AI Overviews Took Your Traffic. Here’s How to Take It Back

AI Overviews are not a future concern. They are already reshaping which pages get clicks and which get skipped entirely. If your organic traffic has dropped in the past twelve months and you cannot point to a technical issue or a penalty, AI Overviews are almost certainly part of the explanation. The question worth asking is not whether this is happening, but what you can actually do about it.

Recovering traffic lost to AI Overviews requires a shift in how you think about SEO visibility. Ranking on page one is no longer sufficient if Google answers the question above your result. The strategies that work in 2025 are about earning citation inside AI-generated answers, capturing demand that AI Overviews cannot satisfy, and building content formats that pull readers past the summary and onto your site.

What AI Overviews Actually Do to Your Traffic

Before you can recover lost ground, you need an honest picture of what is happening. AI Overviews appear at the top of Google search results for a growing range of queries, particularly those that are informational in nature. They synthesise answers from multiple sources and present them directly on the results page. For users who get what they need from the overview, there is no reason to click through to any individual site.

The traffic impact is real, but it is not uniform. Queries where the user needs to make a decision, compare options, or take action tend to generate clicks regardless of whether an overview is present. Queries where the user just wants a quick answer are the ones most exposed. If your content was built around answering simple questions efficiently, you are feeling this more acutely than sites built around depth, opinion, or proprietary insight.

I have spent time this year looking at Search Console data across a range of sites in different verticals, and the pattern is consistent: impressions remain relatively stable, but click-through rates on informational queries have declined. That gap between impressions and clicks is where AI Overviews live. Google is showing your content as a reference point but not always sending users to it. Understanding that distinction matters for how you respond.

One important caveat on measurement: your analytics data is giving you a perspective on this problem, not a precise account of it. Search Console counts an impression when your result appears in the results page, but AI Overviews load dynamically and the interaction model is different from a standard ten-blue-links page. The numbers you are looking at are directionally useful, but they are not a complete picture. Treat them as a signal, not a verdict. This is a principle I apply to almost every analytics conversation I have had over twenty years: the tool shows you a version of reality, and your job is to interpret it with appropriate scepticism.

For a broader view of how AI is reshaping search behaviour and what it means for SEO strategy, the Complete SEO Strategy hub covers the full landscape, from technical foundations to content and channel diversification.

How to Get Cited Inside AI Overviews

The most counterintuitive response to AI Overviews is to try to appear inside them rather than around them. When Google’s generative layer cites a source, that citation carries a link. Traffic from AI Overview citations tends to be lower volume than traditional organic traffic, but it comes with a meaningful signal of authority. Being cited tells the user that Google considers your content trustworthy enough to source.

The pages most likely to be cited share a set of characteristics. They have clear authorship with visible credentials. They are structured in a way that makes individual claims easy to extract, with headers, short paragraphs, and direct answers to specific questions. They carry strong E-E-A-T signals, meaning they demonstrate genuine experience and expertise rather than just keyword coverage. And they tend to be on domains that have established topical authority over time.

Structured data plays a role here too, though it is not a guarantee. FAQ schema, HowTo schema, and Article schema all help Google understand the structure and intent of your content. They do not force a citation, but they make it easier for the generative layer to parse what your content is about and how it relates to a given query. Ahrefs has published useful thinking on this in their AI and SEO webinar series, and it is worth reviewing if you have not already.

When I was running iProspect and we were building out content programmes for enterprise clients, the instinct was always to produce volume. More pages, more keywords, more coverage. What the AI Overview era is forcing is a different instinct: fewer pages, but pages that are genuinely better than anything else on a given topic. That is a harder brief to sell internally, but it is the right one.

Which Query Types Are Worth Fighting For

Not every query that triggers an AI Overview is a lost cause, and not every query type deserves the same response. Prioritising where to focus your recovery effort is as important as the tactics themselves.

Informational queries with simple, factual answers are the most exposed. If your content answers “what is X” or “how does Y work” at a surface level, an AI Overview will likely answer it well enough that most users will not click. These pages are worth either upgrading significantly or deprioritising in favour of content that serves a different intent.

Comparison and evaluation queries are more competitive. When a user is trying to decide between two products, two approaches, or two providers, they often want more than a summary. They want specifics, they want to see the reasoning, and they want to trust the source. AI Overviews do appear on these queries, but click-through rates tend to hold up better because the stakes are higher for the user.

Transactional queries, where the user is ready to act, remain largely outside the AI Overview model for now. Google has not moved heavily into generating answers for purchase-intent queries, and the commercial risk of doing so incorrectly is significant. This is where traditional SEO and paid search still deliver reliably.

Local queries are another area where AI Overviews have limited reach. Searches with local intent, “near me” queries, and searches tied to specific geographic areas still resolve primarily through local packs and organic results. If you have a local component to your business, this is worth protecting and investing in. The Semrush guide to SEO traffic generation covers intent segmentation in useful detail if you want a framework for thinking through your query portfolio.

Building Content That AI Cannot Summarise

The most durable long-term strategy is producing content that resists clean summarisation. AI Overviews are good at synthesising publicly available information into a coherent answer. They are not good at replicating original research, genuine opinion, lived experience, or proprietary data. These are the content types that create a reason to click through even when an overview is present.

Original research is the clearest example. If you have conducted a survey, analysed your own data, or produced findings that do not exist elsewhere, an AI Overview cannot reproduce that content. It can reference it, but the user has to come to you to see the full picture. This is not a new idea in content marketing, but it has become significantly more valuable in the AI Overview era.

Opinion and analysis carry similar value. A page that takes a clear, defensible position on a topic, backed by reasoning and experience, offers something that a synthesised summary cannot. When I write about marketing strategy, I am not trying to be comprehensive. I am trying to be specific about what I have seen work and what I have seen fail. That specificity is what makes the content worth reading rather than skimming.

Case studies and examples are underused in this context. Concrete stories about how a specific problem was solved, with enough detail to be genuinely useful, are difficult for AI to replicate because they are grounded in particular circumstances. A case study that names the challenge, the approach, and the outcome is more valuable than a generic explanation of the same concept, and it is harder to summarise without losing the point.

Moz covered this shift well in their MozCon 2025 presentation on stopping traffic loss, and the core argument holds: content that reflects genuine human expertise and perspective is more durable than content that is optimised purely for keyword coverage. The two are not mutually exclusive, but the balance has shifted.

Rethinking What “Traffic Recovery” Actually Means

There is a version of this conversation that frames AI Overviews purely as a problem to be solved, and a correct answer that involves recovering every lost click. I do not think that is the right frame. Some of the traffic that AI Overviews have absorbed was never going to convert. It was surface-level curiosity traffic that landed on informational pages, read for thirty seconds, and left. Losing that traffic is not the same as losing revenue.

The more useful question is which traffic actually mattered, and whether you have a clear enough picture of that to know what to prioritise. In my experience running agencies and managing large performance marketing programmes, the gap between reported traffic and commercially meaningful traffic is almost always wider than clients expect. GA4 tells you a story about sessions and users. It does not tell you which of those sessions were ever going to become customers.

Recovery efforts should be calibrated to business outcomes, not raw traffic numbers. If a category of content was driving newsletter sign-ups, lead form submissions, or product page visits that converted, that is worth recovering. If it was driving traffic that bounced without any meaningful engagement, the loss is less significant than the impression count suggests.

Semrush has published useful analysis on how Google’s AI mode is affecting SEO across different content categories, and the data there reinforces the point that impact is highly uneven by query type and industry. Worth reviewing before you make assumptions about your own exposure.

Diversifying Traffic Sources Without Abandoning SEO

One response to AI Overview disruption that I have seen gain traction is a renewed interest in traffic diversification. Not as a replacement for SEO, but as a hedge against over-dependence on a single channel that is changing in ways you cannot fully control.

Email is the obvious starting point. An email list is an audience you own. It is not subject to algorithm changes, AI Overviews, or ranking fluctuations. If your content programme has not prioritised list building, this is a good moment to start. The mechanics are straightforward: content that is genuinely useful, a clear reason to subscribe, and a consistent publishing cadence. The value compounds over time in a way that search traffic does not.

YouTube is worth considering for content that benefits from demonstration or explanation. Search on YouTube operates differently from Google search, and AI Overviews do not currently appear in YouTube results. If your content covers topics that work well in video format, there is an audience there that is not being intercepted by generative summaries. Mailchimp has a solid primer on YouTube SEO basics if you are starting from scratch.

Paid search remains a reliable channel for capturing high-intent demand, and it is worth revisiting your paid strategy in light of organic changes. If organic traffic on commercial queries has declined, paid search can fill part of that gap. The economics change, but the demand does not disappear.

Direct traffic, meaning users who come to your site because they know you and sought you out, is the most valuable traffic of all and the least discussed in SEO conversations. Building a brand that people remember and return to is not a short-term play, but it is the most durable response to any algorithm change, including the ones we have not seen yet.

Technical Foundations That Support AI Overview Visibility

None of the content strategies above work without solid technical foundations. AI Overviews pull from pages that Google can crawl, index, and parse efficiently. If your site has technical issues that prevent efficient crawling or create ambiguity about what a page is about, you are starting at a disadvantage.

Page speed matters more than it used to, not because it is a direct ranking factor for AI Overview citations, but because slow pages signal poor user experience and Google’s quality assessments take user experience signals into account. Core Web Vitals are worth auditing regularly, not obsessively, but enough to catch regressions before they compound.

Internal linking structure affects how Google understands the relationship between your pages and the depth of your topical coverage. A site with a clear hierarchy, where pillar content links to supporting content and vice versa, signals topical authority more clearly than a flat structure where every page exists independently. This has always been true for SEO, but it matters more when Google is trying to assess whether your site is a reliable source on a given topic.

Structured data is worth auditing across your key content types. If you have FAQ content that is not marked up with FAQ schema, or article content without Article schema, you are leaving signals on the table. Search Engine Land has covered the foundational principles of what harms SEO performance in their analysis of common SEO mistakes, and technical neglect consistently features prominently. The basics are still the basics.

Canonicalisation and duplicate content issues deserve attention too. If Google is uncertain which version of a page to index or cite, it will often default to neither. Clean up any ambiguity in your URL structure, ensure canonical tags are implemented correctly, and make sure that your most important pages are not competing with near-duplicates for the same queries.

Measuring Recovery Without Being Misled by the Numbers

Measuring the impact of your recovery efforts is genuinely difficult, and I want to be direct about that rather than offer a tidy measurement framework that implies more precision than is available. The honest position is that you are working with imperfect signals and making directional judgements rather than precise calculations.

Search Console is your primary diagnostic tool. Track impressions and clicks separately, and look at click-through rate by query type and by page. If your recovery efforts are working, you should see CTR stabilise or improve on the pages and queries you have prioritised. Impressions may not change much, because AI Overviews do not reduce how often your pages appear, they reduce how often users click through after seeing them.

Conversion tracking matters more than traffic tracking in this context. If you are recovering the right traffic, the commercial metrics should reflect it. More newsletter sign-ups, more form submissions, more product page visits from organic search. If traffic is recovering but conversions are not, you may be recovering low-value traffic rather than the traffic that actually matters.

I have been in enough client reviews to know how easily teams get anchored to the wrong metric. Traffic is visible, easy to report, and emotionally satisfying when it goes up. But I have seen traffic grow by 40% while revenue from organic search stayed flat, because the growth was in informational queries that never converted. The number that matters is the one connected to a business outcome, not the one that looks best in a slide.

Track AI Overview appearances for your target queries manually, or use tools that monitor SERP features. Knowing whether an overview is present on your key queries, and whether your content is cited within it, gives you context that Search Console alone cannot provide. This is still a developing area of tooling, but the major platforms are adding this capability progressively.

The complete picture of how to build an SEO programme that holds up through changes like this, from technical foundations to content strategy to measurement, is covered across the articles in the Complete SEO Strategy hub. If you are rethinking your approach in light of AI Overviews, it is worth working through the full set rather than treating this as an isolated tactical problem.

About the Author

Keith Lacy is a marketing strategist and former agency CEO with 20+ years of experience across agency leadership, performance marketing, and commercial strategy. He writes The Marketing Juice to cut through the noise and share what works.

Frequently Asked Questions