Tag: AI Overviews traffic drop

Your Rankings Didn’t Drop. Your Clicks Did.

You open Search Console. Average position: unchanged. Impressions: up, if anything. Traffic: down 30%.

Nothing is broken. You weren’t penalised. Your content is fine. What happened is that Google started answering the question before anyone reached your page — and most SEO dashboards are not built to show you that.

This is the single most common problem we’re asked to diagnose at AlphabetServ, and it has a name: the AI Overviews traffic drop.

What the Data Actually Says

This is not vibes. Four independent studies, four different methodologies, same direction.

Source	Method	Finding
Seer Interactive	Brand-level GSC data, 3,119 queries, 42 organisations	Organic CTR on AI Overview queries fell from 1.76% to 0.61%
Ahrefs	300,000 keywords, Dec 2025	Position-one CTR down 58% when an AI Overview is present
Pew Research Center	68,879 real searches, 900 US adults	Users clicked a result 8% of the time with an AI Overview vs 15% without
SparkToro / Datos	Clickstream panel, 5M+ users	58.5% of US Google searches already ended without any external click

Ahrefs re-ran its study with December 2025 data and found the presence of an AI Overview correlated with a 58% lower average click-through rate for the top-ranking page. That finding is corroborated by Seer Interactive (organic CTR down 49.4% to 65.2%), Kevin Indig (over 50%) and Authoritas (47.5%).

The Pew study is the most methodologically rigorous of the set because it tracked real browsing behaviour rather than estimating from keyword tools. It found users clicked a traditional result 8% of the time when an AI Overview was present versus 15% without — a 47% relative decline — and only 1% clicked a link inside the AI Overview itself.

One nuance the panic coverage usually omits: the floor appears to have been reached. Seer’s April 2026 update documented organic CTR on AI Overview queries climbing from 1.3% in December 2025 to 2.4% in February 2026 — which Seer describes as levelling off rather than a genuine recovery. The remaining gap between AI-Overview-present queries and AI-Overview-absent queries — roughly 37% — is now the structural baseline to plan around.

Translation: it is not getting dramatically worse. It is also not going back.

Why Your Dashboard Hid It

Standard SEO reporting is built on two numbers: average position and sessions. AI Overviews break the relationship between them.

• Position is unchanged. You still rank third. The AI Overview sits above you and answers the query.
• Impressions may rise. Being included in an AI Overview’s source set can register as an impression. The dashboard reads that as growth.
• CTR collapses quietly. Clicks per impression is the metric that moved, and it’s the one nobody puts on the monthly report.
• The loss is uneven. It concentrates on informational queries. Your branded and transactional terms look healthy, which masks the damage in the blended average.

If you want to see the real picture, segment Search Console by query type and chart CTR over time for informational queries only. That’s where the hole is.

Which Pages Are Losing

Not all content is equally exposed. The pattern is consistent:

Risk level	Content type	Why
Severe	Definitional “what is X” posts	AI answers these completely in-SERP
Severe	How-to and step-by-step guides	Trivially synthesised
High	Generic listicles and comparison posts	AI compiles its own list from several sources
High	FAQ-style content	Literally the format AI Overviews replicate
Moderate	Product and category pages	Commercial intent still drives clicks
Low	First-party case studies with real numbers	Cannot be synthesised — the data exists nowhere else
Low	Proprietary research and original data	Same reason, plus it earns citations
Low	Tools, calculators and interactive assets	Cannot be answered in text

The rule is uncomfortable but simple: if an AI can fully answer the query using publicly available information, that page’s traffic is already gone. Rewriting it more thoroughly does not help. Thoroughness is exactly what makes it easy to summarise.

The Two Things That Actually Work

1. Get cited instead of getting clicked

Being the source an AI Overview names is now a distinct and winnable objective. Seer’s data indicates brands cited within an AI Overview earn meaningfully more organic clicks than uncited brands on the same query — the citation itself functions as an endorsement.

What drives citation:

• Self-contained answers. Put the direct answer in the first 40–60 words under each heading, before the context. AI extracts the first clear, complete statement it finds.
• Specific, attributable figures. “Processing takes 5–15 minutes in typical conditions” gets cited. “Fast processing” does not.
• Structured data. Organization, Article, FAQPage, HowTo and Product schema give machines an unambiguous reading of the page.
• Entity consistency. Your business name, description and category should be identical across your site, LinkedIn, directories and press. Inconsistency makes you a weak entity, and weak entities don’t get cited.
• Crawler access. If GPTBot, ClaudeBot, PerplexityBot and Google-Extended are blocked in robots.txt — often inherited from a template or a plugin default — you are invisible to those systems by your own configuration. Check this first. We find it more often than you’d expect.

2. Rebuild the content mix toward what AI can’t replicate

Shift the publishing budget away from definitional content and toward:

• First-party case studies with real numbers
• Original research and proprietary survey data
• Opinion and analysis tied to named expertise
• Tools, templates and calculators
• Commercial and transactional pages, which retain click intent

This is not “write better content.” It’s writing content that contains information which does not exist elsewhere on the internet.

What This Looks Like in Singapore and Malaysia

Two market-specific notes:

• AI Overview trigger rates vary sharply by vertical. B2B technology and healthcare queries trigger them far more often than local service queries. A Singapore B2B SaaS company is materially more exposed than a KL dental clinic.
• Multilingual queries are less saturated. AI Overview coverage on Bahasa Malaysia and Mandarin queries currently lags English. That gap is a temporary, and closing, advantage for businesses publishing in those languages.

A 30-Day Diagnostic

Before changing anything, measure:

1. Export 16 months of Search Console query data. Segment informational vs branded vs transactional.
2. Chart CTR, not clicks. Identify queries where impressions held or rose while CTR fell 40%+.
3. Check which of those queries trigger an AI Overview. Manual sampling works; a rank tracker with AI Overview flags is faster.
4. Audit robots.txt and meta robots for AI crawler blocks. Confirm GPTBot, ClaudeBot, PerplexityBot, Google-Extended and Bytespider are permitted, or deliberately blocked with a reason.
5. Test your top 20 queries in ChatGPT and Perplexity. Record whether you’re cited, who is cited instead, and what those pages do differently.
6. Score your top pages against the risk table above. You now have a rebuild list ordered by exposure.

That’s your baseline. Everything after it is prioritisation.

The Honest Summary

Traffic is not returning to 2022 levels. Anyone telling you otherwise is selling something.

But the traffic that remains converts better, and the citation channel is genuinely open — most competitors in Singapore and Malaysia have not started optimising for it. The window where “AI SEO” is a differentiator rather than table stakes is open now and will not stay open.

If your rankings are flat and your traffic isn’t, request an audit. We’ll tell you which pages are exposed and what’s recoverable before you spend anything.

FAQ