The Shift From Keyword Search to Conversational Discovery
INDUSTRY AND COMPETITIVE CONTEXT
For roughly two decades, online discovery ran on one model. A consumer typed a few keywords, received a ranked list of links, and chose where to click. Marketers built entire disciplines around that model: search engine optimisation to earn a position on the page, and paid search to buy one. The click was the unit of value, and the results page was the shelf.
Between 2024 and 2026, three developments changed that model. Each is documented by the companies concerned.
The first was the arrival of conversational search products from outside the incumbent. OpenAI launched ChatGPT search on October 31, 2024, first to paid subscribers. Its announcement described fast answers with links to web sources and follow-up questions that draw on the full conversation. It reached all logged-in users on December 16, 2024, and became available to everyone on February 5, 2025, with no sign-up required. OpenAI also named launch partners among news publishers, including the Associated Press, Axel Springer, Financial Times, Le Monde, News Corp and Reuters.
The second was the response from the incumbent. At Google I/O on May 20, 2025, Google announced that AI Mode, a conversational search experience, would roll out to everyone in the United States. Google described it as a way to ask complex, multi-part questions and follow up. Its related Deep Search feature breaks a question into subtopics and issues dozens or even hundreds of queries to assemble a cited report. Alongside it, AI Overviews, the generated summaries that appear above conventional results, expanded to more than 200 countries and territories.
The third was the growth of independent answer engines. Perplexity's CEO, Aravind Srinivas, said at Bloomberg's Tech Summit in June 2025 that the company handled 780 million queries in May 2025. He said it was growing more than 20 percent month over month and handled about 30 million queries per day, against 3,000 on its first day in 2022.
The competitive structure is therefore not a simple displacement story. The incumbent, the largest AI developer and a specialist startup are all competing to become the interface through which consumers ask questions. Each frames the product as a conversation rather than a list.

BRAND SITUATION PRIOR TO THE SHIFT
The "brand" at the centre of this case is the keyword search model itself, together with the marketers who depended on it.
Under that model, a brand's discoverability was largely a function of two controllable inputs: the pages it published and the positions it earned or bought. Measurement was built around rankings, impressions, clicks and the traffic that followed. Because the consumer saw a list and made the selection, the brand's task was to win a place on the list.
Google's position as the dominant gateway made this a stable system. No verified public information is available in the sources reviewed on how any individual brand's marketing budget was divided between organic and paid search before 2024. This case therefore does not make claims about the scale of that dependency for any named advertiser.
What is documented is the state of marketer readiness at the point of change. McKinsey's October 2025 report, "New front door to the internet: Winning in the age of AI search," drew on a September 2025 survey of about 30 Fortune 500 consumer-brand chief marketing officers. It found that just 16 percent of brands systematically track AI search performance. In other words, the dominant measurement systems were designed for a list of links, and most large brands had not yet built equivalent systems for answers.
STRATEGIC OBJECTIVE
Google has stated a direction of consolidation. In its Q2 2026 earnings remarks, CEO Sundar Pichai said Google had "brought together AI Overviews and AI Mode into one seamless Search experience." In Q3 2026, Google again described "building one seamless experience across AI Overviews, AI Mode, and traditional Search." The stated objective is to keep conversational capability inside Search rather than leave it to a separate product.
OpenAI's stated objective for ChatGPT search was to combine a conversational interface with live information. Le Monde's CEO, quoted in OpenAI's launch announcement, said that "AI search will be, in a near future and for the next generations, a primary way to access information." That is a partner's view, published by OpenAI. It is not an OpenAI forecast.
Perplexity's CEO said in June 2025 that its Comet browser would move AI from giving answers to completing actions for users. He described a single AI action as a full browsing session completed from one prompt, versus the four or five searches a typical AI answer represents.
CAMPAIGN ARCHITECTURE AND EXECUTION: HOW THE SHIFT WAS ROLLED OUT
Because this is a platform transition, the relevant "execution" is how each company introduced conversational discovery to consumers. The sequencing reveals three distinct strategies.
OpenAI used staged expansion from a position of existing usage. ChatGPT search began with paid subscribers, widened to all logged-in users within seven weeks, and then to everyone without an account. Search was added to an assistant that already had a large audience. OpenAI reported 800 million weekly active users in October 2025 and 900 million in late February 2026, alongside 50 million paying subscribers. Those figures describe ChatGPT as a whole. No verified public information is available on what share of ChatGPT usage is search, beyond what OpenAI has stated.
Google used a layered approach built into its existing product. AI Overviews came first, appearing within standard results. Google reported 1.5 billion monthly users of AI Overviews in May 2025 and 2 billion in its Q2 2025 earnings call. In that same call, Google said AI Overviews drove over 10 percent more queries for the types of search that display them. AI Mode, a more fully conversational layer, followed. Google reported over 100 million monthly active users of AI Mode in the United States and India in July 2025. In its Q2 2026 remarks, it said AI Mode had surpassed 1 billion monthly active users.
Perplexity built its position as a dedicated answer engine and then moved toward owning the access point. The company's CEO tied future growth to its Comet browser, which he said would create "infinite retention" once users were in it.
Two architectural points matter for marketers. First, the answer is now assembled rather than retrieved. Google's description of Deep Search as issuing "dozens or even hundreds of queries" on a user's behalf means the consumer's single question can generate many background searches. Second, the interface has moved beyond text. In its Q3 2026 remarks, Google said that queries in AI Mode are on average three times longer than traditional searches and that one in five AI Mode searches globally use voice, images or documents. Google also said it has crossed over 50 billion visual searches per month.
POSITIONING AND CONSUMER INSIGHT
Conversational queries differ in kind from keyword queries. The Q3 2026 figure that AI Mode queries are on average three times longer than traditional searches supports this. A longer query carries context, constraints and intent that a short keyword string cannot. The consumer is describing a problem rather than naming a product.
Consumers are asking questions early in their decision process. McKinsey's August 2025 AI Discovery Survey, fielded among a representative US consumer panel (n = 1,927), found that half of consumers use AI-powered search. Among those users, more than 70 percent ask top-of-funnel questions. Discovery and consideration are the stages where brands are most exposed to being included or left out.
Users also appear to treat generated answers as sufficient. A Pew Research Center study analysed browsing data from 900 U.S. adults collected in March 2025. Of the Google searches analysed, 18 percent produced an AI summary. When a summary appeared, users clicked a traditional result link in 8 percent of visits, compared with 15 percent of visits without one. Users clicked a link inside the summary itself in 1 percent of visits. They ended their browsing session after 26 percent of visits with a summary, compared with 16 percent without. The study measures behaviour in one month, in one country, on one search engine. It does not measure outcomes for any brand.
The behaviour is not confined to the West. A YouGov India Web Search and AI Report 2026, based on more than 950 online searchers in India surveyed between April 21 and May 29, 2026, found that 65 percent had used search engines in the past 30 days and 57 percent had used AI assistants or chatbots. Respondents reported using AI assistants to get a direct answer (49 percent), to verify information from elsewhere (44 percent) and to summarise or compare information (43 percent). Trust in AI assistants was 69 percent, below the 75 percent recorded for both search engines and maps and navigation apps. Two-thirds expected to use AI assistants more in the next 12 months.
On shopping specifically, a McKinsey survey of about 21,000 consumers across 11 markets was reported by the news agency ANI in October 2026. It found that 76 percent of consumers in India use AI tools to search for products, compare choices or make purchase decisions, second to China at 80 percent, against a global average of 53 percent. This figure comes from a secondary news report, and the original McKinsey publication was not independently reviewed for this case.
The strategic reading is that trust and usage are growing together, but trust remains lower than for established tools. A consumer who verifies an AI answer elsewhere, as 44 percent of Indian respondents reported, is still passing through other sources before deciding.
MEDIA AND CHANNEL STRATEGY
What is publicly documented concerns the structure of the channel. McKinsey reported that a brand's own sites make up only 5 to 10 percent of the sources AI search references. By implication of that finding, a brand's visibility in AI answers depends heavily on third-party content: reviews, publishers, forums and other independent sources. McKinsey also reported that about 50 percent of Google searches already have AI summaries.
The monetisation model is also documented only in part. In June 2025, Perplexity's CEO
Publisher relationships form a second channel layer. OpenAI's launch announcement stated that any website or publisher can opt in to appear in ChatGPT search, and that it worked with a group of news publishers. Google stated in its Q2 2026 remarks that it is "now sending billions of clicks to websites every week through AI features in Search." Pew's click data and Google's statement describe different measures.
BUSINESS AND BRAND OUTCOMES
For Google, the shift has coincided with reported query growth rather than decline. Google said in July 2025 that AI Overviews drove over 10 percent more queries for the search types that show them. In Q2 2026, Google said AI Mode is driving an incremental increase in Search queries overall, and that it reduced the cost of AI Mode responses to its lowest level since launch. In Q3 2026, it said its year-over-year query growth rate was the highest in the last five years and that AI investments were driving growth including in commercial queries. Google also reported that the Gemini app reached one billion monthly active users. These are Google's own statements and have not been independently audited in the sources reviewed.
For OpenAI, the reported outcomes are scale metrics: 900 million weekly active users and 50 million paying subscribers as of February 27, 2026, announced alongside a $110 billion private funding round at a $730 billion pre-money valuation. In February 2026, OpenAI CEO Sam Altman said India had 100 million weekly ChatGPT users, making it the second-largest market after the United States, as reported by Digit.in.
For Perplexity, the documented outcome is query growth as reported by its CEO: 780 million queries in May 2025.
For consumers and the wider market, the documented outcomes are behavioural. These include Pew's click-through and session-ending differences, McKinsey's finding that half of US consumers use AI-powered search, and the YouGov India findings on adoption and trust.
On forecasts: Gartner stated in a February 2024 press release that traditional search engine volume would drop 25 percent by 2026, with search marketing losing market share to AI chatbots and other virtual agents. No verified public information is available confirming that this prediction was realised. Google's reported query growth in 2026 points in a different direction for Google's own platform, though the two measures are not like-for-like.
On revenue at stake, McKinsey stated that by 2028, $750 billion in US revenue will funnel through AI-powered search, and that unprepared brands may see traditional search traffic fall by 20 to 50 percent. These are McKinsey's projections, not realised outcomes.
STRATEGIC IMPLICATIONS
First, the unit of competition is moving from the page to the answer. In a keyword model, a brand competes for position among ten results. In a conversational model, a brand competes for inclusion in a single synthesised response. The Pew data suggest that when a summary appears, fewer users click through to results. A brand that is absent from the answer may therefore lose the chance for a click it would previously have contested. The study does not show how many of those lost clicks were commercially valuable.
Second, control over visibility is shifting from owned to earned and third-party content. If brand-owned sites account for 5 to 10 percent of cited sources, as McKinsey reports, then reputation across independent publishers, review sources and communities becomes a primary lever. This changes the role of marketing from optimising owned pages to managing the whole information environment around the brand.
Third, measurement is the first capability gap. With only 16 percent of surveyed large brands systematically tracking AI search performance, most are managing a growing channel without instruments for it. The strategic implication is that organisations able to measure AI visibility earlier gain information that competitors lack. That is a reading of McKinsey's finding, not a documented outcome for any firm.
Fourth, the incumbent is adapting rather than ceding the market. Google's integration of conversational capability into Search, and its reported query growth, show that conversational discovery and the incumbent's scale are not mutually exclusive. The competitive question is less whether keyword search disappears and more who owns the conversational interface.
Fifth, trust is the constraint on adoption. YouGov India found that AI assistants were trusted by 69 percent of online searchers, below search engines and maps. Many users verify AI answers elsewhere. Brands should treat AI answers as one stage in a multi-source journey rather than a replacement for it. Credibility signals, such as official sources and multiple sources side by side, were among the factors respondents named as raising trust.
Sixth, the evidence base has limits. Most usage figures are reported by the companies that benefit from showing growth. Monthly and weekly active-user metrics are not directly comparable. The Pew study covers one country and one month. The Gartner forecast has not been confirmed. Strategic decisions should rest on directional agreement across sources, not on any single number.



Comments