The May 2026 core update finished rolling out on June 2, running about 12 days, with dictionary and utility sites showing the sharpest movement — but you should wait a week before drawing conclusions. In the same week, Google wired Preferred Sources into AI Overviews and AI Mode, and data from 40 major sites showed that AI traffic and AI citations are two completely different measurements.
This issue covers three things only: how the update wrapped up, what Preferred Sources actually changes, and why AI traffic can’t stand in for AI value.
The Three Storylines This Week
Define the three terms first. Every judgment below rests on these definitions.
May 2026 core update: A core algorithm update Google released on May 21, 2026 and completed on June 2, running roughly 12 days. It’s a reallocation of relevance weighting, not a penalty.
Preferred Sources: A Google Search personalization setting where users pick the sites or creators they trust most and want to see first. It went live in AI Overviews and AI Mode on May 27.
AI citations vs AI traffic: AI citations are the URLs an AI system cites, mentions, or indexes inside an answer — even when nobody clicks. AI traffic is the URLs users or AI agents actually click through to. The first shapes discovery and evaluation; the second shapes visits and action.
What the three storylines point to
Content value increasingly depends on whether real users and real sources vouch for it — not on how many keywords a page stacks.
| Storyline | What actually changed | Stage affected |
|---|---|---|
| May core update | Relevance weighting reallocated | Rankings |
| Preferred Sources in AI search | Real user endorsement enters rankings and AI answers | Exposure and trust |
| AI citations ≠ AI traffic | Impact happens before the click | Discovery and evaluation |
May 2026 Core Update: Timeline, Volatility Data, What To Do
Timeline
| Milestone | Time | Notes |
|---|---|---|
| Release | 2026-05-21, 08:40 PT / 11:43 ET | Confirmed via Google Search Status Dashboard and official X |
| Early site volatility | From 2026-05-23 | Ahead of the official announcement |
| Complete | 2026-06-02, 05:40 / 05:43 PT | Roughly 12 days, consistent with the March core update |
Official dashboard: https://status.search.google.com/incidents/wdAXJk6LRRihEjpzEeWE
Volatility data (SISTRIX)
SISTRIX ran a visibility analysis on this update. Source: https://www.sistrix.com/blog/may-2026-core-update-visibility-analysis-and-data-updates/
Key points: the UK reacted sharply, US data lagged, and the rollout released in stages by country.
| Direction | Sites |
|---|---|
| Ranking gains | wiktionary.org, wordhippo.com, wordreference.com |
| Ranking losses | merriam-webster.com, cambridge.org, thesaurus.com |
The dictionary shakeup shows that “homogeneous content + strong brand” no longer guarantees rankings on its own.
What to do
- Don’t rush to edit pages. A core update is usually a reallocation of relevance weighting, not a penalty. Wait at least a week before concluding anything.
- Break volatility down by country and language. Don’t judge from site-wide averages — the UK and US released on different schedules, and mixing them causes misreads.
- Check whether the search intent behind volatile pages matches the landing page. Fix intent mismatch first, not keyword density.
- Record this volatility baseline as the comparison point for the next core update.
Algorithm impact conclusions need a week of observation.
Preferred Sources in AI Search: Mechanics, Three Changes, What It Means for SEO/GEO
Mechanics
Preferred Sources is a personalization setting in Google Search where users designate the sites or creators they trust most and want to see first — essentially a subscription. On May 27, Google’s official blog announced it now feeds AI Overviews and AI Mode.
The eligibility bar is low: any site that consistently publishes new content is eligible to be selected as a Preferred Source. Sites don’t need to enable any subscription feature themselves.
Google shared two data points:
- Users are 2x more likely to click through to a Preferred Source than other sources.
- More than 345,000 unique sources have already been selected by users as Preferred Sources.
Three core changes
| Change | What it is | What it means for sites |
|---|---|---|
| Preferred Sources in AI results | Links from Preferred Sources inside AI answers are explicitly labeled “Preferred” and get more exposure | Being actively subscribed to = a new kind of endorsement weight |
| Timely topic link carousels | AI answers give background first, then show a carousel of timely articles and diverse viewpoints highlighting Preferred Sources; a separate carousel aggregates first-hand views from forums, communities, and social | First-hand sources and community discussion enter AI answers |
| Highly Cited label | Articles cited by many other reports get a label, helping identify original reporting and first-hand sources, and marking which articles cite those highly cited sources | The value of original reporting becomes visible |
What it means for SEO/GEO
I read Preferred Sources as Google building a universal subscription mechanism for the whole web. Whether or not your site has its own subscription feature, Google has effectively built a web-wide subscription entry point for you.
Preferred Sources plus repeated citations look closer to a vote from real users and real reporting, and it’s harder to game than backlinks. Backlinks are increasingly a vote between site owners — and full of manipulated votes. Preferred Sources and citations come from real users and real reporting, and at least for now are harder to manipulate.
Long term, this could gradually replace traditional backlink PR as an evaluation dimension.
For SEO and GEO it’s a turn in the same direction — endorsement weight is rising.
AI Traffic vs AI Citations: Comparison Tables and Action Items
Data source: Aleyda Solis, https://www.aleydasolis.com/en/ai-search/ai-traffic-citations-research/. Underlying data from Semrush, covering 40 major sites across travel, finance, real estate, and ecommerce retail, April 2026, US.
Definition comparison
| Metric | Definition | Stage affected |
|---|---|---|
| AI citations | Which URLs an AI system cites, mentions, or indexes in an answer, even without a click | Discovery / evaluation |
| AI traffic | Which URLs users or AI agents actually visit when a click happens | Site visits + action |
Traffic scale comparison (40 sites combined)
| Channel | Visits | Share |
|---|---|---|
| SEO search traffic | 793 million | 20.45% |
| AI traffic | 7.36 million | 0.19% |
SEO search traffic is roughly 108x AI traffic. The claim that AI traffic is replacing SEO traffic is misleading.
AI traffic vs AI citations distribution (nearly mirrored)
| User journey stage | AI traffic share | AI citations share |
|---|---|---|
| Brand entry (homepage / landing page) | 57.7% | 3.0% |
| Discovery / evaluation (search / category pages / guides / product detail) | 8.9% | 57.0% |
| Action / task (cart / checkout / account) | 19.9% | 1.4% |
| Other deep / non-standard pages | 13.4% | 38.6% |
The pages users click most in AI are homepages, but the pages AI pulls most to build answers are discovery/evaluation pages.
Looking only at AI traffic in GA4 is looking at the tip of the iceberg — 57% of citation impact never shows up in your reports.
Action items
- Don’t dismiss the value of SEO traffic. Its scale is still far above AI traffic.
- Don’t judge AI performance from AI traffic alone. Always read it alongside AI citations.
- Prioritize the pages in the AI citation stage:
- Ecommerce: product data feeds, product detail pages, category pages, reviews, merchant trust signals
- Travel: listings, search, policies, booking data
- Finance: products, rates, support, educational content, plus clear task-flow segmentation
- Real estate: listings, local, entity data, plus cited city and category pages
- Watch the conversion stage. Don’t just publish content and chase citations and exposure while ignoring the back half of the user journey.
This Week’s Action List
- Pull country- and language-level ranking data from May 21 to June 2 and flag volatile pages.
- Run an intent-match check on volatile pages, list the intent mismatches, and fix the top 10 this week.
- Check whether your site meets the conditions to be selected as a Preferred Source: consistent publishing, a clear author or brand, and a subscribable entry point.
- Build a separate “AI citations” tracking sheet outside GA4, logging which URLs and page types get cited in AI answers.
- Compare your industry against the discovery/evaluation page checklist and fill gaps (category pages, guides, product detail, policy pages).
- Tag first-hand reporting and original data to pursue the Highly Cited label.
- Review core update impact after a week and update your baseline.
FAQ
Q1: How do you set up Preferred Sources?
In Google Search personalization settings, users designate the sites or creators they trust most and want to see first. Any site that consistently publishes new content is eligible to be selected as a Preferred Source, and sites don’t need to enable any subscription feature themselves.
Q2: What does Preferred Sources do for SEO?
Users are 2x more likely to click through to a Preferred Source than other sources, and Preferred Source links are labeled “Preferred” in AI Overviews and AI Mode, gaining more exposure. Long term, it looks closer to a real user vote and could gradually replace traditional backlink PR as an evaluation dimension.
Q3: What’s the difference between AI citations and AI traffic?
AI citations are the URLs an AI answer cites or mentions, even without a click, affecting the discovery/evaluation stage. AI traffic is the URLs users or AI agents actually visit, affecting the visit + action stage. The two distributions are nearly mirrored: 57.7% of AI traffic lands on brand entry, while 57.0% of AI citations land on discovery/evaluation pages.
Q4: What should I do if traffic dropped after the core update?
Don’t rush to edit pages. A core update is usually a reallocation of relevance weighting, not a penalty. Wait at least a week, break volatility down by country and language, and fix intent-mismatched pages first rather than stuffing keywords.
Q5: What share of traffic is AI search traffic right now?
Per the Semrush data cited by Aleyda Solis (April 2026, US, 40 major sites across 4 industries), AI traffic is 0.19% and SEO search traffic is 20.45%. SEO search traffic is roughly 108x AI traffic.
Q6: How should I approach GEO optimization?
Prioritize the pages in the AI citation stage: category pages, guides, product detail, policy pages, first-hand data, and original reporting. Also watch the conversion stage — don’t just publish content and chase citations and exposure.
Conclusion
The May core update is complete, running about 12 days, with dictionary and utility sites showing the sharpest movement. Conclusions need a week of observation.
Preferred Sources feeding into AI Overviews and AI Mode amounts to a universal subscription entry point for the whole web, and long term it’s a new kind of endorsement weight.
AI traffic is not AI value: AI citations and AI traffic act at different stages of the user journey, and looking only at AI traffic in GA4 badly underestimates the impact of AI search.
All three storylines point to one thing: content value increasingly depends on whether real users and real sources vouch for it.
Sources and boundaries
- May core update: Google Search Status Dashboard (https://status.search.google.com/incidents/wdAXJk6LRRihEjpzEeWE)
- Volatility data: SISTRIX (https://www.sistrix.com/blog/may-2026-core-update-visibility-analysis-and-data-updates/)
- Preferred Sources: Google official blog, 2026-05-27
- AI traffic vs AI citations: Aleyda Solis (https://www.aleydasolis.com/en/ai-search/ai-traffic-citations-research/), underlying data from Semrush, April 2026, US
- This site’s own traffic and ranking changes were not measured and are not fabricated; algorithm impact conclusions are marked “needs a week of observation.”