Google's AI Overviews rolled out to US search results in May 2024. By Q3 2024 I was looking at Google Search Console data for a content-heavy client site and watching informational traffic drop 35% over three months with no meaningful change in technical health, no algorithm manual action, and no domain authority decline. The site was not penalized. The search results page had simply changed.
The top of the SERP for most informational queries now had an AI-generated summary that answered the question well enough that a significant portion of users never clicked. The client's content was being used to generate that summary. They were getting cited without getting the click.
This article covers what I observed, what I changed, and what is actually working to grow organic traffic in an AI Overviews environment.
What Stopped Working
Informational "what is" and "how to" content at scale. The clearest casualty of AI Overviews is high-volume informational content targeting definition, explanation, and simple how-to queries. If someone searches "what is a conversion rate" or "how to write a cold email subject line," AI Overview answers the question before any organic result is seen. Click-through rate on these queries has dropped 40-70% in the accounts I monitor.
The irony: this was the content type that drove the entire SEO content marketing boom of the 2010s. Long-form, comprehensive guides targeting informational queries were the playbook for domain authority building for a decade. That playbook does not work the same way in 2025.
Generic listicles without first-hand experience. "10 best tools for X" content that reads like it could have been written by anyone without using the tools has been deprioritized in rankings for 18+ months now. Google's Helpful Content updates accelerated this before AI Overviews arrived. AI Overviews accelerated it further by providing synthesis for exactly this content type at zero cost to the user.
Thin comparison pages. Simple competitor comparison pages ("Tool A vs Tool B") with no original data, no real usage, and no differentiated analysis are largely invisible now. AI Overview creates a serviceable comparison from multiple sources. Unless your comparison page has something AI cannot synthesize from public sources, it does not get the click.
The Click Distribution Shift
Here is the pattern I see across 12 client sites with enough organic traffic to produce statistically meaningful data:
| Query Type | CTR Change (2024 vs 2023) | Traffic Change |
|---|---|---|
| Informational ("what is", "how to") | -42% average | Down 35-60% |
| Commercial investigation ("best X", "X vs Y") | -28% average | Down 20-40% |
| Navigational (branded, site-specific) | -5% average | Roughly flat |
| Transactional ("buy X", "X pricing") | +8% average | Up 5-15% |
| Local intent queries | +12% average | Up 8-20% |
The channels hurt most are the ones that drove "awareness" content strategies. The channels least affected, and in some cases growing, are the ones closest to purchase intent.
What Replaced It
Experience-led content that AI cannot replicate. Google's ranking signals now heavily weight first-person, experience-based content that contains original data, personal observation, or tested results. An article that says "I ran 47 cold email subject line tests across three industries and here is what I found, with the specific language that performed" is producing something AI Overview cannot synthesize from public sources. It contains non-public data. Google values this. Users click for it.
This is not the same as adding "I" pronouns to generic content. It means genuinely having done the thing you are writing about and writing from the results, not from research.
Comparison and alternative pages with real switching context. The comparison pages that do work are the ones that address a specific switching scenario: "I was using Tool A, ran into this specific problem, and here is exactly how Tool B solved it and where it falls short." This is a decision-support document, not a feature matrix. AI cannot generate it because it requires lived context.
Data and research pages with original methodology. Original surveys, proprietary data analysis, internal benchmark reports, and research methodologies produce content that gets cited by AI Overviews and drives clicks to the source for the full methodology. Being cited in an AI Overview is not a consolation prize. For the right content, it can drive significant referral traffic from users who want more than the summary.
Bottom-of-funnel transactional and pricing content. Pricing pages, ROI calculators, case studies with specific outcomes, and "is X right for me" pages are performing better than before. Users with purchase intent skip the AI Overview and want specifics. This content type requires domain authority to rank but produces higher-quality traffic when it does.
The New SEO Workflow I Use
My client SEO workflow has shifted in three concrete ways since mid-2024.
Content audit first, new content second. Before writing anything new, I audit existing content for AI Overview exposure. I use Google Search Console to identify which queries have declining CTR despite stable rankings. Those are the AI Overview casualties. For high-value pages, I rewrite them with first-person experience and original data. For low-value pages, I consolidate or remove.
Intent classification for every new piece. Before briefing a piece of content I classify the primary search intent as informational, commercial investigation, transactional, or navigational. Informational content gets a budget only if it has a differentiated experience angle. Commercial investigation content gets built only with switching context or original testing data. Transactional and navigational get prioritized.
Schema markup for everything, especially HowTo and FAQ. AI Overviews pull from structured data. Adding HowTo, FAQ, and Review schema does not prevent AI Overview from using your content. It increases the probability that your site gets cited as a source, which generates brand visibility even when the user does not click through.
{
"@context": "https://schema.org",
"@type": "HowTo",
"name": "How to calculate LTV:CAC ratio",
"step": [
{
"@type": "HowToStep",
"text": "Calculate your average gross margin per customer per month (ARPU * gross margin %)."
},
{
"@type": "HowToStep",
"text": "Divide gross margin per month by monthly churn rate to get LTV."
},
{
"@type": "HowToStep",
"text": "Divide LTV by CAC to get the ratio."
}
]
}
Technical SEO Changes That Matter More Now
Core Web Vitals are table stakes. Sites with poor LCP or CLS are disadvantaged in an environment where users are clicking through less and Google needs high-quality signals to decide which clicks are worth sending. Page speed is not optional.
Internal linking to bottom-of-funnel pages. With informational traffic declining, the internal linking structure that passes PageRank from high-traffic informational pages to conversion-oriented pages has become more important. Every informational page should link to the relevant pricing, case study, or comparison page. Audit this quarterly.
Entity optimization over keyword density. AI Overviews are entity-aware. Content that clearly establishes author expertise (biographical context, linked social profiles, publisher markup) and topic authority (comprehensive coverage of a topic cluster) ranks better for the queries that AI Overview is not fully satisfying. Building entity authority is a longer-term play but increasingly the right investment.
What I Got Wrong
I underestimated the speed of the informational content collapse. In late 2023 I was still advising clients to produce comprehensive informational guides as a primary SEO strategy, with the caveat that they needed to be "high quality." By mid-2024 I was watching those guides decline in traffic regardless of quality because the query type was being answered at the SERP level.
The shift I should have made earlier: prioritize original data and experience-based content over comprehensive coverage. Comprehensive coverage is now the baseline. It does not differentiate. First-hand experience and original data is the differentiator.
I also underestimated how much the Google Discover channel would partially replace informational organic search traffic for the sites that lost it. Three clients who saw significant informational traffic declines recovered 20-30% of the volume through Discover, which surfaces content to users who have not specifically searched for it but match a topical interest profile. Optimizing for Discover (compelling titles, high-quality featured images, E-E-A-T signals) is now part of my SEO workflow for content-heavy sites.
The Boring Reality
SEO in 2025 and 2026 is harder than it was in 2020, requires more original thought, and produces less volume from informational content. The sites that are growing organically are the ones where actual humans with real expertise wrote things they genuinely know. The content machine that produced 50 articles per month using a template and a keyword cluster is not producing the same ROI. The content craftsperson who produces 5 deeply original pieces per month is outperforming it.
This is inconvenient for agencies that built their model on volume. It is good for practitioners who were doing the work properly anyway.