one piece of original research or one useful free tool.
• Refresh your top 10 informational pages with new facts, not new dates.
• Write answer-first introductions and clear headings, which help readers and AI alike.

Pillar 2: what other websites say about you

This is the pillar most businesses skip, and it is often the biggest one. When an AI tool decides which brands belong in an answer like “best accounting software for a bakery,” it leans on reviews, comparison articles, forums, videos and news coverage.

The numbers back this up. In an Ahrefs study of about 75,000 brands, YouTube mentions had a correlation of about 0.737 with AI brand mentions, and branded web mentions scored between 0.656 and 0.709 depending on the platform. Domain Rating, the classic link-strength metric, scored only 0.266–0.326 (December 2025). As the authors themselves say, “correlation isn’t causation,” so I treat this as direction, not hard science. But the direction is clear: being talked about matters more than being big.

Mentions that place you in the right category

A mention is most useful when it ties your brand to a category, a problem and a type of customer. “Acme is a good company” helps a little. “For small dental clinics, Acme is the scheduling tool most people here use” helps much more, because it matches the way people ask AI tools for recommendations.

Practical sources of these mentions include:

1. Review platforms your buyers actually check (Google, Trustpilot, G2, Capterra, industry directories).
2. Third-party “best of” and comparison articles. See our guide on how to get on best-of lists.
3. YouTube videos and podcasts where you appear as a guest or get reviewed.
4. Community threads on Reddit and niche forums. I cover the Google side of this in showing up in Reddit threads that rank on Google.
5. Digital PR: data stories and expert comments in trade and local press.

Mentions without links still count

Many owners still judge PR only by the backlink. For AI visibility, an unlinked mention in a respected article can still shape the answer. Our post on AI mentions without links explains why, and how to find them.

Pillar 3: letting AI crawlers and agents through the door

You can have the best content in your industry and still be invisible if AI systems can’t reach it. This happens more often than you would think, usually because a security plugin or CDN setting blocks bots by default.

The key is knowing that each AI company runs different bots for different jobs. Blocking one doesn’t block the others, and some don’t follow robots.txt at all because a human triggered the visit.

User agent	Company	What it does	Follows robots.txt?
OAI-SearchBot	OpenAI	Finds pages to show in ChatGPT search answers	Yes
GPTBot	OpenAI	Collects content that may be used to train models	Yes
ChatGPT-User	OpenAI	Visits a page when a user’s question needs it	May not apply (user-initiated)
PerplexityBot	Perplexity	Surfaces and links sites in Perplexity results	Yes
Perplexity-User	Perplexity	Fetches a page for a specific user question	Generally no
Googlebot / Google-Extended	Google	Googlebot crawls for Search (including AI features); Google-Extended controls some use in other Google AI systems	Yes

Sources: OpenAI crawler documentation, Perplexity bots guide and Google’s AI features page (all accessed October 2026).

The strategic decision here is simple: do you want to appear in AI search answers? For almost every business that sells something, the answer is yes. In that case, allow the search bots (OAI-SearchBot, PerplexityBot and Googlebot) even if you choose to block training bots like GPTBot. OpenAI is explicit that blocking OAI-SearchBot keeps your site out of ChatGPT search answers.

Speed matters too. AI fetchers are impatient. The Ahrefs RAG article reports that one analyst observed ChatGPT using roughly a two-second timeout when fetching pages, and that Ahrefs saw many of its own pages return errors to ChatGPT-User in one month. Treat that as a warning sign rather than a rule, and keep server response times low. Our page speed optimization service exists for exactly this kind of problem.

Pillar 4: measuring visibility as an average, not a screenshot

Because AI answers change so often, the only honest way to measure them is across many prompts, several platforms and a period of time. One answer on one day is noise. Fifty prompts tracked weekly for three months is a trend.

The metrics I suggest you report each month:

• AI share of voice: the share of tracked answers that mention you, compared with named competitors.
• Mentions vs citations: how often you are named, and how often your pages are linked as a source. They are not the same thing.
• AI referral traffic and conversions: visits from ChatGPT, Perplexity, Gemini and Copilot, and what those visitors do. GA4 added a default “AI Assistant” channel in May 2026, which makes this easier; our guide to tracking AI traffic in GA4 covers the setup.
• Google AI features: Google’s guide recommends the Generative AI performance report in Search Console. Remember that AI Overview and AI Mode clicks are also counted inside the normal “Web” search type.
• Accuracy: are the facts in AI answers about you correct? Track errors like a support ticket queue.
• Bot activity: are AI crawlers visiting your key pages, and getting a 200 response?

You will need a tool for prompt tracking. Our comparison of AI visibility tools can help you choose. Be careful with any vendor that claims special access to Google’s systems; Google’s own guide warns against exactly that.

Turning the four pillars into goals you can actually hit

A strategy without numbers is a wish list. Here is how I would set goals per pillar for a quarter. The figures in the third column are an example for a hypothetical mid-sized service business, not benchmarks. Set your own based on your starting point.

Pillar	Goal type	Example quarterly target	How you check it
1. Content	Output + quality	8 source-of-truth pages fixed, 1 original study published, 10 guides refreshed	Content tracker; citations of those URLs in your prompt tool
2. Mentions	Earned coverage	40 new reviews, 3 third-party list inclusions, 2 podcast or video appearances	Review platforms, mention alerts, PR log
3. Technical	Access + speed	0 blocked AI search bots, key pages respond in under 1 second	robots.txt check, server logs, speed tests
4. Measurement	Visibility + business	Share of voice up from 12% to 18% on 50 tracked prompts; AI referral conversions tracked	Prompt tracker, GA4, Search Console

A worked example: setting the share of voice target

Let me show the maths with a made-up example. Say you are a regional HVAC company. You track 50 prompts (such as “best heat pump installer in Leeds” or “is a heat pump worth it for an old house”) across ChatGPT and Google AI Mode, once a week. That gives you 100 answers a week, or about 1,300 answers in a quarter.

In the first month, you appear in 156 of 1,300 answers, so your share is 12%. Your main competitor appears in 390, which is 30%. Aiming to match them in one quarter is unrealistic. Aiming for 18% (about 234 answers) is ambitious but possible if you win a few reviews, two local press mentions and a well-cited guide. That is a 50% relative increase, which is a good target to put in front of the team.

Notice that the target is about the average across many answers, not about winning one specific prompt. That keeps the team focused on lasting work instead of chasing a single screenshot.

Who owns each pillar, and how to split the budget

The fastest way to kill an AI search strategy is to give it to “everyone.” Each pillar needs one named owner, even in a small business where that person wears three hats.

For budget, I suggest you think in percentages of the total AI search effort rather than fixed sums, because the right amount depends on your market. Here is an example split I would start with, then adjust after the first review:

• Content (pillar 1): around 35–40%. Writing, original research, refreshes and design.
• Mentions (pillar 2): around 35–40%. PR, review programmes, partnerships, video.
• Technical (pillar 3): around 10%. Mostly a one-off fix, then light monitoring.
• Measurement (pillar 4): around 10–15%. Tools and reporting time.

If you already run strong SEO, you can shift money from content towards mentions, because your pages are likely already findable. If your brand is new, the reverse can make sense: you need pages that explain who you are before anyone will mention you. When you want an outside view of costs you can request a custom quote.

A quarterly review routine for your AI search strategy

AI platforms change their behavior often, so I suggest a proper review once a quarter, with a light monthly check in between. Here is the routine I recommend:

1. Pull the averages. Share of voice, mentions, citations and AI referral conversions for the quarter, compared with the previous one.
2. Check accuracy. Read 20 answers about your brand. List every wrong fact and trace it to a source page.
3. Review the prompt list. Drop prompts nobody asks any more and add new ones from sales calls, support tickets and Search Console queries.
4. Check access. Confirm robots.txt and CDN settings haven’t changed, and look at AI bot hits in your logs.
5. Score each pillar. Did we hit the target? If not, was it effort, budget or a wrong assumption?
6. Move money. Shift 10–20% of budget towards the pillar that showed the strongest link to results.
7. Set next quarter’s goals and write them down with owners and dates.

Two things to watch for in the review. First, don’t over-read a single platform. Perplexity, ChatGPT and Google behave differently, and a gain on one can hide a loss on another. Second, keep a note of platform changes. A drop that lines up with a known product change is a different problem from a drop that lines up with a competitor’s PR campaign.

If you want the tactical, week-by-week version of all this, our 90-day AI search plan turns the four pillars into a sequence of tasks, and our guide to generative engine optimization covers the page-level how-to.

Questions business owners ask about planning for AI search

Do I need a separate team for AI search?

No. Most of the work sits with people you already have: content, PR and whoever looks after the website. What you need is shared goals, one owner per pillar and a single report.

Should I block AI crawlers to protect my content?

Only block what you truly don’t want. If you want to appear in AI answers, allow the search bots like OAI-SearchBot and PerplexityBot. You can still block training bots such as GPTBot if you prefer.

How long before an AI search strategy shows results?

Technical fixes can show up within weeks. Mentions and reviews usually take a few months to shift the averages. Plan to judge the strategy after two full quarters, not two weeks.

Is AI search going to replace my Google traffic?

Not fully. Ahrefs itself says AI search probably won’t replace all the organic traffic websites are losing. That is why the plan should value mentions and lead quality, not only clicks.

Do llms.txt files or special markup help?

Google says its Search ignores llms.txt and that you don’t need special markup or rewriting for AI. Spend that time on content and mentions instead.

Where to start with your AI search plan this quarter

To bring it together: AI search is not a trick, it is a mix of four things you can plan. Pages worth quoting, mentions on sites that matter, open doors for AI crawlers and honest, averaged measurement. Give each one an owner, a target and a share of the budget, and review it every quarter.

If you only do one thing this week, list your 50 most important customer questions and check how often you show up across ChatGPT and Google AI Mode. That number is your baseline.

If you would like help building the plan, we can do it with you. At My SEO Experts we audit how AI tools see your brand today, fix the technical basics and build the content and mentions that move the averages. Take a look at our SEO services, or get a free quote and tell us about your business.

Further reading on ai search strategy

• Retrieval-Augmented Generation (RAG) Explained for Marketers
• The Websites ChatGPT Cites Most (and What That Means for You)
• Ranking Yourself #1 in Your Own 'Best' Lists: Does It Work in AI Search?

Sources for this AI search strategy guide

1. Ahrefs – AI Search Strategy (Mateusz Makosiewicz, July 24, 2026)
2. Google Search Central – AI features and your website (updated December 10, 2025)
3. Google Search Central – Guide to optimizing for generative AI features (updated July 10, 2026)
4. Ahrefs – How often AI Overviews change (November 11, 2025)
5. Ahrefs – Fresh content and AI citations (December 22, 2025)
6. Ahrefs – AI brand visibility correlations (75K brands) (December 12, 2025)
7. OpenAI – Overview of OpenAI crawlers (accessed October 2026)
8. Perplexity – Perplexity crawlers (accessed October 2026)