AI Search
Google's AI Overviews are eating your clicks. Here's how to get quoted instead 🕳️
AI Overviews now take most of the clicks your top rankings used to earn. Here's the answer-first page format that gets you quoted, plus the automation that keeps every page in shape.

Quick answer: to rank in AI Overviews, make each page easy to quote. Put a 2-3 sentence direct answer at the top, follow it with clear question-style headings that each answer one sub-question, add an FAQ section with FAQ markup, and keep the page crawlable and up to date. Google builds AI Overviews from pages that already rank well and answer the question cleanly, so this is SEO with a sharper format, not a separate trick.
You rank number one. Traffic still drops. That’s the new normal.
Ahrefs compared the same queries before and after AI Overviews and found the top-ranking page now gets a 58% lower click-through rate when an Overview sits above it. A year earlier the same study measured 34.5%. The gap is growing, not settling.
So the question isn’t “how do I rank?” any more. It’s “how do I become the source the Overview quotes?”
The idea 💡
Rewrite your top 10 pages answer-first, so an AI can lift a clean, correct answer out of them in one bite.
Why it works 🧠
Two things are true at once.
First, the summary takes the click. Pew Research tracked 900 US adults and found people clicked a regular result on 8% of visits when an AI summary appeared, versus 15% when it didn’t. They also clicked a link inside the summary on just 1% of visits.
Second, Google says it picks the links. Its own guidance on AI features explains that AI Overviews and AI Mode use “query fan-out”: they run several related searches across subtopics, then show supporting pages. There’s no special markup to opt in. The pages that get pulled in are the ones that already meet normal search standards and answer each sub-question clearly.
That second point is the opportunity. If the Overview is going to split your topic into five sub-questions, you want a page with five clean answers waiting.
How to do it ✍️
- Pick your 10 pages. Open Search Console → Performance, filter to the last 3 months, and sort by impressions. Choose pages with high impressions and falling clicks. Those are the ones the Overview is already eating.
- Write the quick answer. Two or three sentences at the very top that fully answer the main search. No warm-up, no “In this article”. If someone read only this, they’d have what they came for.
- Turn your headings into questions. Look at the “People also ask” box for your keyword. Each H2 or H3 becomes one of those questions, and the first sentence under it answers it directly.
- Add an FAQ section. Three or four real questions with short answers, marked up as FAQPage schema. On this site the markup is generated automatically from a “Frequently Asked Questions” heading.
- Prove and date it. Link every number to its source, add your name and role, and show an “updated” date when you revise it. Fresh, sourced, attributed pages are easier to trust.
The short answer wins. That’s what you’re building at the top of every page.
Copy this template for each page:
QUICK ANSWER (2-3 sentences, answers the main search on its own)
## [Main keyword as a question]
First sentence answers it. Then the detail.
## [People also ask question 1]
Direct answer in one sentence. Then 2-4 sentences of support.
## [People also ask question 2]
...
## Frequently Asked Questions
### [Real question]
[Answer in under 50 words]
Updated: [date] · By [name, role]
Automate it ⚙️
Doing this once is a project. Keeping 50 pages in shape every quarter is a workflow. Here’s the one I’d build in n8n:
- Weekly trigger pulls the last 28 days of page data from the Search Console API.
- Flag pages where impressions are steady or up but CTR has dropped more than 30% against the previous period. That pattern usually means an Overview arrived.
- Fetch each flagged page and check it: is there a quick answer in the first 100 words, are the headings questions, is there an FAQ block, when was it last updated?
- An AI step drafts the fixes: a new quick answer and 3-4 FAQs, using only facts already on the page.
- Send the draft to a human in Slack or email for review. Nothing goes live without approval.
The prompt for step 4:
You are editing a B2B web page so it can be quoted in Google AI Overviews.
Main search phrase: {{keyword}}
Page text: {{page_text}}
1. Write a 2-3 sentence quick answer that fully answers the search phrase.
2. Write 4 FAQ questions people genuinely search about this topic, each with an answer under 50 words.
Rules: use only facts that appear in the page text. If a fact is missing, write "NEEDS SOURCE".
Plain words, Australian spelling, no hype.
The “use only facts on the page” rule matters. An AI that invents a statistic is worse than no answer at all.
Map the questions first, then give each one its own clean answer.
If you’d rather not build this yourself, it’s exactly the kind of thing I set up as part of n8n automation or a custom AI agent. For the follow-up side of the funnel, see how to respond to leads in minutes, not days.
Quick wins to try 🧪
- ✂️ Delete the intro paragraph on your top page and start with the answer.
- ❓ Rename one H2 to the exact “People also ask” question.
- 📅 Add a visible “updated” date to your five most-viewed posts.
- 🔗 Replace one unsourced stat with a linked primary source.
- 🤖 Check your robots.txt isn’t blocking Googlebot or Bingbot from key pages.
- 🔎 Search your main keyword in an incognito tab and note who the Overview cites today.
How to measure it 📊
- CTR on the pages you fixed, compared with the same 28 days before the change (Search Console).
- Citations: once a month, search your 10 target phrases and record whether your page is linked in the AI Overview.
- Leads from organic: the number that actually matters. Fewer, better-qualified clicks can still mean more pipeline.
Frequently Asked Questions
How do I get my website into Google AI Overviews?
Rank well for the topic first, then make the page easy to quote: a direct answer at the top, question-style headings, an FAQ section, linked sources and a recent update date. Google says there’s no special markup or opt-in required.
Do AI Overviews reduce website traffic?
Usually, yes. Ahrefs measured a 58% lower click-through rate for the top-ranking page when an AI Overview appears, and Pew found people click regular results about half as often when a summary is shown.
Does FAQ schema help with AI Overviews?
It helps Google understand the page, and it gives the AI clean question-and-answer pairs. It isn’t a guaranteed ticket in, but it costs nothing and supports the answer-first format.
Can I stop Google using my content in AI Overviews?
You can limit snippets with the nosnippet or max-snippet controls, but that also limits how your page appears in normal results. For most businesses, getting cited is the better play.
That’s the move: stop writing pages to be read top to bottom, and start writing them to be quoted.
Sam
If you want this built for you, book a call.
Keep reading: the “Best X for Y” page that ChatGPT keeps quoting, the same answer-first thinking for ChatGPT. Then measure your share of LLM to see if it is working.


