How to Adapt Your SEO Strategy to AI Overview Impacts

What changed when AI Overviews launched in France?

AI Overviews changed the search results page by placing a generated answer and supporting links ahead of many traditional organic results. The underlying pages can remain indexed and ranked, yet a user may get enough of an answer without opening them. That is the shift SEO teams need to understand: a ranking is still an opportunity to be seen, but it is no longer a reliable proxy for a visit.

France gave search marketers a rare clean before-and-after window. Google introduced AI Overviews there on July 22, 2026. Ahrefs then compared Google Search Console data for 963 domains across the 28 days before launch and nine days after it. The study reports that the highly exposed group had median click-through rate fall from 2.29% to 1.76%, a 23.1% relative decline. Ahrefs published the analysis on September 16, 2026.

This does not mean an AI Overview removes every organic opportunity. Google says AI Overviews surface links for people who want to explore a topic further, and the set of linked pages can differ from a classic result page. But an overview can satisfy the first, easy part of a search before a result earns its click. Pages built only to answer a narrow question are therefore more exposed than pages that help a visitor decide, calculate, compare, buy, or complete a task.

The important distinction is between an AI Overview as a search has and a wholesale replacement for search. Google says the has appears only when its systems decide it adds something beyond classic Search. That makes exposure query-specific. A site does not need a single verdict on whether AI Overviews are good or bad. It needs a map of the queries where its existing traffic is vulnerable.

1. Before the launch: median click-through rate for highly exposed domains was 2.29%.
2. After nine days: median click-through rate for highly exposed domains was 1.76%.
3. What to do: compare clicks, impressions, click-through rate, and AI Overview exposure at query level.

Who is most affected by AI Overview impacts?

Sites are most affected when a large share of their important queries trigger an AI Overview and the answer can be consumed without a visit. Ahrefs found the effect was concentrated: 82.2% of highly exposed domains lost click-through rate after the French launch. Across the full panel, the median declines were smaller, at 5.7% for click-through rate and 8.8% for clicks. Exposure matters more than broad anxiety about AI.

Informational pages are often at the front of the queue because they answer questions a reader can resolve quickly. A short definition, a basic instruction, or a simple factual comparison may still create an impression while losing the reason to click. That does not make informational content pointless. It changes the standard. The page needs a clear answer for discovery and a deeper payoff for the visitor who follows the citation.

The risk cuts across business models. B2B software teams can lose traffic from category education, implementation explainers, and comparison queries. Local businesses can lose visits from generic service questions while retaining more intent from searches tied to availability, location, pricing, booking, and specific needs. Publishers can see a growing gap between search visibility and readers arriving on the site.

Do not use vertical alone as a forecast. Ahrefs reported that health was the most exposed sector in its French sample, but the more useful unit of analysis is query intent. A visitor seeking a simple answer may stop at the overview. A visitor assessing a product, reviewing constraints, or preparing to act still has reasons to visit the underlying source.

1. High-risk queries: concise informational questions with a complete answer in the result.
2. Lower-risk queries: needs that require a tool, a transaction, live availability, or detailed evaluation.
3. Priority audience: pages with high impressions, material click loss, and recurring AI Overview triggers.

Why can impressions rise while clicks fall?

Impressions can rise while clicks fall because a page can still be shown in Search while the AI Overview changes what the user does next. Ahrefs observed this pattern in France: rankings broadly held while users clicked less. The study describes an extreme domain with impressions up 183% and clicks down 1.9%. That result is unusual, but the mechanism is not. Visibility records that a result appeared, not that it won attention.

This is why an impressions-only dashboard can mislead an SEO team. A growing impression line may look like success while click-through rate quietly erodes. Rankings can tell you whether Google continues to surface a page. Clicks tell you whether the result still earns a visit. Neither measure tells you whether an AI system has cited the brand when a person asks a related question elsewhere.

Treat the change as a measurement problem before treating it as a content problem. Segment Search Console performance by query theme, page type, country, device, and date. Then add a record of which high-value queries show AI Overviews. Without that layer, a traffic change can be confused with seasonality, a ranking update, demand changes, or an unrelated technical issue.

Google groups traffic from AI has into the existing Search Console Performance report rather than creating a separate AI Overview traffic type. That makes disciplined segmentation even more important. The goal is not to infer a perfect causal number from one weekly chart. The goal is to identify a repeatable pattern: where an overview appears, whether clicks change after it appears, and which pages retain demand.

1. Monitor clicks and click-through rate alongside impressions.
2. Compare affected query groups with comparable groups that do not trigger AI Overviews.
3. Keep a dated record of result-page has changes so the analysis can be repeated.

What should your SEO strategy measure now?

Your SEO strategy should measure exposure, traffic quality, and citation presence rather than relying on average rank alone. Start with a practical inventory of the queries that drive meaningful business outcomes. Mark which queries trigger AI Overviews, which pages rank, which pages are linked in the overview, and how performance changes over time. This turns a broad platform change into a work queue.

Use Search Console to establish the baseline. Export clicks, impressions, average position, and click-through rate before making major content changes. Break the export into query groups that reflect intent rather than only keyword volume. A page about software pricing, for example, should not be bundled with a page that defines the category simply because both contain the same product term.

Then track Citation Share for the questions that matter to your category. Citation Share is the percentage of relevant AI answers in a category that cite you. It is not the same as organic traffic, and it should not be treated as a replacement for conversions. It gives teams a way to see whether their expertise appears when AI systems answer comparison, recommendation, and problem-solving prompts.

A useful measurement framework separates four questions. Are we visible in classic results? Are users clicking when they see us? Do AI systems cite us for important questions? Are those visits and citations connected to qualified actions? This framework avoids the false choice between SEO and AI search visibility. They overlap, but they need different evidence.

Do not create a dashboard with dozens of decorative metrics. Pick a small set of query clusters, establish the current baseline, and review movement on a regular schedule. If a cluster loses clicks while AI Overview exposure grows, investigate page intent and result-page context. If citation presence improves without traffic growth, assess whether the cited queries influence consideration, assisted conversion, or category awareness.

1. Organic visibility: rankings and impressions for priority queries.
2. Traffic response: clicks and click-through rate by query cluster.
3. AI visibility: Citation Share across relevant questions and engines.
4. Business response: qualified signups, calls, bookings, or revenue where attribution is available.

How should you improve pages that AI Overviews affect?

Improve affected pages by making them the best source to cite and the best destination to visit. Those are related goals, but they are not identical. A clean direct answer helps a page serve a reader and gives a system a clear claim to understand. Original analysis, transparent methodology, current details, and useful decision support give the reader a reason to continue beyond the result page.

Start with the page that already earns impressions. Put the answer near the top in plain language, then show the evidence behind it. Replace generic filler with details that a visitor cannot obtain from a short summary: a methodology, a worked example, a comparison that states the criteria, a current process, or a product-specific recommendation with clear limits. Do not manufacture depth by repeating the keyword.

Keep technical SEO solid. Google states that there are no additional technical requirements or special optimizations for appearing in AI Overviews. A page must be indexed and eligible to appear in Google Search with a snippet. Google also says existing SEO fundamentals remain relevant, including helpful and reliable people-first content, internal links, and page experience. The right response is better execution of fundamentals, not a supposed AI Overview loophole.

Freshness needs a purpose. Update a page when the facts, product, market, or underlying source has changed. Add a visible date when it helps a reader assess currency. Cite the primary source for statistics and claims. If the page includes an original analysis, explain the data set, date range, selection criteria, and limits. That makes the content more useful to people and more defensible when another publisher, search feature, or AI answer relies on it.

Do not block content reflexively. Google provides controls such as snippet limits and preview controls, but restricting previews can reduce the material available for Search experiences. Test any restriction against a defined outcome. If a page is essential to acquisition, measure whether the change affects ranking, clicks, citations, and conversions before applying it broadly.

1. Answer the query directly before expanding the explanation.
2. Add source-backed evidence and explain how the evidence was produced.
3. Give visitors a next step that a summary cannot complete.
4. Preserve indexability, snippet eligibility, internal links, and page quality.

How should content planning change after AI Overviews?

Content planning should shift from publishing isolated keyword answers to covering the questions a buyer asks before and during a decision. A page that answers one simple query can still be useful, but it should connect to deeper pages that compare options, explain trade-offs, document implementation, and help a visitor act. The editorial unit is a connected body of evidence, not a pile of loosely related posts.

Build clusters around real question paths. A software category cluster might include a definition, a buyer guide, implementation considerations, pricing questions, security questions, and comparisons. A local-service cluster might cover service scope, location-specific constraints, costs, qualifications, timing, and booking. Each piece should have a specific job, source its factual claims, and link naturally to the next question.

This is where Citation Engineering becomes useful as a framing. The aim is not to game models. It is to create authoritative, current coverage that makes a brand easier to cite when an answer needs evidence. Omnicite measures that outcome as Citation Share across relevant AI answers. The work is quality, coverage, and freshness, not a guarantee of a ranking or a citation count.

Prioritize pages with both strategic importance and demonstrated risk. A page losing clicks on a low-intent query may not deserve the same effort as a page shaping a category decision. Conversely, a cited page with modest direct traffic may matter if it repeatedly introduces the brand in high-intent comparisons. Use the query, the audience, and the business outcome to decide what to improve first.

Keep the publishing standard high. A fast answer needs a source. A comparison needs criteria. A claim about market movement needs a date and a link. This discipline protects credibility when AI answers compress the web into a few visible statements.

1. Map content to decision-stage questions, not only keywords.
2. Link simple answers to pages that support evaluation and action.
3. Prioritize clusters by business impact, traffic risk, and citation opportunity.
4. Retire or rebuild thin pages that give no reason to visit beyond the answer.

What is the practical before-and-after response plan?

The practical response is to establish a baseline before exposure grows, then improve the pages and query clusters where measured performance indicates real loss. The French launch shows why timing matters. Ahrefs had a defined launch date and pre-launch data, allowing it to compare 28 days before AI Overviews with nine days after. Most sites will not receive that clean a market event, but they can still create their own dated baseline today.

First, document the current state of your priority query groups. Record clicks, impressions, click-through rate, average position, pages, locations, and any visible AI Overview trigger. Second, identify the clusters where the page gives an answer but no compelling reason to continue. Third, improve the most important affected pages with clearer answers, evidence, and decision support. Fourth, monitor both organic traffic and citation presence after publication.

Use controlled comparisons where possible. Compare a set of affected informational queries with a similar set that does not show an AI Overview. Do not declare causation from a single page or week. Ahrefs itself notes that exposure is a risk factor rather than a verdict. A strong plan keeps that uncertainty visible while still acting on the measurable signal.

The message for leaders is blunt: more impressions do not automatically mean more opportunity. A search result can be visible and increasingly redundant at the same time. SEO teams that keep measuring only positions will spot the impact late. Teams that measure exposure, clicks, and citations can decide where their work still changes the outcome.

1. Baseline now: export priority-query performance and document result-page features.
2. Diagnose next: identify clusters with AI Overview exposure and falling click-through rate.
3. Improve selectively: add evidence, useful depth, and a clear visitor next step.
4. Review monthly: compare organic performance with Citation Share and business outcomes.

Sources

Source: Ahrefs

AI Overviews launched in France on July 22, 2026. Ahrefs studied 963 domains and reported a 23.1% relative median CTR decline for highly exposed domains after nine days. Ahrefs, 2026-09-16

Source: Google Search Central

Google says there are no additional technical requirements or special optimizations to appear in AI Overviews, and that standard SEO best practices remain relevant. Google Search Central, 2025-12-10