Starting June 3, Google split its June core update into an unusual three-phase rollout, expected to finish June 24. Across that three-week window, tracking firms report that 34% of analyzed pages saw significant shifts in their AI-citation rates. Today, June 14, the update sits in Phase Two — the axis is information extractability and citation-worthiness, and most of this week's ranking and visibility volatility comes from here.
For 12–18 months the center of gravity in search has been migrating from blue links to an answer layer. Google made AI Overviews default in 2025 and surfaced AI Mode more broadly in 2026, and the effect is quantifiable: when an AI Overview appears, the former #1 loses about 58% of its clicks and the zero-click rate jumps to roughly 83%; nearly 60% of searches now produce no click at all (see Semrush's zero-click research). Phasing the core update is really Google scoring originality → extractability → worth-including-in-AI-Overview separately — the first time the platform openly forks writing for machines from writing for humans.
The move is not isolated. In June Google also updated Search Console, separating AI-search visibility from traditional organic for the first time, so GEO (generative engine optimization) can finally be measured independently of SEO; meanwhile Profound, Evertune and Scrunch are racing for the who-gets-cited-by-AI dashboard. The whole category is shifting from estimating clicks to estimating citations.
For SMBs the question is not whether to do SEO but which metric to chase and whether to pay for a GEO SaaS. Below: the phase details and numbers, a GEO-tool comparison, and a zero-cost route that needs no subscription.
Event details + full numbers
The phases read roughly like this: Phase One (Jun 3–10) focuses on content originality and expertise signals; Phase Two (Jun 11–17, live) on information extractability and citation-worthiness; Phase Three (Jun 18–24) refines how Google's systems judge whether content is worth including in an AI Overview. Winners: sites whose facts are clean, self-contained and sourced. Under pressure: keyword-stuffed pages, padded long-form, and anything that buries the conclusion in paragraph eight. The number that hurts brand owners most: rankings hold but traffic falls — because the answer is served on the results page and users stop clicking through. The new Search Console exposes this: for the first time you see AI-citation count and organic clicks as two independent curves.
Immediate actions for three audiences
- Brand owners / SMB bosses: stop reading rank alone. Open the new Search Console AI-search report and compare AI visibility against organic clicks; if rank is stable while clicks fall, the answer layer is eating you — move resources from chasing rank to being cited.
- Marketing / SEO operators: rewrite core pages into an answer-first structure — conclusion then evidence per section, authoritative outbound links on key claims, plus FAQ and structured data. Phase Two is re-scoring extractability this week; changes made now can catch the wave.
- Developers / agencies: make FAQPage/Article structured data and clean semantic HTML a delivery default; building citation-worthiness into client templates is a technical-SEO line item you can quote today.
SaaS comparison
| Tool | Focus | Price band | Best fit |
|---|---|---|---|
| Profound | Brand-citation tracking across ChatGPT/Gemini/Perplexity | Enterprise (high monthly) | Funded brands wanting cross-platform citation share |
| Evertune | Analysis of how models lean when describing a brand | Enterprise | Mid-to-large firms wanting to know how AI describes them |
| Scrunch AI | AI-search visibility monitoring + content suggestions | Mid-tier subscription | Small marketing teams doing GEO monitoring |
| Google Search Console (new) | First-party AI-visibility vs organic split report | Free | Everyone — the zero-cost baseline tool |
What nobody tells you
- GEO SaaS mostly sells visibility estimation, not guaranteed citation. The model's citation mechanism is a black box that changes weekly; the tools give trend and relative position, not a lever you can pull. Treat it as a thermometer, not a steering wheel.
- Being cited by AI does not necessarily convert. Users who get the answer in the answer layer may be satisfied and never visit you. Citation rate up while inquiries flat is entirely possible — do not mistake a vanity metric for a business metric.
The no-subscription SMB alternative
A zero-cost GEO route: use the new Search Console (free) for your citation-vs-click split; list 20–30 questions your customers ask most and, weekly, drop them into ChatGPT, Gemini and Perplexity, logging who gets cited and which passage into a Google Sheet; for questions where you are not cited, rewrite the matching page into answer-first + FAQ + structured data. Built in an afternoon, maintained in half an hour a week — and competitive with entry-level SaaS.
FAQ
What does a three-phase core update mean, and which phase matters now?
The June update splits into originality and expertise (Jun 3–10), information extractability and citation-worthiness (Jun 11–17), and AI Overview inclusion assessment (Jun 18–24). Today is Jun 14 — Phase Two — so how cleanly a machine can lift your content as an answer and whether it is worth citing is being re-scored. Most of this week's ranking volatility comes from here.
Do zero-click and AI Overviews really cut traffic that much?
The data is hard. When an AI Overview appears, the former #1 loses about 58% of its clicks and the zero-click rate jumps to roughly 83%; nearly 60% of searches now produce no click at all. Ranking first no longer equals getting traffic — being cited inside the answer is the new placement.
Can an SMB track this without buying a GEO SaaS?
Yes. Use the new Google Search Console AI-search report (it now separates AI visibility from traditional organic) for your own citation picture, then manually test 20–30 core questions in free ChatGPT/Gemini/Perplexity and log who gets cited. That is a good-enough, zero-cost GEO dashboard.
How should I change content to get cited by AI?
Front-load answers and structure facts. Lead each section with the conclusion then expand, attach authoritative source links to key numbers, put common questions in FAQ structure, and embed Article/FAQPage structured data. Machines extract clean, verifiable, self-contained fact units — not marketing adjectives.
My take
The mainstream conclusion will be add structured data fast, do GEO fast. My read is sharper: for 90% of SMBs the rational move is not chasing citation rate but re-doing the whole traffic math — accept that Google organic is a shrinking asset and diversify marketing weight into owned channels the answer layer cannot intercept: email lists, LINE, community. GEO is worth doing, but those treating it as a lifeline will discover in 18 months they merely moved money from an SEO consultant to a GEO SaaS while the traffic ledger did not improve. The ScriptWalker opportunity: instead of selling un-promisable AI-rank gains, package structured data + FAQ + answer-first templates + owned channels (newsletter/LINE OA) into a deliverable, verifiable technical SEO/GEO offer — sell certainty, not mysticism.
Sources
- Google Search Central — Ranking & Updates (official) (first-party)
- Google — Structured Data docs (official) (first-party)
- Google Search Console (official) (first-party)
- Semrush — Zero-Click Searches study (third-party)
- Profound — AI citation tracking tool (third-party)