How to Rank in Google AI Overviews (Real Examples & Tracking Tips)

For years, SEO experts followed a familiar playbook: target the right keyword, write clean content, build topical authority and hope for a Featured Snippet or a spot on page one.

But since Google officially rolled out AI Overviews in May 2024, the rules have started shifting fast.

In just six months, Google reported over a billion (!) users successfuly engaged with AI Overviews, a strong enough signal to double down on expanding it.

What started as an experiment is rapidly becoming a core part of the search experience, not just another SERP addition, but a fundamental shift in Google’s DNA.

So the real question is: how do we adapt?

That’s exactly what this guide is here to answer – let’s dive in.

What Are AI Overviews?

AI Overviews (previously called SGE – Search Generative Experience) are generative AI-powered summaries that appear at the very top of Google’s results page.

These summaries don’t just pull a quote from one site. They synthesize a complete answer using multiple sources, often blended with original text generated by Google’s Gemini model, and link back to the content they were built from.

So how does it work?

AI Overviews are enhanced SERP features designed to give users a full, AI-generated answer directly on the search page before showing any standard organic results.

Unlike traditional Featured Snippets that extract one paragraph from a single page, AI Overviews blend content from multiple pages, combine it with model-generated text, and present it as a unified response based on search intent, not just keywords.

And here’s the critical part:

Each AI Overview includes a box of clickable links to the pages Google pulled information from – that’s where you want to be!

But here’s the catch – not all sources in that box are visible by default.

In many cases, a user has to:

1. Click “Show more” to expand the AI Overview
2. Then click “Show all” to reveal the full list of sources

So if your site is listed 4th or 5th in that box, users will need to click twice just to even see you and a third time to land on your page.

In other words:

This is a whole new level of search competition.

If you’re not in the top two visible links, your chances of getting traffic drop significantly.

To illustrate, here’s an example from one of our clients, Incredibuild (development acceleration platform), which ranks in the AI Overview for the query “implicit programming”.

Step 1: You see a single visible link inside the source box:

Step 2: Clicking “Show more” reveals two additional sources:

Step 3: Clicking “Show all” finally displays the full list of cited links:

But here’s where things get weird:

By the time you get to Step 3, your link which originally appeared near the top might suddenly shift down the list (I’ve seen this happen a few times).

Bug? Maybe.

But if it’s intentional, here’s my take:

If the user expanded the box twice, that’s a strong signal the top links didn’t deliver. So Google “makes room” for others, hoping one of them will.

Think of it as a reverse version of classic SEO CTR logic:

• In organic search: more clicks → better ranking.
• In AI Overviews: more expansion → lower visibility.

In other words, the AI reacts in real time to user behavior and reshuffles the link order accordingly.

Smart? Yes.
Challenging? Absolutely.

By the way, on mobile, it’s even worse – you typically see just one visible link, tucked at the bottom of the screen.

How Popular Are AI Overviews Right Now?

According to a March 2025 study by Semrush and Datos, AI Overviews now appear in 13.14% of all desktop searches in the U.S., nearly double the rate from just two months earlier (6.49% in January). The vast majority of these queries are still informational (88.1%), but commercial intent is rising, signaling Google’s slow but steady expansion into what used to be more “sensitive” or high-conversion query types.

At the same time, Google officially rolled out Gemini 2.5 two months ago, its upgraded generative model powering AI Overviews. Among the new capabilities: support for more complex queries (like code, comparisons, and math), a new experimental AI Mode, and multimodal reasoning that allows Google to tackle layered searches using text, images, and real-time data.

How Are AI Overviews Different From GEO and Featured Snippets?

Let’s clear things up for a second.

We recently shared a guide on Generative Engine Optimization (GEO), and it’s basically an umbrella term for getting your content into AI-generated answers.

But GEO actually includes two different worlds:

1. External AI platforms like ChatGPT, Perplexity, and Gemini, where the models pull info from the web or their memory to build an answer.
2. Google Search itself, where AI Overviews now show up right inside the results, powered by Google’s own model (Gemini), but still based on the same idea of auto-generated answers.

In both cases, the answer is written by AI. The only real difference? Where that answer lives on Google, or outside it.

Here’s a quick side-by-side to break it down:

So What Does This Mean for You?

AI Overviews are changing the game.

It’s no longer enough to write good content or rank in the top three. Now, you need to be one of the 3–4 trusted sources that Google uses to build its AI-generated answer.

Here’s how that plays out in real life:

• If you’re a private clinic: You used to compete for the Featured Snippet on “how to treat a herniated disc.” Now, you need to be one of a few hand-picked sources feeding the AI Overview.
• If you’re a travel blogger: Queries like “Is Iceland kid-friendly?” may now show a complete answer without your article, unless Google thinks your post adds unique, irreplaceable value.
• If you’re running an eCommerce site: It’s not enough to write “Best summer mattresses.” You need to appear neutral, trustworthy, and well-referenced across the web to be selected as a source.

How to Rank in AI Overviews?

Google hasn’t published an official algorithm for how it selects sources for AI Overviews, but it has shared a basic guideline page on Search Central. Beyond that, real-world SEO tests, community experiments, and several solid case studies have helped paint a clearer picture.

So how do you actually get featured there?

1. Understand the User’s Intent, Not Just Their Keywords

AI Overviews don’t rank based on keywords alone, they aim to solve the searcher’s problem.

That means Google’s language model is looking for content that understands the intent behind the query and responds to it clearly and practically.

Let’s look at two examples from our clients that show how this works in the real world:

Example 1:

Incredibuild, a platform for accelerating game development workflows, ranks for the query “Unity vs Unreal”:

Why?

Because their article mirrors the searcher’s intent by offering an in-depth, side-by-side breakdown. It includes a comparison table, pros and cons, use cases by audience (e.g., indie devs vs AAA studios), and real-world performance considerations. The structure and tone match what someone typing “vs” into Google typically expects: an unbiased, scannable, and helpful comparison.

Example 2:

CoreTigo, which provides wireless automation solutions for industrial settings, appears for the query “wireless industrial automation”:

The article aligns with the intent behind the query: it defines the concept, explains where to begin, walks through implementation steps, and compares key technologies like IO-Link Wireless vs Wi-Fi or BLE.

Instead of a product pitch, the page acts like a mini-guide, exactly the kind of structured, informative content that fits naturally into an AI Overview.

In both cases, the content anticipates what the user wants to understand, not just what they typed. That’s what makes it useful and that’s what makes it rank.

2. Match Google’s Preferred Format: Structure, Style & Tone

According to Yongi Barnard at Backlinko, most content that makes it into AI Overviews tends to follow one of these four formats:

1. Q&A style: clearly defined questions and answers
2. Step-by-step guides: for how-tos and processes
3. Comparisons: especially between products or service types
4. Conversational tone: written in plain, natural language

What to do:

• Include real FAQs in the body (not just in schema).
• Use subheadings phrased as questions (“How does it work?”, “What are the pros and cons?”).
• Write clear, concise, and focused answers.
• Use comparisons or contrasting points wherever relevant.

3. Build Trust with E-E-A-T Signals

Google’s AI model looks for credible, authoritative sources when building AI Overviews. That includes assessing E-E-A-T signals:

• Experience: First-hand use or real-world familiarity
• Expertise: Professional or topic-level knowledge
• Authoritativeness: Strong reputation for both author and domain
• Trustworthiness: Transparency, clear sourcing, and accurate info

What to do:

• Add an author byline with a short, relevant bio, especially if they have direct experience in the topic.
• Reference external sources: research, studies, regulations, or data.
• Showcase social proof: reviews, testimonials, or user comments.
• Use first-person perspective when the writer has direct experience, it boosts authenticity.

Example:

Search for “409A” in Google, and you’ll find an AI Overview answer that pulls from several professional sources. One of them is a detailed article written by Philip Stein, a client of ours (accounting and tax firm), with both Yaacov Jacob and Louis Barr credited by name and title within the article.

What’s interesting? Even Morgan Stanley shows up in the same AI Overview, but their article includes no author attribution at all.

That’s not unusual. Most articles on 409A in big corporate websites don’t mention who actually wrote them. That’s exactly why including a real byline with proven expertise can help you stand out. It’s a simple signal that can give smaller firms an edge over much bigger names.

4. Multi-Dimensional Content – Not Just a Single Paragraph

AI Overviews favor content that explores a topic from several angles, not just a short definition or a basic explanation.

The model is trying to build a rich, well-rounded response. That means pulling from sources that cover not just what something is, but also how it works, why it matters, and what to do next, all in one place.

Example:

If you search for “what is ESPP”, you’ll see another article from Philip Stein:

Here’s why it ranked there:

• Beyond the definition: The article doesn’t just explain what an ESPP is, it also walks the reader through how it works, how to participate, and common mistakes to avoid.
• Layered explanations: It includes both U.S. and international tax implications, making it relevant to a wider audience with different situations.
• Multiple entry points: Subheadings like “Are all ESPPs the same?”, “Key features of an ESPP”, and “Tax complications” offer the model various angles to pull from ideal for AI synthesis.
• Scannable structure: Clear subheadings, bullet points, and definitions make it easy for the model (and user) to parse information quickly.
• Actionable content: The article ends with real guidance who to consult, what questions to ask, and links to related topics, making it a strong candidate for AI answers that aim to assist, not just inform.

In short, this page ranks not because it has the best one-liner definition, but because it gives the model a complete kit to build a helpful, nuanced response.

5. Freshness & Content Updates Matter

Especially in verticals where Google is already allowing AI Overviews to show up (like health, legal, travel, and tech), content freshness is critical.

Looking at the examples we’ve seen so far, outdated pages (older than 2–3 years) almost never show up in the box.

What to do:

• Update your publish dates.
• Include new data, stats, or references.
• Add a line like “Last updated on [DATE]” at the top of your content.
• Make sure your page reflects the current landscape, not the world as it was in 2019.

6. Structural & Semantic Optimization

Google’s LLM doesn’t just read your content, it interprets its structure, semantics, and layout.

Headings, lists, bold terms, and schema markup all help the model parse meaning and identify helpful passages.

What to do:

• Use proper heading hierarchy (H1 → H2 → H3).
• Bold important concepts and terminology.
• Use numbered or bulleted lists instead of long narrative paragraphs.
• Add relevant Schema markup (especially FAQPage, HowTo, WebPage).

7. Off-Page Presence Counts Too

Unlike featured snippets (which come from a single URL), AI Overviews consider the broader context of your domain.

Sites mentioned often across the web, especially in authoritative places, are more likely to be chosen.

What to do:

• Answer relevant questions on Reddit, Quora, and X.
• Build backlinks and mentions from trusted websites.
• Contribute content or insights to industry blogs.
• Partner for content exchanges to earn citations.

8. Keep It Neutral, No Sales Pitching

The SEMrush study referenced earlier shows that AI Overviews are heavily skewed toward long-tail, informative queries with low commercial intent.

Here’s a breakdown:

The following screenshot from the Semrush report clearly illustrates the typical query profile that triggers AI Overviews: usually long, precise, non-commercial, and focused on learnin, not buying.

What to do:

• Write in a comparative, balanced tone: “X is better for Y… Z might be preferred when…”
• Avoid one-size-fits-all claims.
• Offer readers options, not conclusions.
• Ditch language like “the best” in favor of phrases like “among the most popular in [region].”
• Even commercial sites can win here if they offer value first, product second.

9. Support Multistep Reasoning

One of the biggest upgrades in Gemini 2.0 (which now powers AI Overviews) is its ability to handle multi-step reasoning.

As Google noted in March 2025, AI Overviews can now:

• Solve multi-step math problems
• Compare products based on layered criteria
• Combine info from several sources to answer nuanced queries

Behind the scenes, Gemini 2.0 uses a technique called Query Fan-Out: It breaks a query into smaller questions, fetches answers separately, and then merges them into one cohesive result.

Let’s say someone searches: “Smart ring vs smart watch vs mattress with sensors”, Google’s LLM might:

• Run separate sub-queries for each item
• Look for pros and cons across several sources
• Consolidate everything into a final answer that reads naturally

What to do:

• Use step-by-step formatting for guides (“Step 1: Define your goals…”).
• Add comparison tables that highlight differences by feature.
• Include follow-up questions your reader might ask next.

Example:

If you’re writing about college degrees, don’t just explain “how to choose a major”. Walk them through steps like: interest assessment, job market overview, lifestyle alignment, financial aid considerations etc.

This layered thinking mirrors how Gemini now builds its answers, and gives your content a better shot at getting featured.

For more, check out this Whiteboard Friday episode from Moz:

How to Track Your Appearance in Google AI Overviews

Not everything that matters can be measured, but everything you can measure, matters 🙂

If you’re investing time and resources to appear in AI Overviews, don’t rely on guesswork. Start tracking, tagging, and analyzing.

While you may not be able to capture 100% of appearances, you can spot trends, identify top-performing pages, and gain a competitive edge.

1. Track Appearances Using Third-Party Tools

AI Overview placements aren’t traditional rankings, but they are placements.

The first step in understanding your AI Overview visibility is simply knowing for which queries this feature is triggered, and ideally, whether your site is part of it.

Tools like Semrush, Ahrefs, and others now offer SERP feature tracking that includes AI Overviews.

How to Track AI Overviews in Ahrefs:

• Go to Site Explorer → Organic Search → Organic Keywords
• Then filter by SERP Features → select AI Overview

Here’s what that looks like in the dashboard (client name blurred):

Additionally, Ahrefs now offers a beta feature called Brand Radar.

This tool lets you search for a specific brand or product, choose a country, and see how it’s being mentioned across generative engines, essentially functioning as an AI visibility report.

How to Track AI Overviews in Semrush

If you want to know whether your site appears inside an AI Overview, not just whether the feature is triggered, Semrush currently offers one of the most comprehensive monitoring solutions available. Here’s how to access and use that data:

• Position Tracking: Your go-to tool for daily keyword monitoring. Here’s how to set it up:
  • Create a Position Tracking project for your domain with relevant keywords.
  • Go to the SERP Features tab and filter for AI Overview.
  • The tool will show which terms are triggering the AI Overview, and whether your site is featured as one of the sources.
  • If you are included, Semrush will count your site as ranking #1 even if your regular organic position is lower.
  • You can then track changes over time and identify which pages are consistently being pulled in and which aren’t.
• Organic Research / Domain Overview: Use this for a high-level view of your domain’s search visibility:
  • See how many of your keywords trigger the AI Overview feature.
  • And among those, how many include actual links to your site inside the summary box.
• Keyword Overview / Keyword Magic Tool: These tools are great for researching specific terms or building keyword lists:
  • Enter a keyword and apply the filter SERP Features > AI Overview.
  • You’ll be able to see which domains are being featured, and identify new content opportunities.
• Semrush Sensor: Want to understand broader market trends?
  • The Sensor tool shows how often AI Overviews appear broken down by industry (health, law, travel, etc.).
  • You can track daily fluctuations, identify emerging verticals, and monitor how prominent AI Overviews are across search results.

Once you know which pages from your site are showing up in AI Overviews, connect that data to your Google Analytics 4 reports to evaluate impact:

• Did traffic to that page increase?
• Did users stay, read, convert?
• Are specific sections of the content driving clicks? (Use Text Fragments and Google Tag Manager to track.)

This way, you’re not just tracking appearances you’re measuring the real value.

Heads up: If you’re featured inside the AI Overview, Semrush will count it as position #1, even if your organic rank is #5 or #10.

✅ You can try this AI Overview tracking feature using Semrush’s 7-day free trial

Disclosure: this is an affiliate link, if you decide to purchase a plan after the trial, we’ll receive a commission.

2. Tracking Real Clicks from AI Overviews

Even if you’ve earned a spot in an AI Overview, the real question is:

• Did users click your link?
• Did they land on the page?
• And what happened next?

That’s something tools like Semrush won’t show, it requires tracking directly on your site.

The method below is based on research from Brodie Clark, who first identified the unique link format used in AI Overviews, Featured Snippets, and People Also Ask. He also published a tracking script to help measure visits from those sources.

In the following video, Dana DiTomaso expands on Brodie’s method with a more refined solution that works seamlessly with GA4, even under strict character limits for event names:

How does a link from AI Overviews actually work?

When Google pulls a paragraph from your website into an AI Overview, it often creates a deep link that jumps directly to that section, with highlighted text.

The URL will typically look like this:

https://example.com/page#::text=what%20is%20active%20oxygen

This is called a Text Fragment, a URL format that allows browsers to scroll directly to, and highlight the specific passage shown in the AI Overview.

The Problem: GA4 doesn’t track this by default.

Google Analytics (including GA4) doesn’t capture anything after the # in a URL.

So in your reports, you’ll just see a visit to example.com/page, but you won’t know which snippet was clicked or why the user came.

The Solution: track with Google Tag Manager (GTM).

Here’s how to track it:

• Create two custom variables in GTM:
  • One for the beginning of the ::text= fragment
  • One for the end
• Use them in a page_view or custom event: Every time a page loads with a ::text= in the URL, those values will be captured and sent to GA4 as event parameters
• Set up Custom Dimensions in GA4: Register the parameters as custom dimensions (event-scoped) so you can include them in reports
• Build reports by snippet, page, or source: This lets you analyze which text fragments are driving real engagement from AI Overviews, Featured Snippets, or People Also Ask

Limitations to Keep in Mind

• Not every AI Overview link includes a ::text= fragment, some are basic links
• You won’t capture 100% of clicks, but the ones you do will be extremely accurate
• GA4 limits parameter values to 100 characters, so splitting the snippet start and end (as shown in the video) is recommended

Benefits of This Method

• You’ll see which parts of your content are getting clicked
• You can analyze whether AI Overview traffic leads to engagement or just bounces
• You’ll have real data to decide whether to rewrite, boost, or repromote certain content

3. Indirect Ways to Measure AI Overview Impact

Even if you can’t confirm a snippet click (no text fragments), you can still monitor indirect signals:

• Traffic changes to informational pages.
• Drops in CTR for queries now showing AI Overviews.
• Increases in mentions or shares (social proof).
• Query-type analysis in Google Search Console.

Key Takeaways: Where This Is Going, and What to Do Tomorrow

AI Overviews are still evolving, but the direction is clear: Google isn’t just listing answers. It’s generating them.

At the same time, more brands are realizing that content now needs to serve two audiences: humans and machines. If you want to appear in answers from ChatGPT, Gemini, or Perplexity, your content must be usable by language models, not just optimized for people.

This is where GEO (Generative Engine Optimization) comes in and its scope is broader than most think.

GEO spans two ecosystems:

• External LLMs like ChatGPT, Gemini, and Perplexity, where answers are generated from crawled content or embedded knowledge.
• Google Search itself, where AI Overviews synthesize responses on the results page using those same models, but within Google’s platform.

In both environments, it’s the LLM that decides what content to cite. The only difference is where that answer shows up, inside an external generative AI platform or directly in Google search results.

So GEO is about creating content that large language models across the web can recognize, trust, and quote.

Different interfaces. Same rules.

Is It Worth the Effort?

A February 2025 study from Ahrefs examined 3,000 websites and found that 63% received at least one visit from an AI source. While that still represents a small share of traffic (on average, 0.17% of total), here are some key takeaways:

• Small websites received a significantly higher relative share of AI traffic than large ones (opposite trend from Gemini, where large sites tend to dominate).
• ChatGPT accounted for over 50% of total AI traffic, followed by Perplexity and Gemini.
• Some visits show up as “Direct” in analytics, because bots like Copilot don’t pass a referrer.
• Even if people don’t click, being mentioned in answers still builds brand exposure. The visibility effect is real even when traffic is hard to track.

What it means: Even if you’re not seeing hundreds of visits from language models yet, they’re already there, behind the scenes. If you ignore them now, you risk losing a competitive edge that’ll be hard to win back a year from now.

What should you start doing today?

• Map out your informational pages: Review which pages on your site address user questions. Identify the ones that look like “answer candidates”, and rewrite them with AI Overviews in mind.
• Use structures that LLMs can understand: Add clear Q&A headings, tables, bullet lists, comparisons and use supporting schema like FAQPage or HowTo.
• Start tracking your presence in AI Overviews: Use tools like SEMrush to see if your site appears there, and GA4 + GTM to measure actual clicks from these features.
• Apply GEO principles early: Write for both humans and machines. Use natural language, maintain semantic consistency, ensure topical density, and structure content for easy extraction.
• Play the long game: This isn’t a sprint. It’s a deep shift in how content is evaluated and surfaced. The sooner you start, the better your long-term advantage.