Hicham Souilmi

AI Search advertising isn't one channel, it's three very different games.

Fresh data from Similarweb shows just how differently ChatGPT, Google AI Mode, and AI Overview handle ads:

ChatGPT — full conversation context, 1 brand per turn, 56% of ads fire later in the chat

AI Mode — keyword match, 98.5% on Turn 1, multiple ads per response

AI Overview — search query based, shared carousel of product cards

And the brand mix tells the story: HubSpot, Fiverr, Preply and Cursor dominate ChatGPT, while Amazon, eBay, Walmart and Temu own Google's AI surfaces.

Different targeting, different timing, different winning brands.

okay so Wix just dropped a study analyzing 75,000 AI answers and 1M+ citations and the findings are genuinely interesting if you work in SEO or content

the tldr: the #1 predictor of what gets cited in AI search is not the model, not the industry

it's user intent

like. what the person is actually trying to do completely changes what content type gets cited:

— trying to learn something? articles dominate (45% of citations) — comparing options before buying? listicles take 41% of citations — ready to buy right now? product and category pages win — using Perplexity specifically? it cites Reddit and community discussions 2x more than ChatGPT or Google AI Mode

also the listicle thing is wild: 80.9% of listicle citations come from third-party sites. self-promotional "we're #1" lists barely get cited at all. getting featured in someone else's roundup is what actually moves the needle

the takeaway for anyone doing GEO: stop optimizing for the model. start optimizing for what the user is trying to accomplish

Content Types & Formats That Earn Mentions in LLMs How to get cited by AI in 2026:

Tables beat text (2.5x more citations).

FAQ schema is king (3.2x advantage).

Why everyone is wrong about the death of Search.

We've all heard the rumors: "ChatGPT is killing Google." But data is better than anecdotes.

A massive new study analyzing the top 40,000 websites in the US just dropped some truth bombs:

Search isn't dying: Traffic to search engines actually went UP slightly (+0.4%) in 2025.

The "Crash" is a myth: Real traffic decline is -2.5%, compared to the hype of -25%.

Ads aren't winning: Organic results still get 90% of the clicks.

The Lesson: Human memory is unreliable (surveys). Server logs are honest (data). Keep building great content.

How to Rank in AI Overviews: What Actually Works (Based on Data, Not Speculation)

If you've noticed your traffic dipping, you aren't alone. New studies show AI Overviews are eating up to 61% of clicks.

What actually works to get cited?

Answer "Why" questions: These trigger AI results almost 60% of the time.

Cut the fluff: Google's AI prefers concise answers (around 500 words) over long essays.

Get famous: Paradoxically, being mentioned on other sites and YouTube helps your own site rank in AI snapshots more than almost anything else.

🚨 AI Traffic Alert: 1.08% of ALL web visits = 1 in 100 ULTRA-qualified leads

🔥 Fresh 2026 AEO/GEO Benchmarks Report from Conductor 📊 10 industries | 13k+ domains | 17M AI responses | 100M citations 👉 https://www.conductor.com/academy/aeo-geo-benchmarks-report/

The Data Drops:

✅ AI = 1.08% total traffic (IT leads 2.8%) ✅ ChatGPT owns 87.4% referrals (Gemini spikes 21% Utilities) ✅ +1% growth EVERY MONTH ✅ Publishers crushing brands (NerdWallet > banks!)

The Shift: AI = "parallel visibility surface" Brands discovered INSIDE answers before clicks Citations > clicks as the new KPI

AIO Bonus: 25% Google searches trigger AI Overviews Health Care = 48% highest rate

💡 Reality Check AI expands SEO, doesn't replace it Track citations like you track rankings

📋 Enterprise Action Items

AEO platform monitoring

Structured content at scale

Multi-engine benchmarking

SEO + AEO alignment

🚨 AI models give wrong phone numbers 36% of the time

A recent study by Seer Interactive analyzed 178 branded queries across 7 AI platforms and found something concerning.

AI models return a phone number 91% of the time when users ask for contact information.

But only 64% of those numbers match official brand customer service pages.

That means more than a third of AI-generated phone numbers are incorrect.

The risk is real. In one case, ChatGPT surfaced a fraudulent Chase Bank phone number that later spread on Reddit. This isn’t theoretical. It’s already impacting real users.

High-trust industries like financial services, airlines and telecom are especially exposed, with phone numbers returned nearly all the time.

Key takeaway

AI search is not Google.

It’s newer, less protected and far more vulnerable to bad data.

Brands now need to treat AI visibility as a brand safety and customer experience issue, not just an SEO concern.

🔗 Study link

Semrush just dropped fascinating data on how AI search uses Reddit 🔍

Great data & insights from Ahrefs 🔍

Google’s AI Overviews are a fascinating paradox.They change constantly every 2 days on average.

70% of the content shifts.

45% of citations disappear.

Half the entities swap out.

Yet the meaning stays the same (0.95 semantic similarity).

Dynamic on the surface.

Stable underneath.

A perfect snapshot of how AI search actually works.

🌟 Reddit Keyword Research: How to Find Hidden SEO Opportunities

Most keyword tools miss what people really ask online.

Reddit, with its raw and authentic discussions, is a goldmine for discovering:

🧩 Long-tail keywords

💬 Real user pain points

🚀 Unmet content opportunities

This in-depth guide by @semrush shows how to:

1️⃣ Explore niche subreddits where your audience lives

2️⃣ Find recurring questions and struggles

3️⃣ Use Google operators (site:reddit.com [keyword]) to surface hidden threads

4️⃣ Refine keywords with ChatGPT

5️⃣ Cluster and prioritize them using Semrush or Ahrefs

💡 The result: SEO content that speaks your audience’s language — not just search engines.

 E-E-A-T Decoded

The Google Content Warehouse leak, combined with DOJ testimony, reveals how E-E-A-T (Experience, Expertise, Authoritativeness, Trust) is transformed into machine-readable ranking signals.

This isn’t just a theory — it’s a glimpse into how Google actually evaluates content quality, authority, and trust at scale. It’s the blueprint behind what gets rewarded in search, especially for YMYL (Your Money or Your Life) topics.

🔗 Read the full analysis here

🛍️ ChatGPT Shopping is live

700 M users now get product recommendations with images, pricing, and instant checkout 🚀

👉 Read the full Backlinko guide:

📊 Only 14% of the Top 50 Sources Overlap Across ChatGPT, Perplexity & Google AI Overviews

Ahrefs’ Brand Radar analyzed:

🔹 76.7 million AI Overviews

🔹 957,000 ChatGPT prompts

🔹 953,500 Perplexity prompts

Only 7 websites appeared in the Top 50 across all three platforms. That’s right — just seven.

So why the huge difference in citations?

🔸 Google AI Overviews prefer authority sites and Google-owned content

🔸 ChatGPT leans into licensed media partners (think news, entertainment)

🔸 Perplexity pulls from a more diverse, global pool — local and regional sources dominate

💡 The same question to different AIs = completely different sources. Why?

Because behind the scenes, it’s all shaped by search infrastructure, licensing deals, indexing strategies, and platform-specific biases.

This breakdown by @ahrefs is a must-read for anyone in SEO, content strategy, or AI development.

🔗 Read the full article here

🔍 Which websites are AI assistants citing the most?

Ahrefs analyzed 78.6 million AI queries, including:

✅ Google AI Overviews

✅ ChatGPT

✅ Perplexity AI

Top insights 👇

📌 Wikipedia dominates across all platforms

📌 YouTube is a favorite for Google & Perplexity

📌 ChatGPT prefers news sites like Reuters & Apple News

📌 Google highlights user-generated and health content (Reddit, Mayo Clinic)

📌 Each AI assistant has a unique “content personality” — very few overlaps!

📈 Impression Share = citations × Google search volume

→ reveals true brand visibility across AI results.

🧠 If LLMs are the future of search, this is where brand strategy starts.

📎 Source: Ahrefs – “The 10 Most Mentioned Domains…”

🚨 Have AI chatbots replaced search engines?

Not quite.

A study by OneLittleWeb1, using Semrush data, shows that from April 2024 to March 2025:

🔹 ChatGPT and other AI chatbots generated 55.2 billion visits

🔹 Search engines still dominate with 1.63 trillion visits

🔹 Users are still using search engines 34× more than chatbots

Despite massive growth, AI chatbots are still far behind in daily web usage.

🚨 Google’s AI Overviews are reshaping SEO in 2025 — and Semrush just dropped the data.

After analyzing over 10 million keywords, here’s what they found:

🔹 13.14% of all queries now trigger an AI Overview (vs. 6.49% in January)

🔹 88.1% of those queries are informational

🔹 Top impacted industries: Health, Science, Society

🔹 And surprisingly… zero-click rates are not always going up.

📉 It’s no longer just about ranking.

📈 It’s about being the answer.

Whether you’re a content strategist, SEO specialist, or marketer, this shift matters. Google’s turning into a true answer engine—and visibility means earning a place within the AI itself.

🔗 Read the full study by Semrush:

📊 From Search to AI: Why mastering SEO, AEO & GEO matters now more than ever.

Remember when SEO was just about ranking on Google?

Well… those days are over.

Now it’s about showing up in voice search, being the answer to the question, and yes — even training AI models like ChatGPT to recognize your content.

We’re entering a new era:

🔍 SEO = Search visibility

🎙 AEO = Instant answers

🧠 GEO = AI fluency

The smartest strategies mix all three.