Google Just Quietly Redrew The Internet: This Week In Marketing

For 25 years, the search box has looked roughly the same. A rectangle. Ten links. A little autocomplete if you were lucky. That was the contract: type a query, get a page, pick a link. Most of marketing was built on top of that contract.

This week, Google quietly tore it up. Gemini 3.5 Flash now powers AI Mode globally. The search box itself has been redesigned for the first time in a quarter of a century. AI Mode just passed one billion monthly active users. And in the same five days, the rest of the industry caught up: Publicis spent $2.2 billion on data infrastructure, Google's own AI search guide told the AEO and GEO industry to calm down, and Kantar reminded us that the world's most valuable brands are worth $13 trillion for one reason only. Trust.

The pivot is no longer about "preparing for" AI search. The pivot has happened. The question is whether your marketing function is designed for the new shape of the internet, or still operating on the old one.

Here is what mattered, why it mattered, and what to do with it on Monday.

For the highlights in 5 minutes, subscribe to Board Analysis, our weekly newsletter for UK marketing leaders. We cut through the noise and tell you what is worth your time, and what is not.

Search & Intelligence: Google ended the AEO and GEO debate. It is still SEO.

Stop paying for new acronyms. Start building for citation.

Google released an official AI Search optimisation guide this week stating, in its own words, that "optimizing for generative AI search is optimizing for the search experience, and thus still SEO." The guide explicitly debunks several industry tactics: llms.txt files, content chunking, AI-specific rewrites, special schema markup and artificial mention-seeking are all unnecessary for Google Search visibility, according to Google. The company recommends focusing on "non-commodity content" rather than generic articles like "7 Tips for First-Time Homebuyers."

The framing matters more than the tactical detail. For the last 18 months, "Answer Engine Optimisation" and "Generative Engine Optimisation" have been sold as new disciplines with new price tags attached. Google's guidance says the same fundamentals apply: good content, clear structure, real expertise, and a brand that the wider web already trusts. If your agency is selling you something different, you are entitled to ask exactly what.

Why this matters:

• The brands winning AI citations are the brands already doing SEO well. If your SEO foundations are weak, no amount of "GEO" will fix the citation problem.
• Non-commodity content is the brief. Anything that could plausibly be written by 200 other sites about the same topic will not get cited.
• Branded mentions across the wider web (PR, reviews, communities) are now load-bearing for AI search visibility, even when no link is involved.

Read more: "Google's New AI Search Guide Calls AEO And GEO 'Still SEO'", Search Engine Journal, 15 May 2026.

Also worth your time this week:

• "Google Quietly Alters Search Terms Reporting For AI Queries In Google Ads" by Brooke Osmundson, Search Engine Journal, 13 May 2026. Google now states that search terms for AI Mode, AI Overviews and Lens "may reflect the inferred meaning or intent behind a search instead of the literal query itself." Negative keyword lists and search-term audits become a lot less precise overnight.
• "SERP FAQ Removal & New Data Challenge Schema's AI Search Value" by Matt G. Southern, Search Engine Journal, 16 May 2026. An Ahrefs study of 1,885 pages found schema additions produced a 2.4% citation lift in Google AI Mode and a statistically significant 4.6% loss in AI Overviews. Schema is not the GEO magic bullet.

Paid Media & Performance: Google launched Ask Advisor. The agent era for marketers is here.

The four-tabs-and-a-spreadsheet workflow is being replaced by a conversation.

At Google Marketing Live 2026, Google introduced Ask Advisor, a Gemini-powered agent that spans Google Ads, Google Analytics, Merchant Center and Marketing Platform. It connects insights, workflows and recommendations across the entire stack. A marketer can ask it to "find new customers for my hair care products" and it will automatically pull product details from Merchant Center, build a campaign in Google Ads, and surface performance back through Analytics, all without switching products.

The biggest open question is how much operational control advertisers will actually delegate to an agent, and how much transparency there will be around recommendations and reporting accuracy. But the direction is set. Ask Advisor is currently in beta for English-language accounts, which gives UK advertisers an early-mover window. Get on it before your competitors do.

Why this matters:

• Campaign setup, optimisation and reporting collapsing into one conversation will reset the "how long it takes" benchmark inside paid media teams.
• The marketers who learn to brief an AI agent well will outperform the ones who keep clicking through menus. New skill set, same goal.
• Beta-only access in English means UK accounts can pilot before global rollout. First in, first to find out where it breaks.

Read more: "Google launches Ask Advisor across Ads, Analytics and Merchant Center", Search Engine Land, 20 May 2026.

Also worth your time this week:

• "Publicis buys LiveRamp for $2.2B to improve AI agent sophistication" by Peter Adams, Marketing Dive, 18 May 2026. A $2.2 billion all-cash deal that gives Publicis access to LiveRamp's 25,000-plus publisher network and 1,300 employees. As Arthur Sadoun put it: "agents built on co-created data learn and improve with every signal, separating them from competitors that train their agents on stagnant, generic data."
• "Why Incrementality Testing Alone Won't Fix Your Paid Media Budget" by Tony Adam, Search Engine Journal, 19 May 2026. A reminder that Marketing Efficiency Ratio (MER), incrementality testing and attribution all answer different questions, and a three-layer measurement framework will outperform any of them alone.

Web Development & Design: Your next visitor might not be human. Build accordingly.

Agentic browsing is coming. Sites that cannot be navigated by an AI agent will lose revenue, not just traffic.

Google's Universal Commerce Protocol is the first production implementation of agent-ready website architecture. It requires a discovery endpoint at /.well-known/ucp, three REST endpoints for checkout (create session, update session, complete session), transport flexibility across REST, Model Context Protocol and Agent-to-Agent, and an open specification. The principles apply far beyond ecommerce: any site that expects to be touched by an AI agent in the next 24 months will need to publish capability manifests, structured actions and machine-readable state.

The cart-abandonment numbers in Slobodan Manic's piece are the clearest way to think about this. Human cart abandonment sits at 70.22% across 50 studies. Estimated agent abandonment on non-UCP sites is roughly 100%, because an agent cannot complete a transaction it cannot understand. The shift mirrors what happened when sites without mobile layouts started losing organic traffic in 2015. Different decade, same dynamic.

Why this matters:

• Discovery endpoints will become the new robots.txt. If your dev team is not aware of UCP, get them aware.
• Every revenue-critical user action on your site (add to basket, request a quote, book a demo) should be re-evaluated for whether an agent could complete it without a human in the loop.
• "Citation gets you into the AI's answer. Discoverable actions get you into the AI's revenue." The next site rebuild brief should reflect both.

Read more: "What Google's UCP Tells Us About Agent-Ready Websites" by Slobodan Manic, Search Engine Journal, 19 May 2026.

Also worth your time this week:

• "Pinterest wants brands to start thinking about Christmas" by Andrew Hutchinson, Social Media Today, 18 May 2026. Pinterest's data shows campaigns running for six months or more deliver 33% higher ROAS than those under three months, and Performance+ catalog sales campaigns using API data deliver 20% lower cost per action. Q4 planning starts in May, not October.

Brand & Connection: $13 trillion of brand value, built on trust not algorithms.

The world's most valuable brands grew 22% in a year while everyone else was busy chasing AI. Read the room.

Kantar's 2026 BrandZ Most Valuable Global Brands ranking topped $13 trillion in combined value this week, up 22% year on year. 93 of the top 100 brands grew their value. Four brands are now worth more than $1 trillion. Google's brand value alone rose 57% to $1.48 trillion, suggesting AI confidence is now baked into brand equity.

The line from Jessica Sibley, CEO of Time, at the Kantar Brand Summit summed up the entire week: "If you don't have trust, you don't have any of it." Every AI search update this week, from Google's AEO/GEO guidance to the rise of agent-led browsing, is converging on the same truth. The algorithms reward brands the wider web already trusts. Brand investment is no longer the soft alternative to performance marketing. It is the system performance marketing runs on.

Why this matters:

• Brand value compounds in an AI era. 93 of the top 100 grew last year. The brands not on that list are losing share of voice in places no one is measuring yet.
• AI confidence is now part of brand equity. Google's 57% rise to $1.48 trillion is partly the market saying "we trust you to lead this." Plan your AI narrative.
• If your 2026 budget cut brand to fund performance, this is the data point to revisit it with.

Read more: "$13tn and a room full of proof: inside Kantar's BrandZ top 100 reveal" by Reuben Webb, The Drum, 18 May 2026.

Also worth your time this week:

• "The oldest truth in marketing, still standing at the Kantar Brand Summit" by Reuben Webb, The Drum, 19 May 2026. 71% of incremental brand value over 20 years came from brands that disrupted their category while staying true to core identity. AI will commoditise performance. Brand judgment is the competitive moat that remains.
• "Social insights fail to reach-decision-makers: Here's what the numbers say" by Sara Karlovitch, Marketing Dive, 15 May 2026. 93% of professionals say social data is important, but only 17% are extremely confident their organisation uses it well. 33% have missed cultural shifts because the insight never made it to leadership.

AI & Automation: Gemini 3.5 Flash is the engine running Google Search.

Faster generation. Deeper answers. Fewer clicks. Plan for the compression.

At Google I/O, Gemini 3.5 Flash was deployed as the default model for AI Mode globally. Google claims it is four times faster than other frontier models on output tokens per second, scores 76.2% on Terminal-Bench 2.1, 83.6% on MCP Atlas, 84.2% on CharXiv Reasoning multimodal understanding, and 1656 Elo on GDPval-AA. The MCP Atlas score is the one to watch: it measures how well a model operates tools, not just how well it answers questions.

For marketers, the implications are practical. Faster generation means deeper, longer AI answers staying inside the user's flow, which compresses click-through opportunities further. A model designed to use tools means agentic browsing is the next mainstream behaviour, not a niche use case. And Google's commercial pitch ("frontier intelligence at Flash speed") is unambiguous: stay inside our ecosystem.

Why this matters:

• Click-through rates on traditional organic listings will continue to compress. Plan content for citation as well as click.
• The MCP Atlas score tells you what to prepare your site for: an AI that uses tools on behalf of a user.
• Faster, deeper AI answers mean longer-tail informational queries are now the front of the funnel, not the bottom.

Read more: "Google Search Now Powered by Gemini 3.5 Flash" by Barry Schwartz, Search Engine Land, 19 May 2026.

Also worth your time this week:

• "YouTube launches digital training initiative in the UK" by Andrew Hutchinson, Social Media Today, 17 May 2026. A YouTube, BBC and National Film and Television School partnership to train UK creators, journalists and producers. The UK creator economy is already worth £2.2bn to GDP.

Case study of the week: +190% visibility for KOREC Group

Trust, authority and findability still come from the same hard work. AI just makes them count more.

When KOREC Group came to us, they had redirect loops, slow page speeds, thin backlinks and duplicate content all working against them. We rebuilt the technical foundations, ran a content audit focused on expertise-led pages, and used digital PR to win the right links to the right pages.

The result speaks to exactly the week we have just had: brand authority and technical fundamentals working together.

• +190% visibility growth
• +19% organic traffic year on year
• +48% goal completions
• -38.5% bounce rate

See the full KOREC case study →

Bringing it together

Every story this week points to the same conclusion. Search, paid, web and brand are no longer separate disciplines. They are one system, and the operating layer is AI.

Google's AI Search guide says rankings are still SEO. UCP says your site needs to be readable by agents as well as humans. Ask Advisor says campaign management is becoming a conversation. Kantar says the trust that underpins all of it is worth $13 trillion and growing. The brands that recognise this convergence and reorganise around it will compound. The ones that keep running 2022 playbooks will quietly lose ground in places they cannot measure yet.

The good news is that the fundamentals have not changed. Be useful. Be specific. Be findable. Be trusted. The bad news, if you want to call it that, is that the standard has gone up. "Good enough" SEO, "good enough" creative and "good enough" data are no longer good enough.

Three takeaways for the week ahead

1. Treat AI Mode as your homepage. With one billion monthly active users and Gemini 3.5 Flash now powering it, write your priority pages so they answer a real question well enough to be cited. Test how AI Mode summarises you and fix what is missing.
2. Audit your site for agent readiness. Whether or not you sell online, agentic browsing is coming. Map your most important user actions, then ask whether an agent could complete them without a human in the loop. If the answer is no, fix it before competitors do.
3. Reinvest in brand, deliberately. The world's most valuable brands grew 22% in a year because trust is now the operating system AI runs on. Set a brand budget. Measure brand health. Protect it from short-termism.

Want help putting any of this into action?

We help ambitious UK brands turn marketing news into measurable revenue.

If you want a clear view of what to do next, book a call with us. We will tell you what is worth your time, and what is not.

Want help putting any of this into action?

We help ambitious UK brands turn marketing news into measurable revenue. If you want a second opinion on your AI search readiness, paid media efficiency or technical SEO foundations, book a call.

This roundup is also published every Friday in our subscriber newsletter, Board Analysis. If you'd rather have it land in your inbox than scroll the website, sign up here.