The Cannibal Strategy

More people searched Google last week than at any other point in the company's twenty-seven-year history.

Fewer websites benefited from it.

Those two sentences contain the most important strategic fact in marketing right now.

Google I/O 2026 produced a decade's worth of announcements in 48 hours. Most coverage treated it as a product event. It wasn't. It was a strategic declaration and the most revealing detail wasn't in any keynote slide. It was in the number nobody led with. 93% of queries in AI Mode now end without a single click to any website. Position one in organic search earns an 11% click-through rate, down from 27% before AI features became default. Total search volume is at an all-time high. Website traffic is not.

Google is growing. The open web is not. That paradox has a specific name. Clayton Christensen identified it in 1997. He called it the Innovator's Dilemma. Understanding why Google has resolved it (when almost no incumbent in history has managed to) is probably the most important strategic question your marketing function will face this decade.

You need to remember 2024 to understand what happened in May 2026. I know it feels like a world ago, already.

In 2024, Google was being discounted. Not mildly. Genuinely. Perplexity AI was taking search queries from a standing start with no infrastructure and a fraction of the budget. Microsoft had integrated GPT-4 into Bing and was running victory lap press releases about market share. ChatGPT had crossed 100 million users faster than any application in history. The narrative was clear and widely accepted: the search era was ending, and Google had been too slow, too cautious, too wedded to its existing model to lead what came next.

Gemini's early performance gave that narrative oxygen. The image generation controversy. The comparisons to GPT-4 that landed badly. The sense that the company which invented the transformer architecture underpinning every major language model had somehow been outmanoeuvred by a startup it could have acquired for a rounding error on its balance sheet.

The share price reflected the uncertainty. Analysts were asking, for the first time in two decades, whether the search business was structurally impaired.

Google did not panic.

It fixed Gemini. Systematically. Gemini 1.5 outperformed the early models substantially. Gemini 2.0 was competitive with the frontier. Gemini 2.5 was not catching up, it was leading. Meanwhile, search revenue continued to grow throughout what was supposed to be a death spiral.

Not because the threat wasn't real. Because Google understood something that the commentary did not.

The threat wasn't to Google. The threat was to the version of Google that delivered blue links.

Christensen's insight was deceptively simple. Incumbents don't fail because they're incompetent. They fail because they're rational. When a disruptive technology arrives, the rational response is to protect the revenue model that made you. Your customers are happy with the existing product. Your margins depend on it. Your organisational structure was built around it. Every incentive points toward defending what you have rather than cannibalising it before someone else does.

Kodak invented the digital camera in 1975. They filed the patent, looked at the margins on film, and put it in a drawer.

Blockbuster's executives understood streaming. They could see exactly what Netflix was doing (and rejected the idea of buying it when they could). Their own data showed the direction of travel. They chose to stay in the late fees business.

Nokia's engineers were building touchscreen smartphones in 2004. I was at Nokia when the iPhone launched. I remember the reaction from the engineering teams: 'Ha! We know people hate touchscreens.' The rest is now history.

The pattern is consistent. The insight is usually present. The resolve is almost never there.

What makes Google remarkable is not that it understood the Innovator's Dilemma. Every major technology company has read Christensen. What makes Google remarkable is that it internalised it and then acted on it at a scale that should not have been possible for a company of its revenue dependency.

AI Mode generates 93% zero-click results. Google knows this. Google built it anyway. The click-based advertising model that generates approximately $175 billion in annual revenue is being systematically undermined by a product Google designed, built, and deployed deliberately. The $180–190 billion infrastructure investment Sundar Pichai announced at I/O is not being spent to preserve the click. It is being spent to own what comes after it.

That is the most difficult strategic decision a public company can make.

The trade press covered I/O as eight product announcements. It is more useful as an inventory of what Google decided to absorb.

Creative production. AI Brief and Asset Studio with Gemini Omni take a natural-language brief and generate campaign-ready creative variants. What took two days through a design studio now takes hours inside the platform. The brief is the only remaining human input. Vague brief, vague creative. Sharp brief, shippable creative without the production layer in the middle.

Media planning and performance analysis. AI Max replaces keyword architecture with intent-matching AI. Every Dynamic Search Ads account migrates automatically in September 2026 by platform default. Ask Advisor unifies Google Ads, Analytics 4, Merchant Centre, and the full Marketing Platform under a single Gemini agent that surfaces recommendations, identifies problems, and guides campaign decisions across the entire stack without a human moving between platforms. The media analyst, the performance planner, and the cross-platform reporting specialist are not being threatened. They are being absorbed.

Commerce infrastructure. The Universal Cart which is live this summer, launched with Nike, Sephora, Walmart, Target, Ulta Beauty, Wayfair, Fenty Beauty, and Steve Madden as founding partners. It enables consumers to add products from multiple retailers into a single checkout experience without leaving Google's surfaces. Search, YouTube, Gmail. The purchase funnel no longer requires a brand's website. It requires a brand's presence inside Google's commerce infrastructure.

Measurement. Meridian, Google's open-source marketing mix modelling tool, is now native inside Google Analytics 360. The timing is precise: as AI Mode reduces click-path data and Universal Cart removes website visits from the funnel, the only credible framework that captures the full picture is MMM and Google has now made itself the most accessible provider of it.

Each of these was previously a service provided by an agency, a martech vendor, or an independent consultancy. Each is now native platform infrastructure. Infrastructure is not optional. Infrastructure is not replaced by a competitor in a restructuring cycle. Infrastructure is the floor on which everything else is built and Google just poured it.

Here is where the Innovator's Dilemma becomes something else entirely.

When Christensen described disruption, the mechanism was consistent: the new entrant takes the market, the incumbent loses it. Two companies, one outcome. What Google is constructing is structurally different. It isn't disrupting its competitors. It is mediating the relationship between its customers (brands) and their customers (consumers). At both ends. Simultaneously.

Gemini Spark, Google's always-on personal AI agent announced at I/O and available to AI Ultra subscribers from the week of May 25, manages the consumer's inbox, researches purchases on their behalf, compares products, confirms compatibility, surfaces promotions, and executes checkout through Universal Cart, inside Google's surfaces, before the consumer has visited a single brand website. The consumer has delegated discovery, consideration, and purchase to an agent.

AI Max generates the brand's creative, identifies the audience, optimises the bid, and manages campaign performance. The brand has delegated acquisition to an agent.

Both agents belong to the same platform.

The relationship between the brand and the consumer is now mediated at both ends by Google. What used to be a conversation between a brand and a human is becoming an automated exchange between two Google systems, with the brand and the consumer as downstream participants rather than active ones.

This has a specific consequence that most I/O coverage did not articulate.

The consideration phase made up of the research, the comparison and the story a brand told across multiple touchpoints before a human made a decision was where brand equity was actually built. It was where challenger brands disrupted incumbents by being more compelling at the point of attention. It was where premium brands justified their prices by being more convincing during evaluation. That phase is being compressed, sometimes eliminated, by agents that have already formed a recommendation before the consumer is consciously involved in the process.

If your brand is not in the AI's consideration set, you are not in the consumer's consideration set.

The distinction that used to separate brand strategy from performance marketing has effectively dissolved.

Alongside Anthropic, the only other organisation in this space demonstrating both genuine frontier capability and long-term strategic seriousness, Google is the most interesting business operating in AI right now. Not because of the announcements at I/O. Because of the judgment required to make them in that sequence, from the position they were in eighteen months ago. When the critics were writing the search era's obituary, Google was quietly building what search becomes.

Most companies understand Christensen. They cite the examples. They reference the framework in board presentations.

Google did the thing.

The question your digital marketing function is currently optimised to answer, which is how do I rank in search results, how do I acquire clicks, how do I drive traffic to my website is the 2019 question. The 2026 question is different, and most organisations do not yet have a team structured to answer it.

How does my brand get understood, trusted, and selected by an AI agent making decisions on behalf of humans I am no longer reaching directly?

That is a brand authority question, not a performance marketing question. It is answered by the quality and specificity of your product data, by the clarity of your positioning relative to the signals AI agents use to evaluate category fit, by what appears in AI-generated summaries when someone describes a problem your category solves. It is not answered by bidding higher on the keywords you have always bid on.

Google held a two-day conference last week. The trade press covered it as a product launch.

It was a map.

The brands that read it as a map and start building toward the territory it describes will be in the agent's consideration set when the decisions that matter are being made. The ones that read it as a keynote and wait for the next one will discover, sometime in 2027, that they were excluded from decisions made without them.

The click was always borrowed time.

Google just called it in.

Brandflow is written by Justin Billingsley, who has spent his career on all three sides of the industry's table: senior client, global agency leader, technology founder. First published 28 May 2026 in the Brandflow newsletter on LinkedIn.