AI Overviews Cost 47% of Your Clicks — 7 Actions for SMEs

As of May 2026, Swiss SMEs have been seeing fewer Google visitors for months — and many do not know why. The answer is in three numbers. 47 percent lower click rate on queries with AI Overview (Pew Research, July 2025). 25 percent projected decline in classic search volume by end of 2026 (Gartner). At the same time, plus 393 percent AI-driven traffic in US retail according to Adobe Analytics. Classic SEO alone is no longer enough — the answer is Generative Engine Optimization (GEO), also called AI search optimization or SEO for ChatGPT. This article shows seven concrete Generative Engine Optimization actions SMEs can use to soften the traffic drop and build new visibility in AI search systems.

What the current studies show

Three independent data points:

• Pew Research, July 2025: 8 percent click rate on searches with AI Overview, 15 percent without. A relative drop of 47 percent.
• Gartner forecast: by the end of 2026, classic search engines will lose 25 percent of search volume to AI chatbots and virtual agents.
• Adobe Analytics, Q1/Q2 2026: AI-driven traffic in US retail grew 393 percent year-over-year — and converts 31 to 42 percent better than organic traffic.

Translation for Swiss SMEs: the traffic is not disappearing — it is shifting. Whoever is visible in the new channels benefits.

Why classic SEO alone is no longer enough

Three mechanisms work against classic SEO and AI search optimization gets more important by the day:

• Zero-click searches make up around 69 percent of all Google queries (Similarweb). The answer is right there in the SERP.
• AI Overviews cannibalise position 1: even the best organic rank sits below the AI box.
• AI search systems cite different signals: structured data, entity consistency, fresh content beat pure backlinks. The detailed difference is covered in GEO vs. SEO: what is the difference.

SEO remains necessary, but no longer sufficient. Generative Engine Optimization is the layer on top.

The 7 actions against traffic loss

1. Structure content for chunk retrieval

AI models split web pages into chunks and cite individual ones. 44 percent of all LLM citations come from the first 30 percent of a text. In practice: core answer in the first 200 words, then depth. Each H2 section should stand on its own.

2. FAQ sections with FAQPage schema

AI models pull FAQ answers into citations disproportionately often. Rule of thumb: 4 to 6 questions per page, answers of 40 to 60 words, valid FAQPage schema in JSON-LD.

3. Entity consistency across all platforms

Name, address, phone — identical across website, Google Business Profile, local.ch, moneyhouse.ch, LinkedIn. Plus: `sameAs` links in your Organization schema. Fundamentals are covered in our GEO introduction article.

4. Freshness as a ranking signal

50 percent of AI citations come from content less than 13 weeks old. Recommendation: quarterly refresh cadence for core articles, visible update date, refresh note in the first paragraph.

5. Embed statistics and source attributions

The Princeton GEO study (KDD 2024) shows: two to three numerical statistics per 500 words increase citation rate by up to 41 percent. Sources must be reputable and linked.

6. Expand structured data

Minimum set for SME websites:

• `Organization` with `sameAs` to LinkedIn, Wikidata, Crunchbase
• `Article` with `datePublished` and `author`
• `FAQPage` for FAQ sections
• `BreadcrumbList` for navigation
• `HowTo` for instructional content

7. Third-party visibility

AI models weight multi-source mentions higher than single-domain strength. Concretely: LinkedIn articles in your topic cluster, industry forums, Reddit communities (Perplexity cites Reddit in around 24 percent of answers). For multilingual visibility, see GEO for multilingual websites.

What to tackle first? Prioritisation by effort × impact

Week 1 (immediate, high leverage):

• Add FAQ sections + FAQPage schema
• Check NAP consistency
• Make update dates visible

Week 2 to 4 (medium effort):

• Enrich top 10 articles with statistics and sources
• Extend Organization schema with sameAs
• Rewrite the first 200 words of all top pages

Month 2 to 3 (strategic build):

• Build content clusters around core expertise
• LinkedIn distribution plan
• Set up GEO monitoring

The exact prioritisation depends on the status quo — our free GEO check reveals it within 72 hours.

How do we measure success?

Three KPIs SMEs should track:

• AI Share of Voice: percentage share of brand mentions in ChatGPT, Perplexity and AI Mode answers to industry queries.
• Citation rate: share of your own URLs among cited sources per query.
• AI-driven traffic: sessions in Google Analytics with referrer chat.openai.com, perplexity.ai, gemini.google.com — reported separately.

Honest caveat: measurement in 2026 is still maturing. Manual sampling (20-30 queries weekly) remains standard. Tools like Peec AI, Otterly or Profound automate it step by step. For Swiss specifics, see the Swiss Google AI Mode.

Practice sketch: B2B software vendor in the AI Overviews storm

An illustrative before/after from the Swiss B2B software segment — anonymised, but structurally typical for SaaS vendors with an information-heavy top-of-funnel strategy:

Starting point Q3 2025:

• 65,000 organic sessions per month, around 40 percent from top-of-funnel guides
• Main keywords ranked position 2 to 5 on Google
• AI Overviews appeared on 14 percent of top queries from August 2025
• Within four months: minus 28 percent sessions, minus 41 percent clicks on guide articles

Diagnosis:

The guides answered standard questions ("what is X?", "how does Y work?") and were fully summarised by AI Overviews — no click. Position 2 no longer helped because the user already had the answer in the AI box.

Generative Engine Optimization actions (Q4 2025):

• 30 top guides rewritten for chunk structure — direct answer in the first 200 words, then depth
• 2 to 3 numerical statistics per article added, each with a source link
• FAQPage schema rolled out across all guide pages
• Organization schema extended with `sameAs` to LinkedIn, Crunchbase and G2

Status Q1 2026:

• Citation rate in Google AI Mode for top 50 queries: 34 percent (previously zero)
• AI-driven traffic (chat.openai.com, perplexity.ai, gemini.google.com): 8 percent of total
• AI traffic conversion rate: 4.2 percent (organic: 1.8 percent) — confirms Adobe's 31-42 percent better-conversion observation
• Total sessions still declining (-12 percent YoY), but qualified leads stable

Key insight: a sessions drop is not automatically a revenue drop when the right users come via new channels.

Industry table: typical traffic loss from AI Overviews

Based on BrightEdge, Semrush and Pew Research data 2025/2026 — Swiss market conditions trend slightly earlier and stronger than the US/EU average:

• Informational B2B (SaaS, consulting, fiduciary): typical loss 30 to 55 percent organic clicks on AI Overview queries; citation rate 25 to 40 percent for well-optimised vendors
• Health & Wellness: loss 35 to 50 percent — AI provides standard answers directly; citations favour medical authority (hospitals, professional bodies)
• Finance & Insurance: loss 25 to 45 percent — AI cautious on advice but happy to cite structured comparisons
• Education & Training: loss 40 to 60 percent — comparison queries fully replaced by AI; differentiation via case studies and concrete numbers
• E-commerce (information long-tail): loss 20 to 40 percent — product-specific searches less affected
• Local services (trades, gastronomy, emergency): loss 5 to 15 percent — transactional intents largely remain classic

If you operate in a high-risk industry, the GEO sprint should not be postponed. An SEO agency in Switzerland with GEO profile delivers industry-specific benchmark data.

Quick-win checklist: the first 7 days against AI Overviews

If you want to start fast, go through these seven points in one week:

• Day 1: add an FAQ section with at least 5 questions on your top 3 pages; implement FAQPage schema and validate with Rich Results Test
• Day 2: extend the Organization schema in your site footer with `sameAs` to LinkedIn, Moneyhouse and Wikidata
• Day 3: check NAP consistency — website, Google Business Profile, local.ch, kununu; document differences and harmonise
• Day 4: enrich your five most important articles with two to three numerical statistics plus source links
• Day 5: rewrite the first 200 words of the three most important pages as a direct answer — no branding intro, immediate core message
• Day 6: add "Updated: May 2026" to the first paragraph of all top 10 articles; set up a quarterly refresh plan
• Day 7: sample 25 typical industry queries in Google AI Mode and ChatGPT Search; document citations as baseline

After these seven days the Generative Engine Optimization base hygiene is in place. Deeper actions (content clusters, third-party platforms, monitoring setup) build on this. A GEO agency in Switzerland typically accelerates the first two steps and lets the team focus on the editorial work.

FAQ — AI Overviews and traffic loss

How much traffic does AI Overviews cost?

Pew Research measured minus 47 percent click rate on queries with AI Overview compared to searches without. Industry-dependent traffic loss of 20 to 60 percent is realistic — early Generative Engine Optimization action significantly softens the effect.

What is the difference between SEO and Generative Engine Optimization?

SEO targets the organic results list; Generative Engine Optimization targets the AI answer box above it. SEO works with keywords and backlinks, GEO adds schema data, entity consistency and content structures for LLMs. The two complement each other.

Is SEO still worth it?

Yes. Most SEO levers — technical performance, content quality, backlinks — also work in AI search. Classic SEO is the foundation, Generative Engine Optimization the additional layer. Both together yield maximum visibility.

What can be done about it?

The seven actions in this article: chunk structure, FAQ schema, NAP consistency, content freshness, statistics, structured data, and third-party presence. Priority: quick wins in the first 30 days.

Which industries are most affected?

Informational-heavy industries: legal, tax, health, education, B2B software. AI answers the question fully in the SERP. Transactional industries (trades, local services, e-commerce) are less affected but not immune.

Do we need an agency for this?

Not strictly, but useful for SMEs without dedicated SEO resources. A specialised GEO agency in Switzerland combines AI search optimization with classic SEO. A good first step: free GEO check that reveals status quo and key levers.

How do we measure AI-driven traffic in Google Analytics 4?

Filter in GA4 by session source/medium and include the referrers `chat.openai.com`, `perplexity.ai`, `gemini.google.com`, `copilot.microsoft.com`. Set up a dedicated "AI Search" segment. Important: many AI clicks have no UTM, so referrer filtering is the only option. Adjusting the tracking settings takes about half a day.

Which tools suit an SME budget?

For micro and solo SMEs: Otterly (from USD 29/month) for simple brand monitoring. For SMEs 20-200 employees: Peec AI (from EUR 89/month, EU GDPR-compliant, multi-locale for DE/FR/IT) — clear advantage in the Swiss revDSG context. For mid-market: Profound (from USD 499/month) with enterprise features. Manual sampling remains important in any setup.

Can we compensate lost traffic via other channels?

Partially yes. Three levers: AI search visibility (ChatGPT, Perplexity, Gemini), LinkedIn content in your topic cluster, industry-relevant Reddit communities. Adobe data shows: AI-driven traffic converts 31 to 42 percent better than classic organic — quality often compensates the quantity gap.