How AI is reshaping search intent, and what to do about it
Search intent has always been the foundation of good SEO. Before you write a word, build a link, or touch a title tag, you need to understand what the person typing that query actually wants. Not the keyword. The intent behind it.
That part has not changed. Everything else around it has.
AI is reshaping where intent gets satisfied, which content types survive, and what it actually takes to rank now. If your strategy was built before AI Overviews became a fixture on the SERP, it is probably running on outdated assumptions. Some of those assumptions are costing you traffic right now. Others will cost you more over the next twelve months.
This piece covers what is actually happening, why it hits differently depending on your content type, and what to do about it in practical terms.
What AI Overviews are actually doing to intent
Google's AI Overviews are not a new SERP feature. They are a structural change in where search intent gets resolved.
For informational queries, the intent is increasingly satisfied before anyone reaches your site. Question asked. Question answered. No click required. Google is not trying to hurt publishers with this. It is just building the product it wants to build, and that product does not need your page to function.
Worth knowing
Tracking data now tells a more severe story than early estimates suggested. Ahrefs found a 58% CTR drop for top-ranking pages when AI Overviews are present. Seer Interactive's study of 3,119 informational queries across 42 organisations found organic CTR down 61% since mid-2024. Even the most conservative studies put the decline at 35% or more. The direction is consistent across every data source and still trending down.
A significant chunk of informational content that currently drives traffic will drive considerably less of it in the coming years. Not because it is bad. Because the SERP itself became the destination.
That is the threat. There is also an opportunity most people are missing, and we will get to that.
The four intent categories and how AI affects each
Most SEOs still work with the three classic intent categories: informational, navigational, transactional. That model was designed for a pre-AI SERP. There is a fourth category that AI is hitting harder than anything else, and most people are not paying enough attention to it.
Informational
Queries seeking a direct answer or explanation. "What is hreflang", "how does PageRank work", "what does noindex mean."
AI Overviews are resolving these on-SERP. Clicks declining now, will accelerate.
Commercial investigation
Queries comparing options before a decision. "Best SEO tools for agencies", "Semrush vs Ahrefs", "top casino sites UK."
AI is surfacing summaries and comparisons directly. This category is being disrupted faster than most realise.
Transactional
Queries with clear purchase or conversion intent. "Buy X", "sign up for Y", "download Z."
Conversion still happens off-SERP. AI can influence consideration but the transaction itself remains yours to win.
Navigational
Queries seeking a specific destination. "Gmail login", "Ahrefs pricing page", "Sky Sports fixtures."
No AI summary substitutes for the actual page. Brand and destination queries remain largely unaffected.
Commercial investigation is where most affiliate and comparison content lives. It is also where AI summaries are doing the most damage. Search "best VPN for streaming" or "top iGaming affiliate programs" and Google can now pull together a comparison from multiple sources without anyone clicking anything. The comparison happens on the SERP.
The content is not dead. The bar for earning a click just got a lot higher.
How to get cited inside AI Overviews
Here is what most pieces about AI and SEO get wrong: they frame the whole thing as a loss. Traffic goes down, clicks disappear, content becomes pointless. That is only half the picture.
AI Overviews cite sources. Cited sources get attributed links. Those links get clicked, and the people clicking them are usually the ones who want more than a summary. Qualified traffic. The kind that converts.
Getting cited in an AI Overview is a new ranking category with its own signals. Here is what moves the needle.
Structure your content for extraction
AI Overviews pull concise, clearly structured answers. Use H2s and H3s that directly match the question being asked. Write a clear, direct answer in the first paragraph under each heading rather than building to it. Think of it like writing for a featured snippet, because the underlying mechanism is similar.
Include the things AI cannot generate itself
Original data, proprietary research, named expert opinions, and specific statistics with attribution are significantly more likely to be cited because they represent information the model cannot produce from its own training. If you have original findings, lead with them.
Establish entity authority
Google needs to know who you are and why your site is a credible source on a given topic. This means a clear About page with real author information, consistent authorship across your content, and ideally some external references to your site or its authors from trusted sources in your niche.
Cover topics with genuine depth
AI Overviews tend to cite sources that demonstrate comprehensive topical coverage rather than surface-level takes. A site with twenty well-researched articles on a specific subject area is more likely to be cited than a site with one article on the same topic.
Use schema markup correctly
Structured data helps Google understand and extract your content more reliably. Article, FAQ, HowTo, and Speakable schema are all relevant depending on your content type. More detail on this below.
What AI-resistant content actually looks like
"AI-resistant content" gets repeated constantly and explained almost never. So here is what it actually means.
Content survives AI displacement when it contains things a language model genuinely cannot produce. Not because AI writes badly. It does not. But because the value in the content comes from somewhere outside its training data.
First-hand experience
Content written by someone who has actually done the thing being described. A review of an SEO tool written by someone who uses it daily reads fundamentally differently from one synthesised from existing reviews. That difference is legible to both readers and, increasingly, to Google.
Original data and research
Numbers, findings, or observations that do not exist anywhere else online. Your own test results, your own analysis of a dataset, your own survey findings. These are citation magnets and genuinely cannot be replicated.
Specific, named opinions
A clearly attributed perspective from someone with demonstrated expertise. Not "many experts believe" but "here is what I have found after twelve years in iGaming SEO." The specificity and the attribution are both part of what makes it valuable.
Recency and timeliness
Content about things that have just happened, or that requires knowledge of current events, has a natural advantage because training data has a cutoff. Industry news, product updates, regulatory changes, and market shifts all benefit from this.
Community and conversation
Content that aggregates real perspectives from real people: forum discussions, community roundups, expert interviews. The value is in the genuine variety of human viewpoints, not just the information itself.
Depth that exceeds the summary
Content where reading the full piece delivers significantly more value than the summary version. Step-by-step guides with genuine nuance, case studies with full context, analyses that require the reader to follow a line of reasoning.
The structured data opportunity
Schema has been around long enough that most SEOs treat it as a nice-to-have. In the AI era it is closer to essential.
When Google's systems extract, attribute, and present your content in an AI Overview, schema is one of the clearest signals they have about what the content is and who wrote it. Less ambiguity means a higher chance of being cited correctly.
| Schema type | When to use it | AI benefit |
|---|---|---|
| Article | All editorial content | Establishes authorship and publication context |
| FAQPage | Q&A sections within articles | Makes individual Q&A pairs extractable |
| HowTo | Step-by-step guides | Structures process content for direct extraction |
| Person | Author pages | Builds entity authority for cited authors |
| Speakable | Key passages in articles | Flags priority content for voice and AI extraction |
| Review | Product or service reviews | Signals first-hand evaluation and expertise |
Start with the basics. Article schema on every editorial piece, with a named author tied to a Person entity. FAQ markup on any Q&A section. HowTo on anything step-based. None of this is complicated to implement and most competitors have not bothered, which means there is still easy ground to take here.
One caveat worth knowing: FAQ and HowTo schema no longer reliably trigger rich results on all page types following Google's March 2026 update. The AI extraction benefit remains real, which is why they are still worth implementing, but do not expect the visual SERP features they used to generate.
What this means for your content strategy
The answer is not to stop producing informational content. It is to be honest about what role it plays and stop expecting it to pull traffic weight it no longer can.
Informational content still builds topical authority. It still supports the internal linking structure that helps commercial pages rank. It is just not a traffic channel in the same way it once was. If you are measuring it like one, you are going to keep being disappointed.
The shift in how to think about it
Informational content is infrastructure now. It makes your commercial content more credible. Before publishing anything informational, ask: what does this make possible elsewhere on the site? If the answer is nothing, reconsider whether it needs to exist.
Commercial investigation content needs depth and a point of view. If your comparison article reads the same as every other comparison article covering the same products, AI can replicate it easily. If it contains your actual testing, your specific recommendation, your take on who each option actually suits, that is harder to replace.
Transactional pages need to stand on their own. The funnel from informational content to conversion is narrowing. Make sure your commercial pages are competitive in their own right, not just benefiting from traffic flowing through them.
A practical audit framework
Enough theory. Here is how to actually assess where you stand.
Categorise your traffic by intent type
Pull your top 50 to 100 organic landing pages and classify each by the primary intent the content serves. Informational, commercial investigation, transactional, or navigational. If you are unsure, look at the query data in Search Console for each page.
Identify your exposure
What percentage of your organic traffic is currently coming from informational and commercial investigation content? That is your exposure number. For many content-heavy sites this will be 60% or more. That is not a reason to panic, but it is a number that needs a plan.
Check which queries are triggering AI Overviews
Search your key informational queries manually or use a tool that tracks AI Overview presence. Queries already triggering AI Overviews are your most immediate risk. Queries not yet triggering them may be next. Prioritise accordingly.
Assess your content for AI-resistance attributes
For each high-exposure piece, honestly evaluate whether it contains first-hand experience, original data, specific opinions, or depth that goes beyond what a model could produce. If the answer is no, it is a candidate for either upgrading or deprioritising.
Audit your structured data
Use Google's Rich Results Test or a schema validator to check what markup is currently in place. Look for missing Article schema on editorial content, missing author entities, and any FAQ or HowTo content that is not marked up.
Build your response roadmap
Based on the above, prioritise three things: content that can be upgraded to become genuinely AI-resistant, transactional content that needs to be strengthened independently of informational traffic funnels, and schema implementations that can be added quickly.
The AI production opportunity
Everyone is talking about AI as a threat to content. Fewer people are talking about it as the production advantage it actually is.
Publishing at volume used to mean either a big budget or a quality compromise. AI-assisted workflows remove that trade-off, if you build them properly. The emphasis is on the last part.
The distinction that matters
Prompting a language model and publishing what comes out is not an AI content workflow. It is a thin content factory. The teams making this work have quality control built into every stage, not just the generation step.
The tasks AI handles well: structure, first drafts of templated content, metadata, internal linking suggestions, content briefs. The tasks it does not replace: editorial judgment, genuine product knowledge, original perspective, the decision about whether something is actually worth publishing. Keep that division clear and AI becomes a serious competitive lever. Blur it and you are just producing content Google is increasingly good at ignoring.
The sites that will be hardest to compete with are the ones building proper AI workflows now, not the ones using AI as a shortcut.
Key takeaways
- AI Overviews are resolving informational and commercial investigation intent on-SERP. Clicks are declining for these query types and will continue to do so.
- Getting cited inside AI Overviews is a new ranking opportunity. Structure your content for extraction, lead with original data, and build entity authority.
- AI-resistant content contains things a model cannot generate: first-hand experience, original research, specific named opinions, and genuine depth.
- Structured data is more important than ever. Article, FAQ, HowTo, and Person schema all improve your chances of being correctly attributed and cited.
- Run the six-step audit to understand your actual exposure and build a prioritised response roadmap.
- AI is also a production opportunity. Teams that build proper AI-assisted workflows have a meaningful scale advantage over those that do not.