AI Overviews Cut Top-Ranking CTR by 58%

• Ahrefs December 2025 study: 300,000 keywords, aggregated Google Search Console data, comparing December 2023 (pre-AI Overviews) against December 2025
• Core finding: position-one CTR fell 58% when an AI Overview was present — from 0.073% to 0.016%
• Earlier Ahrefs measurement (March 2024–March 2025) had found a 34.5% drop; the December 2025 update nearly doubled it — trend is still moving in one direction
• Seer Interactive (September 2025, 25.1 million impressions across 42 organizations): organic CTR down 61%, paid CTR down 68% on queries with AI Overviews — corroborating data from a different methodology
• The damage decays down the SERP but no position escapes: position two dropped 50.8%, position three dropped 46.4% (Ahrefs, December 2025)
• Pew Research behavioral study of real users: only 1% of AI Overview search visits result in a click on a link inside the AI Overview itself — the citation bump is largely an awareness event, not a traffic event
• BrightEdge data: click-through rates (CTR) fell roughly 30% while impressions climbed roughly 49% — GSC dashboards surface rising impressions while traffic quietly bleeds, making the problem easy to miss
• Frame for the reader: this is not a ranking problem visible in the usual tools; it is structural decoupling between rank and reach

Why the impact is uneven — and which queries are already gone

• Query-type breakdown creates a gradient of urgency:
  • Informational queries: 70–80% click loss
  • Commercial queries: 35–45% click loss
  • Transactional queries: 20–30% click loss
  • Branded queries: CTR actually increases roughly 18.7% — the one exception, and only because user intent is already brand-directed
• Non-branded queries lose 19.98% CTR on average when AI Overviews appear (Amsive Digital, 2025) — the baseline toll even for mid-funnel queries
• AI Overviews appear in 48% of all tracked Google queries as of March 2026, up from 31% a year earlier — the scope of affected queries is expanding fast
• Vertical skew (Ahrefs, November 2025): science, health, and "people and society" queries see AI Overviews on a large share of searches; real estate and shopping see them on a small fraction — B2B SaaS and content-heavy brands are squarely in the high-exposure zone
• Business Insider case: organic search traffic fell 55% between April 2022 and April 2025, leading to 21% staff cuts — a real-world consequence at scale, not a hypothetical
• Point to make explicit: the queries brands optimized hardest for — long-form informational, how-to, comparison — are the ones hit first and hardest

How AI Overviews decide which sources to cite — and why rank is a weak predictor

• In early 2026, only roughly 38% of pages cited inside Google's AI summaries also held a traditional top-10 ranking; eight months earlier that figure was roughly 76% (Mission Media / Semrush analysis) — the decoupling is recent and accelerating
• Ahrefs analysis of a large set of queries: only roughly 12% of URLs cited by AI tools overlap with Google's top-10 organic results
• What AI citation actually rewards, per current research:
  • Structured data: 81% of pages cited by AI engines use schema markup (AccuraCast / 2025 study)
  • Third-party earned coverage: roughly 85% of brand mentions inside AI search results come from third-party pages, not the brand's own site
  • Named-author credibility and E-E-A-T signals
  • Answer-first content structure — direct response in the first paragraph, semantic heading hierarchy, HTML tables over images
• Citation instability compounds the problem: roughly 70% of pages cited in AI Overviews change over a two-to-three month window, and those changes are not correlated with traditional organic ranking shifts — a citation earned today is not guaranteed next quarter
• The implication: teams optimizing only for rank are running a strategy that no longer reliably connects to visibility in the channel now handling nearly half of all queries

AI search is not one platform — it is four diverging channels with different citation logic

• Scale context: AI Overviews serves roughly 2 billion people monthly; ChatGPT grew from 400 million weekly active users in February 2025 to 900 million by February 2026; Gemini sits at 750 million monthly active users; these are not niche channels
• Referral share is shifting fast: ChatGPT held 89% of B2B AI referrals roughly eight months ago; by March–April 2026 that share had fallen to 63%, while Claude grew from 1.4% to 18.5% (higoodie.com Wave 2 panel, May 2026) — the dominant platform is no longer dominant alone
• Platform-by-platform citation behavior:
  • Google AI Overviews / Gemini: mirrors traditional index signals, heavy reliance on Reddit, LinkedIn, Wikipedia; rewards schema and structured content already performing in Google's ecosystem; JavaScript rendering means it sees pages other models miss
  • ChatGPT: largest citation surface, least traceable; paraphrases without sourcing unless SearchGPT is active; anything post-training-cutoff requires retrieval-layer presence, not just training-layer familiarity
  • Claude: most conservative citation approach; Anthropic's Citations API (early 2025) reduced source hallucinations in early testing; rewards formal authoritative tone and explicit sourcing in underlying content
  • Perplexity: the outlier — roughly one in three cited links points to a Google top-10 page; the assistant most aligned with traditional SEO signals, but still not identical to it
• Cross-platform divergence: only an estimated 11% of domains are cited by both ChatGPT and Perplexity — a brand can be prominent in one model and invisible in another on the same query
• Sentiment risk: if the most prominent third-party content about a brand is a critical review thread, models performing retrieval-augmented generation will often surface that framing — review management is now a direct input into LLM sentiment

What the brands getting more clicks from AI Overviews are doing differently

• Open with the asymmetry: brands cited in AI Overviews earn 35% more organic clicks and 91% more paid clicks than equally-ranked non-cited brands on the same SERP (Seer Interactive, 2025) — the question is no longer "how do I protect my ranking" but "how do I get inside the summary"
• AI-referred visitors also show stronger downstream behavior: 12% more pages per visit, 23% lower bounce rate compared to non-AI referrals
• Tactics that earn citations, based on current research:
  • Answer-first structure: lead with a 50–70 word direct response to the primary query; AI models extract the most extractable content
  • Complete JSON-LD schema (Organization, FAQPage, Product) — technical hygiene that signals trustworthiness to AI crawlers
  • Outbound citations to authoritative sources within the content — nearly all pages earning AI citations include them
  • Proprietary data and named-author credibility — original research and E-E-A-T signals that third-party models treat as authority markers
  • Content freshness: updated statistics and expanded topic clusters strengthen topical authority signals that AI retrieval layers use
• Readiness gap: only 18% of businesses have optimized for conversational search, despite brands cited in AI Overviews earning 35% more organic clicks than non-cited brands — most of the competitive field has not yet adapted
• Budget signal: a Branch survey of 300 enterprise leaders found 89% say AI-powered search improved their marketing performance in 2025; 65% are dedicating at least 25% of their 2026 marketing budget to AI search optimization — the investment case is already being made at the enterprise level

Why click-through rate is now an incomplete measure of brand visibility in search

• The structural problem: when an AI Overview synthesizes a brand's content without a click, that brand has influenced the user's decision — the influence is real, the session in GA4 is not
• Gartner 2025 B2B buyer survey (n=645–646, August–September 2025): 45% of B2B buyers reported using GenAI during a recent purchase, primarily to gather information on vendors and products — these are not fringe interactions
• The impression/click decoupling: BrightEdge data shows impressions rising while clicks fall — a dashboard that reports on impressions and rankings looks fine while brand presence in AI-generated answers goes unmeasured
• What CTR misses entirely:
  • Whether a brand is named in the AI Overview at all
  • Where in the answer it appears (first mention vs. buried fifth)
  • Whether the framing is positive, neutral, or negative
  • How that treatment compares to competitors on the same query
  • Whether visibility differs across ChatGPT, Gemini, and Claude for the same prompt
• The old proxy (rank → traffic → brand exposure) has three weak links now; each one needs its own measurement

The metrics that replace CTR: share of voice, citation rate, sentiment, and prominence in AI answers

• AI Share of Voice defined: the percentage of AI-generated answers that mention or recommend a brand across a defined set of prompts and models — comparable to SOV in paid media, measurable in the same systematic way
• Example calculation: 100 industry-relevant prompts across three AI models; brand appears in 35 responses, competitor in 50 — AI SOV is 35% vs. 50%
• Benchmarks from AthenaHQ's State of AI Search 2026 report:
  • Average brand mention rate: 17.2%
  • A brand-specific SOV above 30% is strong in a competitive market
  • In highly competitive categories, 5–10% can represent a leadership position
• Citation rate alone is insufficient — prominence matters: first mention in an AI answer carries more weight than appearing fifth in a list; being the recommended option is different from being one of several mentioned
• Sentiment as a first-class metric: appearing in an AI answer negatively is worse than not appearing — negative third-party content that models surface via RAG actively harms brand standing
• Cross-model tracking matters: a brand's visibility, prominence, and sentiment can differ substantially between ChatGPT, Gemini, and Claude on identical prompts — aggregate figures mask per-model gaps
• One-off manual prompting is not a strategy: citation sources shift (roughly 70% of cited pages change over a two-to-three month window), meaning a snapshot is obsolete quickly; systematic, scheduled monitoring is required to see trends and act on them

How to build a monitoring system that tracks AI brand visibility the way SEO teams track rankings

• Start with prompt strategy: craft prompt variations that reflect how real users discover your category — not branded queries, but the informational and commercial queries where AI Overviews displace clicks; run them across ChatGPT, Gemini, and Claude, not just one model
• Audit your current AI visibility gap first:
  • Cross-reference GSC for pages with rising impressions but falling clicks — those are the pages AI Overviews are absorbing
  • Query your top category prompts in each model and note whether your brand appears, where, and with what framing
  • Check for competitor presence on the same prompts
• What systematic monitoring needs to produce, per query and per model, over time:
  • Visibility rate (does the brand appear at all)
  • Prominence (where in the answer, in what context)
  • Sentiment (positive, neutral, negative framing)
  • Share of voice vs. named competitors
• Lettertrace is built for exactly this: open-source, MIT licensed, self-hostable, bring-your-own-key — tracks brand visibility, share of voice, sentiment, and competitor benchmarking across ChatGPT, Claude, Gemini, Google AI Overviews, and Perplexity; automates prompt variation, multi-model querying, and aggregation into trend data; no usage markup, no vendor lock-in, data stored in user-owned Supabase
• Data sovereignty matters here: AI monitoring data captures your brand's competitive position across models — it belongs in your own database, not a vendor's; Lettertrace enforces that by design
• Connect monitoring to action: identify which content earns citations and retroactively apply those structural patterns — answer-first copy, schema, named-author credibility — to under-performing pages; re-run prompts after updates to close the feedback loop
• The parallel to SEO tooling is exact: rank trackers made position monitoring systematic rather than manual; AI visibility tools do the same for brand presence in AI-generated answers — AEO and GEO teams deserve the same quality of tooling SEO teams have had for years