On May 20, 2026, Google reshuffled search in Mountain View. AI Mode is becoming the default search experience, AI Overviews reach 2.5 billion users a month, and Universal Cart is set to replace the classic e-commerce listing this summer. This roadmap shows where things are heading, and which 15 measures you as a Swiss SME should put in place now so that your brand gets cited in ChatGPT, Perplexity, Gemini and Google AI Overviews.
At SEOX, we have worked on online visibility in Switzerland for 25 years. What is different in 2026: visibility can no longer be reduced to a single ranking position. A ChatGPT answer, a Perplexity snippet, a Gemini recommendation all run on citation and mention logic that is related to classic SEO but not the same. This article is not a trend scan. It is the roadmap we are using right now to guide more than 40 Swiss brands through the transition.
If you are in a hurry, scroll straight to the 15-point checklist. If you want to understand why it looks the way it does, read the next 3 sections. The reasoning behind it matters more than any single to-do item.
What Google actually announced for search at I/O 2026
On May 20, 2026, Sundar Pichai summed up the keynote: “Google Search is AI Search.” Gemini 3.5 Flash is the new default model for AI Mode (12 times faster). AI Mode has more than 1 billion monthly users, AI Overviews 2.5 billion. Universal Cart and Generative UI roll out free for everyone in summer 2026. The redesigned search box accepts text, images, files, videos and Chrome tabs, and is already live worldwide.
Four days before the keynote, the Sistrix briefing had anticipated the market assessment. We checked seven Sistrix claims directly against the Pichai keynote and the list of 100 announcements. Six are confirmed by primary sources. One (Ask Maps) cannot be clearly found in the official recaps. Here is what is confirmed and what is not:
| Sistrix claim | Status | Primary source |
|---|---|---|
| AI Mode 1 billion MAU | ✓ confirmed | blog.google Pichai keynote |
| AI Overviews 2.5 billion MAU | ✓ confirmed | blog.google Pichai keynote |
| Gemini 3.5 Flash new default | ✓ confirmed | List of 100 announcements |
| Ask YouTube (US-only start) | ✓ confirmed | List of 100 announcements |
| Ask Maps | ! not in primary sources | Sistrix secondary source only |
| Universal Cart (summer 2026) | ✓ confirmed | List of 100 announcements |
| Generative UI free (summer 2026) | ✓ confirmed | List of 100 announcements |
Why this check matters to us: in the 48 hours after the keynote, reactions from SEO voices in German-speaking Europe ranged from “nothing significant” to “SEO is dead”. Both miss the mark. Google’s official story is more sober, and more promising for SMEs: GEO, explained in detail, is no longer a discipline for later. It is now required.
The 3 shifts that hit Swiss SMEs now
Three shifts weigh most heavily for SMEs. Search becomes hyperpersonalized, so ranking alone is no longer enough. Two new search channels open up: Ask YouTube and Ask Maps. E-commerce visibility moves from the classic Shopping listing to a Universal Cart across all Google properties.
Shift 1: Hyperpersonalization beats ranking
Google now uses Gmail, Drive, Calendar, Maps and Photos as a context layer. As Pichai put it: “Search is learning more than what someone wants. It is learning who someone is.” For SMEs, this means a single fixed ranking per keyword no longer exists in that form. Your content has to be quotable for a range of personas, not only for an average searcher.
In practice: if you used to have exactly one money page per search intent, in 2026 you need several levels of answer depth on the same page, for example the keynote-level answer, the answer for a senior marketer and the implementation answer. The differences between LLMO and GEO mostly come down to how deep this semantic layering goes.
What many competitors in German-speaking markets underestimate: hyperpersonalization goes beyond demographics and search history. Google now actively combines calendar context (“You have a trip to Lugano next week”), Drive context (“You are working on a B2B pitch”) and Maps context (“You are in Ticino right now”) in real time. For Swiss brands, this means context factors such as industry, region and languages (de-CH, fr-CH, it-CH) become visible in the AI answer, provided your content signals them clearly. If you neglect hreflang tags, local schema markup and region-specific author bylines, you drop out of the context match.
Shift 2: Ask YouTube and Ask Maps become new search channels
Conversational search in videos (Ask YouTube) launches in the US in summer 2026. German-speaking markets will follow later, the same pattern as with AI Mode. Ask Maps was mentioned in the Sistrix briefing but is not confirmed in the official Google recaps we checked. We will verify it directly again when it rolls out in German-speaking markets.
In practice: Swiss SMEs should build a video library (how-to videos, product demos, founder stories) now, not when it launches here. YouTube is already the most-linked source in German-language AI Mode, ahead of Wikipedia. If you have no YouTube presence today, you are not building a reserve for 2027.
Building for 2026 concretely means at least 6 structured videos per service or product line over the next 90 days. Format mix: 2 how-to tutorials (3–5 minutes), 2 behind-the-scenes or founder story clips (1–2 minutes), 2 product demos or comparison videos (5–8 minutes). Upload a transcript file, write a detailed description with the exact keywords from your keyword sheet, and set chapter markers. Ask YouTube parses this metadata. A nice-looking video without a transcript will stay invisible in 2027.
Shift 3: Universal Cart replaces classic e-commerce rankings
With the Universal Commerce Protocol (UCP), Google bundles shopping visibility across Search, Gemini, YouTube and Gmail. Rollout is in summer 2026, starting with a US focus and Google Pay as the backbone.
For Swiss e-commerce, visibility shifts away from organic product page rankings. Product schema, Merchant Center health, GA4 conversion paths and reviews become more important than classic SEO signals on the product page. An LLMO setup can prepare for this before competitors react.
Concrete preparations we are setting up with e-commerce clients right now: product feeds extended with brand, gtin, mpn and identifier_exists. Review schema with aggregateRating and verified review snippets per SKU. Return and shipping policies as structured data, because Universal Cart will embed these fields directly in the AI answer. If you still produce hand-written product copy without schema in 2026, you will lose visibility through missing machine-readable signals, not through Google updates.
Where things are heading: 3 scenarios for 2026–2027
Three scenarios are realistic: conservative (classic SEO remains a 60% traffic source), hybrid (citation rate becomes KPI #1, GEO and SEO carry equal weight) and AI-first (70% of searches run through generative answers). SEOX is clearly betting on scenario B. Here is why.
Scenario A: Classic SEO stays the 60% traffic source
AI Overviews affect the top of the funnel, while classic blue links remain the traffic source for money pages. Discovered Labs data shows a 54% overlap between AIO citations and classic top-10 rankings. If you rank there, you get cited in AIO as well. If you rank outside, you stay invisible, in the AIO slot too.
What speaks for this scenario: AI Overviews have been live in Switzerland for 12 months, and click-through rates to the cited sources are stable in our client dashboards. For our money page clusters, Google Search Console shows no significant loss of clicks compared to 2025. The default search path stays, even as the visible space is being rebuilt.
Probability: medium to high for 2026, dropping clearly in 2027 once AI Mode goes live in Switzerland.
Scenario B: Hybrid, with citation rate as the new KPI (consensus)
SEO foundation plus a GEO layer. KPIs in 2026: visibility index, citation rate and mention rate per platform. Olaf Kopp’s definition (LLM readability plus brand context optimization) becomes standard. In practice, you keep investing in classic SEO (technical, content, backlinks) and add citation optimization: schema stack, answer capsules, FAQ schema, named authors.
What speaks for B: the market in German-speaking Europe has historically trailed the US by 12–18 months. AIO has been live in the US since 2024 and AI Mode since early 2025. AI Mode in German-speaking markets will probably launch in Q3/Q4 2026. This delay gives you a window. If you work through the foundation items now (measures 1–5 below), you will already be quotable when AI Mode launches here. If you don’t, you start catching up only when competitors are already collecting data.
Probability: high. It is also the path we follow by default with our clients from June 2026.
Scenario C: AI-first, classic SERPs become a niche
If AI Mode takes hold in German-speaking markets the way it did in the US (Pichai: “doubling every quarter since launch”), the majority will tip in 2027. If-then logic: as soon as AI Mode adoption in Switzerland passes 40%, brands have to prioritize their citation rate, much like the switch to mobile-first in 2015.
What would speed up C: Google launching AI Mode in German-speaking markets with an aggressive marketing campaign (TV, print, app ads), Universal Cart being switched on in Switzerland without delay, or Apple Intelligence closing a search partner deal with Google around Gemini. Each is possible, none is certain. Planning for C does not mean switching everything to AI now. It means keeping a second set of measures ready that you can activate within 4 weeks once the threshold is reached.
Probability for 2026: low. For 2027/2028: rising sharply if the EU rollout goes to plan.
For the March snapshot of this development, see 5 GEO trends 2026, our earlier take.
The SEOX GEO checklist: 15 measures that count now
15 concrete measures, grouped by effort: 5 foundation items (≤ 1 week), 5 content pattern items (≤ 2 weeks), 4 brand and trust items (≤ 4 weeks) and 1 KPI setup (ongoing). Each item comes with an effort estimate, a result indicator and help for when you get stuck.
Foundation: required, quick (≤ 1 week)
01
Allow AI crawlers in robots.txt
02
Take an AI visibility baseline
03
Check HTML content without JavaScript
04
Build a schema.org stack
05
Named author byline on every article
Content patterns: citation levers (≤ 2 weeks)
06
TL;DR box at the top of every article
07
Answer capsule under every H2/H3
08
FAQ block with FAQPage schema
09
Show an “Updated” date
10
Original statistics in every article
Brand & trust: mention levers (≤ 4 weeks)
11
Build a Wikipedia presence systematically
12
Reddit and Quora presence
13
Consistent NAP and Organization schema
14
3–5 third-party mentions in 90 days
Tracking & KPIs: ongoing
15
Set up a GEO KPI set
If you want to measure how these 15 levers affect your AI visibility, the KPI mechanics are explained in detail in how to measure AI visibility in practice.
GEO tools compared: what you need now
Five tools lead GEO monitoring in German-speaking markets: aimaco.ai (a Swiss provider), Peec AI, Otterly, Superlines and Rankscale. They all cover the same core functions (tracking a prompt set, counting mentions and citations), but they differ in platform coverage, price level and maturity.
| Tool | Coverage | Prompts | Sentiment | API | Price range |
|---|---|---|---|---|---|
| aimaco.ai Swiss | ChatGPT, Perplexity, Gemini, AIO | 500–10,000 / month | Not public | On request | from CHF 79 / month |
| Peec AI | ChatGPT, Perplexity, Gemini, AIO, Claude | up to 500 / month | Yes | Yes | from €99 / month |
| Otterly | ChatGPT, Perplexity, AIO | up to 200 / month | Limited | Beta | from €49 / month |
| Superlines | ChatGPT, Perplexity, Gemini, Claude | up to 1,000 / month | Yes, with trends | Yes | from €149 / month |
| Rankscale | ChatGPT, Perplexity, Gemini, AIO | up to 300 / month | Yes | No | from €79 / month |
Our recommendation: For Swiss SMEs that keep their books in CHF and want Swiss hosting, aimaco.ai is the obvious starting point (CHF 79 / month starter plan, Google Search Console integration, data stays in Switzerland). Internationally, up to EUR 100 / month: Otterly or Rankscale for a first baseline. From EUR 150 / month: Peec AI for the widest coverage including Claude, or Superlines if you need trend analyses over several months. SEOX uses these tools with clients. Disclosure: aimaco.ai is a Swiss partner tool that we actively use; we are not affiliated with the other four.
Platform cheat sheet: what works best where
Each of the four big AI search systems has its own patterns. ChatGPT mostly cites Wikipedia (47.9%), Perplexity cites Reddit (46.7%), Google AIO relies on a mix of Reddit, YouTube, Quora and LinkedIn, and Claude prefers structured list content.
ChatGPT: Wikipedia and factual sentences
Wikipedia accounts for 47.9% of top citations. Allowing OAI-SearchBot is a must. Structurally, short factual sentences (“X is the most important Y”) beat vague wording (“X could help”). Your own llms.txt file and pinning canonical URLs help with entity resolution.
Perplexity: Reddit and live data
Reddit accounts for 46.7% of top citations. Source quality signals (domain authority, recency, clear signals from external reviews) carry more weight than in ChatGPT. Real-time data (dates, “Updated” stamps, RSS feed) clearly increases your chance of being cited.
Google AI Overviews: classic SEO foundation plus schema
54% overlap with the classic top 10. A schema stack is required, and a top-10 ranking remains the ticket in. Mix of cited sources: Reddit (21%), YouTube (18.8%), Quora (14.3%), LinkedIn (13%). If you want in here, you have to be active in several user-generated content channels, not only on your own domain.
Claude: bullet lists and long structured content
Claude is 30% more likely to cite bullet lists (Discovered Labs). Long, structured content with a clear H2/H3 hierarchy is preferred. Allow ClaudeBot, no captcha wall, no JavaScript rendering required.
Concrete 2-week plans per platform that we are running with clients right now: for ChatGPT, a round of brand mention audits plus 3 Wikipedia edit requests. For Perplexity, 4 Reddit posts in subject-specific subreddits plus a wave of FAQ refreshes on the money pages. For AIO, consolidating top-10 rankings for the 20 most important keywords plus validating the schema stack. For Claude, optimizing the list structure of the 5 most important pillar pages plus output format tests via the Anthropic API. Four platforms, four separate levers: a single general GEO plan does not work.
7 common mistakes: what you should NOT do now
With 40+ GEO migrations over the last 9 months, we have collected the typical anti-patterns. Here are the 7 that trip up roadmaps most often, and what you should do instead.
!
1. Rewriting existing content entirely for AI
In 2026, classic SEO still accounts for 60–70% of visibility. If you now rebuild your evergreen top performer from 2024 entirely in an “AI citation style”, you lose your existing rankings, and the AI mentions are not guaranteed. Instead, add new levels of content depth on top (TL;DR box, answer capsules, FAQ schema). Leave the classic body unchanged.
!
2. Publishing AI-generated content without human depth
Lily Ray calls this the “AI slop loop”: LLMs train on AI-generated content, which in turn feeds AI-generated answers. Originality, real cases and personal observations carry more weight in 2026. Instead, use AI as a research tool, but make sure every article has at least one data point of your own, an observation from a case or a direct quote from a real client conversation.
!
3. Blocking everything in robots.txt out of GDPR fears
Many Swiss data protection officers advise caution and block GPTBot and similar bots across the board. That throws away your chance of being cited entirely. Instead, use allow rules that comply with the Swiss Data Protection Act (FADP), with a whitelist of branding pages (about us, services, blog) and a blacklist for sensitive data (login area, customer portal, account settings). That works.
!
4. Optimizing for one platform only
“We’ll focus on ChatGPT first” sounds pragmatic, but it is wrong. The four big AI systems have structurally different citation logic: Wikipedia for ChatGPT, Reddit for Perplexity, the schema stack for AIO, lists for Claude. If you target only one platform, you give up 75% of the reach. Instead, work through the platform cheat sheet (see above) across all platforms.
!
5. Deploying schema markup without validation
Onely shows that only 22% of sites manage an error-free schema stack. Wrong @type values, missing required fields, broken sameAs links: small errors block rich results completely. Instead, run the Google Rich Results Test, the Schema.org validator and the Search Console schema report before every deploy.
!
6. Starting brand mention tracking only after 12 weeks
If you start without a baseline, you cannot measure the effect of your measures, or defend the budget. The baseline (measure 2 in the checklist) is the required first step before anything else. Instead: day 1 = baseline, day 7 = first re-check, then weekly.
!
7. Treating AI visibility as an IT project
Schema, robots.txt, llms.txt: it sounds like technical SEO and often gets handed to IT or developers. But GEO combines content marketing, brand building and technical SEO. Instead, work cross-functionally: marketing, SEO, content and development together, with a clear weekly sync.
The 90-day roadmap: a weekly plan for your GEO setup
The 15 measures from the checklist above, turned into a concrete 13-week plan. At most 3 focus points per week, so it stays manageable. Suitable for SMEs with at least 0.5 FTE of in-house marketing or SEO capacity.
Weeks 1–2: Foundation
Goal: secure the technical basis so that everything else can be seen at all. Concretely: check robots.txt for AI crawlers (measure 1), take an AI visibility baseline for 20–30 top keywords (measure 2), audit HTML content on your top 10 pages (measure 3). These 3 items take 2–4 person-days and give you the diagnosis for everything that follows.
Output at the end of week 2: a spreadsheet with the current status per platform, keyword and URL, a cleaned-up robots.txt, and a list of JS-only pages with a proposed fix.
Weeks 3–5: Schema stack and named authors
Goal: build trust signals at a machine-readable level. Concretely: roll out the schema.org stack (measure 4), add real author bylines with Person schema (measure 5). Both often need developer support, so book a weekly slot with your IT or web team.
Output at the end of week 5: Rich Results Test passed without errors on all pillar pages, an author page with a linked LinkedIn profile for each real author, and the schema stack built into the CMS template (not only on individual pages).
Weeks 6–8: Content citation patterns
Goal: sharpen existing top performers for AI citations without breaking classic SEO. Concretely: TL;DR boxes on the 10 most important pillar pages (measure 6), answer capsules under the main headings (measure 7), an FAQ schema block on the money pages (measure 8). Showing the updated date (measure 9) runs in parallel.
Output at the end of week 8: 10 pillar pages with TL;DR, answer capsules and FAQ schema, visible date stamps and “Updated” markers in the HTML. A first measurement of the citation lift is possible.
Weeks 9–11: Building brand mentions
Goal: generate external trust signals for the mention rate. Concretely: Wikipedia strategy (measure 11, adapted to your notability status), Reddit/Quora presence with 2–3 strong answers per platform (measure 12), NAP consistency audit on 10+ platforms (measure 13).
Output at the end of week 11: a first third-party mention documented, the NAP audit sheet at 100% target vs. actual, and at least 4 high-quality Reddit/Quora posts live.
Weeks 12–13: PR and KPI setup
Goal: kick off 3–5 third-party mentions (measure 14) and put the KPI set into production (measure 15). First real weekly KPI reports.
Output at the end of week 13: a working KPI dashboard, the first documented third-party mentions, and a re-check of the week 2 baseline with a measurable change on at least 2 platforms.
A note on execution: these 13 weeks are a rhythm, not a rigid plan. If a new competitor suddenly appears in AI search in week 4, you prioritize that, not the schema refactoring. The order above is optimized for the normal case. It is not a doctrine.
If you want external support for the roadmap: the SEOX GEO service starts from CHF 1,800 / month, and a one-off GEO/LLMO audit costs CHF 1,600 (fixed price).
What we are doing at SEOX right now: 5 cases
Five anonymized examples from May 2026: what we are setting up with Swiss clients right now, without revealing any brands.
Case A
Hotel services, tourism sector
AI visibility baseline taken: 0% citation in Perplexity, 12% in ChatGPT. Mix of measures: Reddit strategy for tourism threads, Wikipedia mentions in 3 travel entries, FAQ schema on the 8 top money pages.
First effects expected: 6–8 weeks.
Case B
B2B industrial client, German-speaking Europe
Schema stack fully refreshed (Organization + Person + FAQPage + BreadcrumbList), FAQ block with FAQPage schema live on 12 money pages.
Result after 3 weeks: first AIO mentions in Search Console, mostly on long question keywords.
Case C
Beauty/spa brand, Switzerland
Author bylines linked to LinkedIn put in place (previously: “Editorial team”), 6 articles refreshed with bylines from real people.
Brand mention rate in ChatGPT went from 0 to a consistent 2–3 mentions per test prompt run.
Case D
E-commerce, fashion/lifestyle, German-speaking Europe
Product schema refresh, review schema and Universal Cart preparation. Merchant Center health brought to 98%, GA4 conversion paths cleanly attributed.
First effects expected from summer 2026 (UCP rollout).
Case E
Local trades/services, Switzerland
NAP consistency audit across 14 platforms, Google Business Profile completed, local FAQ schema with “near me” long-tail queries.
Result after 4 weeks: first mentions in local AI Overviews for “near me” queries.
If you want to know where your brand stands today in the four big AI search systems, we check it in our GEO/LLMO audit within 5 days, with a concrete action plan for the next 90 days.
Sources and further reading
Primary (Google I/O 2026):
- Sundar Pichai I/O 2026 Keynote — blog.google
- 100 things we announced at Google I/O 2026 — blog.google
GEO sources from German-speaking Europe (in German):
- Sistrix: “Google I/O 2026: Die Suche wird zum Agenten” (Julia Weissbach, May 20, 2026)
- Olaf Kopp: “Guide to Brand Context Optimization” (February 4, 2026)
Citation patterns (international):
- Surfer SEO: 7 Tips for LLM Citations (Kohli, February 22, 2026)
- Discovered Labs: citation patterns across platforms
- Onely: Ultimate GEO Checklist
More from SEOX: our guide to GEO optimization in 7 steps shows how to put the 15-point checklist into practice, from audience analysis to measuring success in the four big AI search systems.