Short answer
An AI content strategy plans pages around the questions behind a prompt, where a keyword plan targets one search at a time. You collect the prompts your buyers use and the searches assistants run behind them, group them by intent, and give each intent one strong page. AI tools speed up research and structure; the facts, experience and judgement come from you, because Google treats mass-produced pages with no added value as spam.
Key takeaways
- Plan around prompts and the searches behind them, grouped by intent, with one strong page per intent.
- Google’s AI features run related searches behind each question, so the plan has to cover those searches as well as the one that was typed.
- AI in the workflow is normal: in Ahrefs’ April 2025 sample, 74.2% of new pages contained AI content, and in its ranking study AI share showed no correlation with position.
- Google calls generative AI useful for research and structure, and warns that many pages with no added value can break its scaled content abuse policy.
- The advantage is what AI can’t produce: first-hand experience, your own data and your judgement.
What an AI content strategy is
The term covers two things: planning content so AI assistants find and cite it, and using AI tools to make that content. A good strategy decides both, and the first decision shapes the second.
A keyword plan asks which term a page should rank for. An AI content strategy asks which questions a buyer puts to an assistant, which searches the assistant runs to answer them, and which of your pages should supply the answer. It is the planning side of what is answer engine optimization.
Start with the questions behind a prompt
Collect the prompts first: what buyers ask in sales calls and support tickets, what your Search Console queries show, and what you track in the assistants. Then look behind them. Google’s AI features run query fan-out, a set of related searches generated to fetch more results for one question [1], which I explain in how query fan-out works.
Each question then has a place in the plan. Here is how one prompt breaks down for a CRM built for small agencies.
| Layer | Example | Where it goes |
|---|---|---|
| The prompt | Best CRM for a small agency | The main question of a page |
| A search behind it | CRM pricing for small teams | A section on that page |
| A follow-up | Does it connect to Slack? | The same page’s FAQ |
| A separate intent | Moving from spreadsheets to a CRM | A page of its own |
How to plan the pages, step by step
- Build the question set: main keywords for the core intent, satellite keywords for the supporting context, and the prompts behind both.
- Map your existing pages and your top 20 competitors against the set, to find what is missing, thin or doubled.
- Give each main keyword one page, with its satellites as sections. Google warns against separate content for every variation of a search, fan-out queries included, when it is done primarily to manipulate results [1].
- Put each page in a tier, which sets its depth, its internal links and its on-page structure.
- Write a brief for every page: the direct answer first, the sub-questions, the entities to explain and the sources to cite.
- Write for people, with clear sections and headings. Google says there is no requirement to break content into tiny pieces for AI [1].
- Add what only you know before anything is published.
- Publish at a steady pace and refresh what has aged, tracking both.
Step 3 keeps the plan honest, and it is how a subject gets covered in depth, which I set out in topical authority for AI search. The keyword research and content planning page shows how I run this for clients.
Where AI belongs in the workflow
Using AI is normal now. In Ahrefs’ study of 900,000 pages created in April 2025, 74.2% contained AI-generated content, measured by its own detector [4]. In its study of 600,000 ranking pages, 86.5% of the top 20 results contained some AI content, and the correlation between AI share and position was 0.011, effectively zero [3].
Google draws the line at value. It says generative AI can be particularly useful for researching a topic and adding structure to original content, and that generating many pages without adding value for users may violate its scaled content abuse policy [2]. It also lists extensive automation to produce content on many topics as a warning sign [5]. And it reminds you that generative models predict a likely sequence of words instead of retrieving facts, so every AI-generated page needs a manual fact-check [2].
My approach is hybrid. The tools I build, used by 100+ SEO experts, handle research, structure and first drafts, and have generated 140,000+ articles. People choose the inputs, vet the data, pick the sources, rewrite in a human voice and sign off every page.
What AI can’t write for you
Google says content people find unique, compelling and useful will likely influence your presence in generative AI search more than any of its other suggestions [1]. Its example is telling: “7 Tips for First-Time Homebuyers” is commodity content, while “Why We Waived the Inspection & Saved Money: A Look Inside the Sewer Line” brings an experienced take that goes beyond common knowledge [1].
So every brief needs an answer to one question: what can we say here that a model could not? Usually it is one of these.
- Your own data: what your customers do, what your tests showed, what your support team hears.
- First-hand experience: what happened when you used the product, ran the process or made the mistake.
- A position: what you recommend against, and why.
- Named people: the expert behind the page, with a profile that shows why they know.
How to tell whether the plan works
Measure the plan by subject. Track how often the assistants name and cite you across a fixed set of prompts, how many of the searches behind those prompts your pages rank for, and what Search Console reports for your pages in AI features [1]. I set out the method in how to measure AI search visibility.
Then read the gaps as content decisions: a prompt where a competitor is named and you are left out is the next brief to write.
A worked example: a CRM for small agencies
The company and its numbers here are illustrative. A CRM company has 60 blog posts, each aimed at one keyword, eight of them on some version of “best CRM”. Across 20 tracked prompts, assistants name it in one.
The team collects 70 prompts and searches, which group into 10 intents. It merges the eight posts into one page, writes a brief for each intent, and uses AI for research, outlines and first drafts. Its head of support adds what no model has: the five problems agencies hit most when they leave spreadsheets, from 200 migrations. Two quarters later, assistants name it in 6 of the 20 prompts, and the gaps that remain set the next briefs.
Frequently asked questions
What is an AI content strategy?
It is a plan for content that AI assistants find and cite, and for how AI tools help make it. It starts from the questions buyers ask and the searches assistants run behind them, and gives each intent one strong page.
Does AI-written content hurt rankings?
Ahrefs found no meaningful link: across 600,000 pages, the correlation between AI share and position was 0.011. Google’s line is value: generating many pages without adding value for users may violate its scaled content abuse policy, and every AI-generated page needs a fact-check.
How is it different from a keyword strategy?
A keyword strategy gives each term a page. An AI content strategy starts from prompts, adds the searches behind them, groups everything by intent, and answers the supporting questions inside one strong page.
Should I write a separate page for every fan-out query?
No. Group them by intent and answer most of them as sections of one page. Google warns that separate content for every variation, done primarily to manipulate results, violates its scaled content abuse policy.
Do I need to break content into chunks for AI?
No. Google says there is no requirement to break content into tiny pieces for AI, and that its systems understand several topics on one page. Write clear sections with headings for your readers.
Sources
- Google Search Central, “Google’s guide to optimizing for generative AI features on Google Search”, last updated 10 July 2026. developers.google.com
- Google Search Central, “Google Search’s guidance on using generative AI content on your website”, last updated 1 October 2026. developers.google.com
- Ahrefs, “AI-Generated Content Does Not Hurt Your Google Rankings (600,000 Pages Analyzed)”, 7 July 2025, updated 12 August 2026. ahrefs.com
- Ahrefs, “74% of New Webpages Include AI Content (Study of 900k Pages)”, 19 May 2025, updated 28 May 2026. ahrefs.com
- Google Search Central, “Creating helpful, reliable, people-first content”, last updated 5 October 2026. developers.google.com