Article

SEO for AI Overviews and AI Mode in 2026: What Actually Works

Google Search is increasingly shaped by AI Overviews and AI Mode, but that does not mean traditional SEO is obsolete. This guide explains what actually improves visibility in Google’s generative AI experiences in 2026, what tactics you can ignore, and how to measure AI Search visibility using Search Console.

Mahyar Molavi
Written by Mahyar Molavi Manager
SEO for AI Overviews & AI Mode in 2026
IN THIS ARTICLE Table of contents 16 sections

SEO in 2026 has a new source of anxiety.

A business owner searches one of their most important keywords, sees an AI-generated answer above the traditional results, and immediately asks:

“If Google answers the question itself, what happens to my organic traffic?”

The concern is understandable. Google Search now includes AI Overviews, AI Mode, multimodal search, and other generative experiences that can change how users discover information and which websites they decide to visit.

But one of the biggest mistakes businesses can make is assuming that AI Search requires an entirely new form of SEO.

Google’s own 2026 guidance says the opposite.

The company says its generative AI features are built on its existing Search ranking and quality systems. Google also says there are no special technical requirements to appear in AI Overviews or AI Mode beyond being eligible for Google Search itself. Google for Developers

So SEO is not being replaced.

The search experience is changing, and that changes what strong SEO needs to accomplish.

The goal is no longer only to rank a page for one exact keyword. Your content increasingly needs to be useful enough, clear enough, technically accessible enough, and distinctive enough to support a broader set of questions that Google may explore while building an AI-generated response.

That sounds like a major shift.

In practice, much of it comes back to doing SEO better rather than inventing a new discipline.

How Google’s AI Search actually finds sources

To understand what to optimise, it helps to understand what Google has publicly explained about the system.

Google says its generative AI search experiences use techniques including retrieval-augmented generation, often called RAG.

In simplified terms, the AI does not need to generate an answer entirely from what a model already “knows.” Google can retrieve current pages from its Search index and use information from those pages to ground the response.

Those supporting web pages can then appear as links in AI-generated results. Google for Developers

This is the first important point for SEO:

Your page still needs to be discoverable, crawlable, indexable and relevant.

Google also describes another technique called query fan-out.

Instead of treating a complex search as one isolated query, the system can generate several related searches to gather the information needed to answer the user properly.

Imagine someone searches:

“What should I check before redesigning a WordPress website without losing SEO?”

Google might need information about redirects, Core Web Vitals, internal linking, content migration, staging, analytics and mobile UX.

The AI system can explore several related concepts instead of relying on one exact keyword match. Google for Developers

This has an important consequence.

SEO becomes less about repeating one phrase and more about creating a page that genuinely understands the problem.

That does not mean creating 30 thin articles for every possible fan-out query.

Google specifically warns against producing large numbers of pages simply to capture every variation of how a user might search. Its systems can understand relevance even when the exact query wording does not appear on the page. Google for Developers

The better approach is to build useful topical depth.

The most important AI Search optimisation is surprisingly unglamorous

Google’s 2026 generative AI guidance places unusual emphasis on something that has nothing to do with AI-specific markup:

Create valuable, non-commodity content.

That distinction matters.

Commodity content is information that could come from almost anyone.

Consider an article called:

“10 SEO Tips for Small Businesses.”

If the page contains the same advice already repeated across thousands of sites — use keywords, create quality content, improve speed, build backlinks — an AI system has no shortage of alternatives.

Now compare that with:

“How We Diagnose a WordPress Redesign That Lost 40% of Its Organic Landing Pages”

The second topic can contain a specific diagnosis process, technical reasoning, screenshots, Search Console evidence, redirect analysis and conclusions from direct experience.

It provides something that is harder to replace.

Google explicitly recommends content with unique viewpoints and first-hand experience, and says publishers should avoid simply recycling what other websites or generative AI systems could easily reproduce. Google for Developers

This is one of the biggest opportunities for consultants, developers and specialist businesses.

Large publishers can produce enormous quantities of content.

A practitioner can produce something different:

experience.

For an SEO consultant, that could mean showing how a technical problem is diagnosed.

For a WordPress developer, it could mean explaining why one redesign architecture was chosen over another.

For an automation consultant, it could mean breaking down the logic behind an actual n8n workflow.

That kind of content is useful in traditional Search and becomes even more valuable when AI systems need credible sources to ground complex answers.

Stop thinking about “AI content” and start thinking about evidence

There is a subtle change businesses should make to their content strategy.

The question should not be:

“How do we write content that AI likes?”

A better question is:

“What can we publish that makes our explanation more useful and credible than a generic answer?”

Evidence can take many forms.

A consultant might include a decision framework that shows how different situations lead to different recommendations.

A developer might include a real architecture diagram.

An SEO specialist might show the Search Console patterns that indicate an indexing problem rather than a ranking problem.

An automation article might include a Workflow showing what happens when an API call fails.

A WordPress article could compare two implementation approaches and explain why one creates more maintenance risk.

This is Information Gain in practical terms.

The content does not just state the answer.

It helps the reader understand why the answer changes depending on context.

That makes the article useful even after Google has shown a short AI-generated summary.

And that matters because the real competition is not only for visibility.

It is for the click that happens when a user decides the short answer is not enough.

Traditional technical SEO still determines whether AI Search can use your content

A sophisticated content strategy will not help much if Google cannot properly access the page.

Google says pages that appear as supporting links in AI Overviews and AI Mode must be indexed and eligible to appear in Search with a snippet. The company does not require a separate AI-specific technical standard. Google for Developers

That makes many familiar SEO fundamentals directly relevant:

crawl accessibility, canonicalisation, internal linking, JavaScript rendering, duplicate-content control and page experience.

This is easy to underestimate because discussions around AI Search often jump directly to content formatting.

But consider a technically weak website.

Its important pages may only be reachable through JavaScript interactions. Several versions of the same page may exist. Canonical tags may be inconsistent. Internal links may not make the important pages obvious.

An AI strategy cannot compensate for that foundation.

For a website that wants more visibility in AI Search, I would still begin with many of the same questions used in a [technical SEO review](/#services):

Can Google crawl the important pages?

Are the correct pages indexed?

Does each important commercial intent have a clear destination?

Can Google understand the relationship between articles, services and supporting content?

Are important pages internally linked?

Are JavaScript-dependent sections actually accessible?

Is duplicate content creating unnecessary ambiguity?

AI Search makes these questions more important, not less.

Internal linking becomes even more useful when searches become more exploratory

Query fan-out is one reason I would pay particularly close attention to internal linking in 2026.

Suppose a business has a strong article about choosing an SEO consultant.

That article can naturally connect to:

an SEO audit guide,
a technical SEO article,
a relevant project,
the SEO service section,
and eventually the contact path.

That structure serves two purposes.

For users, it creates a logical path through the subject.

For search engines, it provides additional context about how pages relate to each other.

The goal is not to insert keyword-rich links into every paragraph.

It is to build a site where related expertise forms a coherent network.

That is why isolated blog publishing is increasingly weak as an SEO strategy.

If a business publishes 100 articles but none of them support its services, case studies or each other, the result is essentially a content archive.

A stronger architecture looks more like:

Problem → Guide → Related Expertise → Proof → Service → Contact

You can see the same principle in the way [services](/#services) and [portfolio work](/#portfolio) should support educational content rather than exist as unrelated sections of a website.

Do you need AEO or GEO instead of SEO?

This is where the marketing around AI Search becomes confusing.

You will frequently see terms such as:

AEO — Answer Engine Optimization

and:

GEO — Generative Engine Optimization

These terms can be useful when discussing specific aspects of AI visibility.

The problem begins when they are presented as completely separate systems that require proprietary hacks.

Google's position in its current guidance is explicit: from its perspective, optimising for generative AI experiences in Google Search is still part of optimising for Search. Google for Developers

The company also warns businesses to evaluate third-party AEO/GEO claims carefully.

Google states that third-party tools and services do not have access to its internal ranking data and cannot guarantee performance in Google Search or its AI experiences. Google for Developers

This does not make every GEO tool useless.

Some can help with research, monitoring, competitive analysis or workflow efficiency.

The problem is treating an external visibility score as though it were an official Google ranking metric.

A tool can produce an estimate.

It cannot give you Google's internal data.

That distinction should influence how businesses spend money on AI Search optimisation.

You do not need an llms.txt file for Google AI Overviews

One of the more interesting parts of Google's 2026 documentation is how directly it addresses several popular AI Search tactics.

The first is llms.txt.

Some websites have begun publishing files designed to make content easier for language models to understand.

That may be relevant for other systems.

But Google states that Google Search does not use llms.txt as a special signal for its generative AI Search features. Creating one will neither improve nor damage Google Search visibility. Google for Developers

That makes the decision simple.

If another platform you care about uses the file, maintaining one may be reasonable.

Do not create it because someone claims it is required for Google AI Overviews.

It is not.

You do not need special AI schema either

Structured data remains useful.

It helps Google understand certain types of information and can make pages eligible for supported rich results.

But there is currently no special Schema.org markup required for AI Overviews or AI Mode.

Google explicitly says not to overfocus on structured data for generative AI visibility. Google for Developers

The sensible strategy is therefore unchanged.

Use valid structured data when it accurately represents the visible content and when it has a legitimate purpose.

Do not fill the site with invented schema types because somebody promises they will make an AI model cite your business.

Technical SEO should reduce ambiguity.

It should not create markup theatre.

You do not need to break every paragraph into “AI-friendly chunks”

Another popular recommendation is to write every article as dozens of tiny answer blocks so language models can extract individual passages.

Google says this is not required.

Its systems can understand multiple topics and nuances within a page, and there is no ideal page length or mandatory chunk size for generative AI Search. Google for Developers

This is especially important because content written entirely for machine extraction often becomes unpleasant for humans.

An article can quickly turn into:

one sentence,
one heading,
three bullets,
another heading,
another definition.

That structure may be easy to scan, but used excessively it removes depth and natural explanation.

The better standard is the same one we use for the content strategy on MahyarMolavi.ir:

Write for an intelligent reader first.

Use sections where they help navigation.

Use direct answers when the query deserves one.

Use tables when comparison benefits from a table.

But do not destroy the natural flow of the article to satisfy a theory about AI parsing.

Images and video matter more than many AI SEO strategies acknowledge

Generative Search is not purely textual.

Google's AI Search guidance specifically recommends supporting appropriate content with useful, high-quality images and video. Google for Developers

That became even more relevant in September 2026 when Search Console added reporting for multimodal web search.

The new reporting can include searches originating from Google Lens, Circle to Search on Android, image uploads and Chrome image search. Google for Developers

This creates a useful opportunity for businesses whose expertise can be demonstrated visually.

A WordPress article can include a before-and-after performance comparison.

An SEO guide can contain an annotated Search Console example.

An automation article can show an n8n Workflow diagram.

A technical guide can visualise system architecture.

This is more useful than adding decorative stock images because the visual itself carries information.

Image SEO also remains relevant: descriptive context, appropriate filenames, accessible alt text and technically optimised media still matter.

The important shift is to think of useful visuals as part of the content, not decoration added after the article is finished.

AI Search performance is finally measurable inside Search Console

This is one of the biggest practical changes for SEO teams in 2026.

Google introduced a dedicated Generative AI performance report in Search Console and completed its worldwide rollout on August 31, 2026. The report covers impressions from AI Overviews and AI Mode. Google for Developers

Instead of relying only on third-party estimates, site owners can now see which pages receive impressions in Google's generative AI features.

The report can break this visibility down by page, country, device and date. It also supports text-based and multimodal search types. Google Help

This changes how I would conduct an AI Search audit.

Start by identifying which pages already receive generative AI impressions.

Then compare those pages with similar content that receives none.

Look for patterns.

Are pages with original examples appearing more often?

Does the site have visibility primarily around informational content but none around commercial topics?

Are old articles appearing while new strategic pages are not?

Which countries and devices generate that exposure?

The goal should not be to obsess over an “AI visibility score.”

The goal is to understand where Google is already finding the site useful and where stronger content or technical improvements may create additional opportunity.

Do not confuse impressions with business results

AI Search visibility is interesting.

Revenue is more important.

This distinction becomes essential as new reports appear.

A page can receive many AI impressions without generating valuable business outcomes.

Another page may receive fewer impressions but introduce the company to people who are much closer to purchasing a service.

That is why AI Search reporting should be evaluated alongside broader business metrics.

For a service business, I would want to understand:

AI visibility → Organic visits → Service exploration → Portfolio interaction → Qualified enquiry

Not every part of that journey will be perfectly attributable.

But the principle prevents a company from spending months chasing an impressive AI visibility graph that has no connection to actual customers.

For MahyarMolavi.ir, for example, visibility around generic “what is SEO?” topics would be less valuable than visibility around searches related to SEO audits, WordPress redesigns, technical problems or business automation.

The useful metric is not visibility alone.

It is qualified visibility.

Google now also gives sites control over participation in generative AI Search

Another important 2026 change is the Search generative AI control inside Search Console.

Google now allows site owners to choose whether their site's links and content can participate in AI Overviews, AI Mode and generative AI features in Discover. As of August 31, this control is available worldwide. Google Help

The default is inclusion.

When included, content can appear as links and can be used to ground AI responses.

A site can also choose to exclude its content from those generative AI experiences without necessarily using noindex to remove itself from Google Search entirely. Google Help

For most businesses pursuing organic discovery, exclusion will not be the obvious default.

But it is useful that site owners now have an explicit control.

It also reinforces an important point:

AI Search is no longer a hypothetical future SEO channel.

Google now provides both dedicated visibility reporting and dedicated participation controls for it.

What should an AI Search content workflow look like in 2026?

I would not create a separate “AI SEO department.”

I would upgrade the normal SEO workflow.

Start with the real user problem and search intent.

Understand the traditional SERP as well as whether the topic is likely to benefit from an AI-generated synthesis.

Then ask what expertise you can add that is not already available in dozens of generic pages.

That may be first-hand experience, a real workflow, technical evidence, a framework, a calculation, an original example or a meaningful visual.

Build the page around the complete problem rather than one exact keyword.

Make the important answer clear, but give users a reason to continue reading.

Then make sure the page can actually be crawled, indexed and found through the site's internal architecture.

Finally, measure the page in both normal Search performance and the Generative AI report.

This is much more sustainable than trying to reverse-engineer which sentence length an AI system supposedly prefers.

What actually matters — and what does not

A useful way to summarise AI Search SEO in 2026 is to separate foundations from distractions.

Focus on thisDo not mistake this for a requirement
Unique, experience-led contentllms.txt for Google
Crawlability and indexing“AI-only” technical SEO
Strong internal linkingHundreds of fan-out pages
Clear answers with real depthKeyword repetition
Original evidence and examplesGeneric AI summaries
Useful images and videoDecorative media added only for SEO
Valid structured data where appropriateSpecial AI schema
Search Console AI reportingThird-party “Google AI ranking scores”
User satisfactionForced tiny content chunks
Service and business relevanceAI visibility with no conversion value

Google's own current guidance strongly supports this direction. Google for Developers

AI Search makes strong SEO more valuable, not less

AI Overviews and AI Mode undoubtedly change Google Search.

Some informational searches can now be answered without the user needing to visit several pages.

That creates pressure on content that only repeats information available everywhere else.

But it also creates an opportunity.

When Google needs current information, specialist insight, supporting evidence or sources that can help users explore a topic more deeply, the open web is still part of the system.

The advantage therefore moves toward pages that deserve to be referenced.

Not pages that merely contain the keyword.

That means the strongest SEO strategy for AI Search looks surprisingly familiar:

Build a technically accessible website.

Publish content that genuinely knows something.

Connect related pages clearly.

Use real experience where you have it.

Support explanations with evidence and useful media.

Avoid shortcuts that exist mainly because “AI SEO” sounds new.

And measure what Google actually reports rather than relying exclusively on third-party predictions.

If your website already ranks in traditional Search but you do not know whether it is gaining visibility in AI Overviews or AI Mode, the next step is not necessarily to rewrite every page.

Start by checking your Search Console generative AI performance, identify the pages already appearing, review the site's technical foundations and then decide where better content can create a meaningful advantage.

You can review the broader [SEO and digital services](/#services) and [portfolio](/#portfolio), or [send your website details](/#contact) for an AI Search SEO review focused on technical eligibility, content quality, generative AI visibility, internal linking and the commercial value of the queries where your site appears.

Mahyar Molavi
About the author

Mahyar Molavi

Manager

I have been developing websites and creating content for over ten years, and this is just the beginning...

EMAIL[email protected]
WORK TYPERemote / Project Based
RESPONSE TIMEUsually within 24–48 hours

Start a collaboration

Collaboration type (optional)

I’ll respond as soon as possible.

Start a collaboration

Have a project in mind? Let’s talk about it

If you need help designing, developing, optimizing or growing a digital product, share the first details and we can discuss the most practical path.

Available for new projects