GEO 2.0: what changed in AI Overviews in 2026

Six months ago, I wrote a post about optimizing your site for AI. Back then, Google AI Overviews was new, still had bugs, and nobody knew how to prioritize.

Today it’s different. Data I collected over 6 months shows clear patterns. Google AI Overviews gained volume, ChatGPT Search appeared, Perplexity got more aggressive. And each one works differently.

I’ll update you.

Recap: what is GEO

GEO is when an AI search engine (Google, ChatGPT, Perplexity) answers your question right on the results page without you clicking. The answer comes from multiple sources compiled into one.

Before, you’d search “best toothpaste for sensitivity”. Google showed 10 links. You clicked one. Read. Went back.

Now? You type the same question. Google shows a paragraph with the answer compiled from 3 sites, with citations. You read and done. Never click anything.

For your site, is this good or bad? Depends if you’re being cited.

What changed in 6 months

Google AI Overviews

In January 2026, Google AI Overviews covered about 8% of searches. Today it covers about 22%. Not going back.

The algorithm also changed. Google now prioritizes:

1. Verifiable facts (with dates, citable sources)
2. Clear structure (headings, lists, tables)
3. Authority (backlinks, domain authority still matter)
4. Freshness (updated content ranks better)

Not “which site has most traffic”. But “which site answers the question clearly with sources”.

Real examples:

• Question: “What’s the savings account interest rate in 2026?”

• Google pulls from banks (authority), not blogs.

• Question: “How do I make chocolate cake?”

• Google pulls from sites with structured recipe (Schema Recipe), not just text.

• Question: “What’s the best dental implant?”

• Google pulls from certified clinics, dated posts, reviews.

ChatGPT Search

Launched 3 months ago. Works like this:

1. You type a question in ChatGPT
2. ChatGPT searches the web in real time
3. Shows results with citations

Difference from Google: ChatGPT is more “conversational”. You can follow up. AI understands context.

Example:

• You: “Who is the CEO of OpenAI?”
• ChatGPT: “Sam Altman”
• You: “Where did he study?”
• ChatGPT: “Stanford”

ChatGPT is citing sites, but the user interacts more. If you’re answering simple, factual questions, ChatGPT might pick you up.

Volume: about 5-10% of searches migrated to ChatGPT Search. Not the majority yet, but growing.

Perplexity

Perplexity is the most aggressive of the three. It cites sources, but generously.

Question: “What are the positives of social media?”

Perplexity: “Social media enables global connection, facilitates business…” and cites.

But Perplexity searches blogs, not just authority sites. If you published a post about social media with good SEO, Perplexity will cite it.

Volume: Perplexity grew 40% in 6 months. Still smaller than Google, but aggressive curve.

How each one picks which sites to cite

Google AI Overviews

Google favors:

1. Domain Authority (above 40 is safe)
2. Quality backlinks (citations from important sites)
3. Structured schema markup (Recipe, Article, FAQSchema)
4. Verifiable facts (dates, numbers with sources)
5. Content freshness (recent updates improve ranking)

If you want to be cited in Google AI Overview:

• Structure content with clear H2, H3 headings
• Add comparison tables (Google loves this)
• Include Schema Article + datePublished + dateModified
• Cite your sources (helps Google understand you researched)
• Update old posts (last update date matters)

Example structure that works:

<article>
  <h1>Best toothpastes for sensitivity</h1>
  <p class="published">
    Published: <time datetime="2026-05-01">May 1, 2026</time>
    Updated: <time datetime="2026-05-15">May 15, 2026</time>
  </p>

  <h2>Toothpaste 1: Sensodyne</h2>
  <p>Contains potassium nitrate that reduces sensitivity in 48 hours...</p>
  
  <h2>Toothpaste 2: Colgate Sensitive</h2>
  <p>...</p>

  <table>
    <tr>
      <th>Toothpaste</th>
      <th>Price</th>
      <th>Effectiveness</th>
    </tr>
    ...
  </table>
</article>

Google will pull this for AI Overview because it’s well structured.

ChatGPT Search

ChatGPT favors:

1. Clear, well-written content (ChatGPT understands nuance better)
2. Direct answers (no filler)
3. Logical structure (answer first, then details)
4. Recent updates (similar to Google)

ChatGPT searches like a human would. If your site answers directly, ChatGPT will cite it.

Difference: ChatGPT values Domain Authority less. A new blog can be cited if the answer is good.

Example: you have a development blog, 1 year old. New post about “How to use embeddings in Supabase”. ChatGPT will cite you because the answer is good and clear, even with few backlinks.

Perplexity

Perplexity favors:

1. Traditional SEO (keywords, readability)
2. Recency (new posts rank higher)
3. Simplicity (less jargon)

Perplexity is the most “democratic” of the three. A small blog has a real chance of being cited.

Downside: Perplexity has fewer users still. But growing.

How to measure if your site gets cited

Setup Google Search Console

Configure Search Console to see “AI Overviews” as a metric.

Problem: Google doesn’t have a direct metric yet. You need to:

1. Search your content in Google Search (incognito)
2. See if it appears in an AI Overview
3. Note how many times it appears

I did this for 1 month with 10 keywords. Of 10 keywords, my site appeared in 4 AI Overviews. Rate: 40%.

Better metric: compare traffic 3 months ago with now. If it dropped, Google AI Overview might be “stealing” clicks.

Setup ChatGPT Insights

If you have a site with Analytics, integrate with ChatGPT.com. OpenAI offers a simple dashboard showing how many times your site was cited.

(Note: this feature is new, may not be available to everyone yet in May 2026)

Track Perplexity manually

Do about 20 searches related to your niche on Perplexity. See how many times your site appears in citations.

Record in a spreadsheet:

Query | Perplexity cited my site?
"web development 2026" | YES
"advanced react hooks" | NO
"custom typescript types" | YES
...

Approximate rate: positive citations / total queries.

The bad side: traffic dropping

Here’s the uncomfortable truth. Sites that had 30% traffic from “long-tail” saw it drop.

Before: “What’s the best way to learn React?” brought 200 visitors/month. Now: Google AI Overview answers it directly. Same query brings 40 visitors/month.

That drop isn’t your fault. It’s market change.

How to adapt:

1. Focus on questions that don’t fit in AI Overview (opinions, guides, nuance)
2. Attract via social (not just search)
3. Create content that complements the AI answer (go deeper)

Example:

• AI Overview answers: “What’s React basic syntax?”
• Your post: “50 React patterns nobody teaches you”

Google might cite the first. But who wants to learn React seriously, clicks the second.

The good side: more direct traffic

Paradox: even with AI Overviews, some sites gain traffic.

How? Being cited in the AI Overview with clickable links.

Google favors authority sites. If you’re being cited (with a link), users click to see the full article.

I saw a site go from 5k visits/month to 8k after optimizing for AI Overview citations. Why? Because when cited, users click and consume the full article.

Comparison: Google vs ChatGPT vs Perplexity

Aspect	Google AIO	ChatGPT	Perplexity
Search coverage	22%	5-8%	3-5%
Visible citations	Yes	Yes	Yes
% of users who click	40%	55%	50%
Favors DA	Yes	No	No
Favors freshness	Yes	Yes	Very
Best for	Authority	Quality	Democratic

Next steps: what to do now

1. Audit your top 10 content

   • Is it structured with headings, tables, Schema?
   • Does it have publish and update dates?
   • Does it cite sources?

2. Optimize for citation

   • Add Schema Article to all posts
   • Structure direct answers in H2s
   • Keep content updated

3. Track

   • Monthly check: “Is my site being cited?”
   • Compare traffic before/after

4. Diversify

   • Don’t depend only on search. Everything’s changing.

Linking to the previous post

If you want to understand better how to specifically optimize for AI, see my previous post at Geo optimize site for AI search. I detail Schema, content structure, and benchmarks.

The difference: that post was “how to prepare for the future”. This is “the future arrived, how to adapt”.

Checklist: optimize for GEO 2026

• Top 10 posts have structured Schema Article?
• Each post has clear datePublished and dateModified?
• Content uses H2/H3 headings with direct answers?
• You have comparison tables where it makes sense?
• You cite your data sources?
• You track if you’re being cited in AI Overviews?
• Google Search Console configured and monitoring?
• Domain Authority above 30 (or working on it)?
• You updated main posts in last 30 days?
• Your content competes with authority sites or is complementary?

Read also: GEO: optimize site for AI search | Schema.org in Astro | SEO for devs who hate marketing