Search Central Live Barcelona was Google's first three-day Deep Dive event in the EU, held 30 September to 2 October 2026. Gary Illyes, John Mueller, Cherry Prommawin and other Googlers walked through crawling, indexing and ranking, and the message was consistent: AI Overviews and AI Mode run on the same systems as classic Search.
Key takeaways
- There were no product launches. Google used the three days to explain how Search works, step by step, from crawl to ranking.
- Google said Search, AI Overviews and AI Mode use the same main crawler (Googlebot) and the same index. Gemini uses a separate crawler.
- Gary Illyes shared typical processing times for the first time, including about 1.5 hours for end-to-end indexing and 3 to 6 months for core update recovery.
- Quality came up in almost every session. Duy Nguyen framed it as effort, originality, talent or skill, and accuracy.
A quick note on sources before the detail. Google hadn't published slides or recordings when I wrote this, and Q&A sessions were off the record. Everything below comes from Google's own announcement posts plus detailed attendee recaps, mainly John Campbell's day-by-day notes at ROAST. Where a number came from a single attendee, I say so.
What Search Central Live Deep Dive Europe is
Search Central Live is Google's event series for SEOs and developers. The "Deep Dive" version is different: three days, one track, all technical, and it's free but you have to apply.
Google piloted the format in Bangkok in July 2025. For the European edition it asked the community to vote on a city in June 2026, and on 6 July it confirmed Barcelona in a post titled Apparently we're going to Barcelona. According to PPC Land's coverage, Barcelona beat Budapest, Berlin, Frankfurt, Lisbon and Prague.
| Detail | Search Central Live Deep Dive Europe 2026 |
|---|---|
| Dates | 30 Sep to 2 Oct 2026 |
| City | Barcelona, Spain |
| Cost | Free, application only |
| Language | English, no translation |
| Format | One track; crawling (day 1), indexing (day 2), serving and ranking (day 3) |
| Google hosts | Gary Illyes, John Mueller, Cherry Prommawin |
| Community content | 7-minute lightning talks, poster sessions |
The structure matters for how you read the takeaways. Google didn't come to announce features. It came to teach the pipeline, and it said as much on day 1: John Campbell reported the line "No new things to announce just yet."
Day 1 at Search Central Live Barcelona: crawling
The keynote and the AI Mode numbers
Lino Cattaruzzi, President of Google Iberia, opened with where he thinks Search is heading. Two points stood out to me. Google will keep testing features and fold the ones that work into core Search. And "Made for Search" content, the stuff written to rank rather than to help, won't do well.
Then Gary Illyes shared AI Mode growth figures, as reported by ROAST:
- AI Mode queries have doubled every quarter since launch.
- The average AI Mode query is about 3 times longer than a traditional search.
- About 1 in 6 AI Mode searches is multimodal (voice or images).
- Follow-up queries grew more than 40% a month on average in the US.
Longer questions change the job. A five-part question about planning a trip can't be answered by one URL, so Google builds the answer from several pages. That's the reason passage-level clarity matters more now, a point I cover in how to rank in AI Overviews.
One crawler for Search and AI features
Cherry Prommawin and Gary Illyes confirmed that Search, AI Overviews and AI Mode rely on the same main crawler, Googlebot. Gemini uses a separate crawler ("for now at least", per the ROAST recap).
The practical point: if Googlebot can't reach a page, it can't show up in AI Overviews either. There's no separate AI door.
Crawl errors you might not see
Two problems got extra airtime.
CDNs and bot protection now block crawlers more often. A new Cloudflare rule can quietly stop Googlebot, and you'll only see it as HTTP errors in Search Console's crawl stats.
CAPTCHA pages are the sneakier one. A challenge page that returns 200 OK looks fine to your monitoring, but Google sees a page with no content and may treat it as a soft 404.
Dave Smart added a robots.txt detail in his lightning talk that catches a lot of sites: files loaded from another domain (a CDN or subdomain) follow that domain's robots.txt, not yours.
Crawl budget, explained by Google
Cherry Prommawin's crawl budget session was the clearest version I've seen. Crawl budget is crawl rate limit plus crawl demand.
- Crawl rate limit is shared across all Google crawlers hitting your host, including Ads and Images. It drops when connection time or time to first byte rises, or when you return 429 or 5xx codes.
- Crawl demand depends on quality, how often URLs change and how popular they are. For URLs Google doesn't know well, it looks at the parent path.
Errors, useless pages and infinite URL spaces (calendars, parameters) waste it. If you've got pages stuck in Discovered – currently not indexed, server speed is one of the first things to check.
"What would you do?"
The fun session was a game show. John Mueller and Gary Illyes answered crawling scenarios, often differently. For a WAF blocking Googlebot during a flash sale, Gary said to check server usage and fix the WAF, while John said to do nothing. For a leaked staging subdomain with 50,000 crawled URLs, John pointed to the Search Console removals tool.
The lesson I took from it: even Google's own advocates say "it depends." Anyone selling you one-size rules for crawl problems is overselling.
Day 2: indexing, rendering and canonicals
Robots meta tags
John Mueller covered robots meta directives and shared the tag he'd use for maximum visibility:
<meta name="robots" content="max-image-preview:large, max-snippet:-1">
The first part allows large image previews (useful for Discover). The second lets Google show a snippet of any length.
Rendering: the 10,000-pixel viewport
Erin Sparling explained that Google renders pages in a viewport around 10,000 px tall and doesn't scroll or click. So content that only loads on scroll or click may never be seen. In a lightning talk, Rebecca Yu named the most common rendering failure: robots.txt blocking the JavaScript files a page needs.
Main content wins
Gary Illyes said words in the main content area carry more weight than navigation, footers and boilerplate. He also returned to soft 404s: pages that return 200 but are empty or look like errors.
Duplicates and canonicals
John Mueller walked through how Google clusters similar pages, picks a canonical and consolidates signals onto it. His recommendations were plain: redirect during migrations, return proper status codes, check rel=canonical, use hreflang for localisation and keep canonical signals consistent. If you've seen Duplicate without user-selected canonical in Search Console, this is the system behind it.
Structured data
Ryan Levering, a Google engineer, said Schema.org data is cleaned and filtered, then either powers rich results or is passed to AI Overviews and AI Mode as context. His line, per ROAST: "It's never a negative to have structured data." The real question is whether it's worth the effort. He also said shopping schema updates are coming, with server-side validation "coming soon."
International SEO
Cherry Prommawin listed common hreflang mistakes: mixing implementation methods and using wrong language codes. She also said localisation means more than translation, down to date formats and currency.
Google Trends
Omri Weisman, who manages Google Trends engineering, showed that Trending Now updates within about 10 minutes and that Gemini in Explore can suggest and compare up to 8 terms. No public Trends API was announced.
Day 3: ranking, quality and AI features
How a query becomes results
John Mueller explained query understanding: language detection, stop-word removal, entity recognition and synonym expansion. Synonyms depend on context. Canon and Nikon are "siblings", not synonyms, and Google learns that from "X vs Y" searches.
For AI search there's an extra step. The model creates fan-out queries and sends them to Search to ground its answer. These fan-out queries don't come from users, so they don't appear in Search Console.
Gary Illyes then covered retrieval. Language and country narrow the candidates first. In his words, as quoted by ROAST: "But quality is what makes the final decision."
What Google means by quality
Duy Nguyen said Google uses hundreds of signals that vary by result type and runs tests constantly. He pointed people to the public Search Quality Rater Guidelines and summarised quality as four things: effort, originality, talent or skill, and accuracy.
ROAST also recorded this line: "The point is the quality, not who wrote it." Google is judging the page, not whether a person or a model typed it.
Core updates and spam
Gary Illyes made three points worth writing down:
- Core updates evaluate content at page level, not domain level. That's why some pages on your site rise while others fall in the same update.
- Google filters about 40 billion spam pages a day (a figure reported by attendees, not yet in Google's docs).
- Scaled content is now a bigger problem than link spam.
I track every update in my Google core updates history.
Discover
Eric Murillo said Discover has no separate bot and uses the normal index. Content passes an eligibility gate for policy and quality first. Image guidance shared on the day: at least 1,200 px wide, more than 300,000 total pixels, 16:9 ratio and max-image-preview:large. Skip generic logos and text-heavy images.
Search Console and Shopping
Ariel Kroszynski covered recent Search Console features (Groups with regex, AI-assisted report setup and multimodal reporting) but gave no dates for anything new. Alex Jansen introduced "conversational attributes" and product highlights for Merchant data, which help AI Mode answer product questions. His summary: "the bar for quality product data has been raised."
Google's processing timelines
The most quoted slide of the event. Gary Illyes shared typical times for Google's processes, which ROAST said was the first time Google had shown this data.
| Process | Typical time | Slowest |
|---|---|---|
| End-to-end indexing | About 1.5 hours | Months, or never (quality) |
| Snippet or title update | 1–2 days | Weeks to months |
| Canonical change | 1–3 weeks | Months |
| Manual action removal | 1–2 weeks | 4–6 weeks or longer |
| Site move | 1–3 months | 6 months to 1 year+ |
| Core update recovery | 3–6 months | 6 months to 1 year |
I break these down with what each one means for your fixes in a separate post: how long Google takes to index, recrawl and recover.
The closing message
Gary Illyes closed with three requests: use AI to work faster, use it responsibly because it hallucinates, and don't add to the pile of AI slop. Then the line everyone posted afterwards: "AI on Google is just SEO."
What I'm changing after Search Central Live Barcelona
Here's my short list. Nothing on it is new, which is the point.
- Check the CDN and firewall logs for blocked or challenged Googlebot requests, and make sure challenge pages don't return 200.
- Watch time to first byte and 5xx rates in Crawl Stats, since both cut the crawl rate for every Google crawler on the host.
- Move key content out of scroll- or click-triggered loading.
- Add
max-image-preview:largeandmax-snippet:-1sitewide unless there's a reason not to. - Audit pages for the four quality pillars, starting with effort and originality.
- Stop chasing separate "AI optimisation" tactics that Google doesn't use. Its own AI optimisation guide says it doesn't use llms.txt, and an attendee knowledge base logged the same answer from the event. My take on the wider debate is in GEO vs SEO.
The bigger idea is the one I wrote about in AI on Google is just SEO: Google spent three days showing that the AI layer sits on top of crawling, indexing and quality, not beside them.
Want the full checklist version? My SEO checklist for 2026 covers every item above and more.
Frequently asked questions
When and where was Search Central Live Deep Dive Europe 2026?
It ran from 30 September to 2 October 2026 in Barcelona, Spain. It was Google's first Deep Dive event in the EU, after a pilot in Bangkok in July 2025. Barcelona was picked from a community vote that also included Budapest, Berlin, Frankfurt, Lisbon and Prague.
Did Google announce anything new at Search Central Live Barcelona?
No major product launches. Attendees reported Google saying there was nothing new to announce. The event focused on how crawling, indexing and ranking work, plus processing timelines, Discover image guidance and shopping schema plans.
Do AI Overviews use a different crawler?
No. Google said Search, AI Overviews and AI Mode use the same main crawler, Googlebot, and the same index. Gemini uses a separate crawler. If Googlebot can't access your page, it can't appear in AI Overviews.
How long does Google take to index a page?
Gary Illyes said end-to-end indexing typically takes about 1.5 hours once Google decides to index a page. The slowest cases take months or never happen, usually because Google doesn't consider the page worth indexing.
Are the Search Central Live Barcelona slides available?
Not at the time of writing. Google hadn't published slides or recordings, and Q&A sessions weren't recorded. Attendee recaps from ROAST, Digitaloft and others are the best sources until Google shares more.
.webp)
Connect with Akif on LinkedIn, Instagram and Facebook.
Driven by advanced SEO expertise, deep marketing analytics, high-impact content strategy
With 5+ years of hands-on experience, I specialize in holistic search strategies that don’t just rank—they drive real, measurable business growth. I’ve worked across industries including healthcare, hospitality, legal, e-commerce, and professional services, helping brands dominate their target markets. My approach bridges the gap between raw data and creative execution. Every strategy I build is rooted in rigorous market analysis, structured SEO frameworks, and tailored content ecosystems—no templates, no shortcuts. Whether you’re a single-location brand or scaling across multiple cities, I create data-driven marketing systems designed to compound results and grow with you.
Hit by a Google Update?
Get an update impact audit that shows which pages lost ground, why, and what to fix first.
Book a Free SEO Audit →