Consumers are increasingly asking AI assistants to compare products, plan holidays and narrow purchase decisions instead of working through pages of search results. Early evidence suggests AI-referred visitors can also arrive with stronger commercial intent. But search is far from being replaced. As Google itself becomes more conversational, the real battle may be shifting from winning the click to becoming part of the answer.
For more than two decades, searching online followed a predictable routine.
A consumer typed a few words into Google, scanned a page of results, opened several links and pieced together an answer.
Artificial intelligence is beginning to compress that journey.
A shopper looking for a smartphone can ask an AI assistant to compare cameras, battery life and prices within a specific budget. A traveller can request a four-day itinerary built around food preferences, hotel budgets and flight timings. Someone buying skincare can describe their skin type, ingredients they want to avoid and how much they want to spend.
The response can arrive as a recommendation rather than a collection of links.
That distinction is turning ChatGPT, Gemini, Perplexity and other AI assistants into discovery channels in their own right. At the same time, Google is putting AI-generated answers and conversational capabilities directly into Search.
It raises a question that would have sounded premature only a few years ago: for some consumer journeys, is asking AI becoming better than searching?
Recent data suggests the answer may be yes for tasks involving research, comparison and multiple constraints. But the evidence does not support declaring the search engine obsolete.
Instead, AI appears to be changing what consumers expect a search experience to do.
From typing keywords to describing the problem
Traditional search works particularly well when people know what they are looking for.
Someone searching for a company’s website, today’s weather, a restaurant nearby or the price of a particular product can express that intent in a few words.
AI becomes more useful when the question is difficult to reduce to keywords.
Consider the difference between searching “South Goa hotels” and asking for “a quiet South Goa hotel for two adults, close to the beach, suitable for remote work, with good vegetarian food and a budget below ₹12,000 a night.”
The second request contains several conditions. A conversational system can retain them, recommend options and respond when the traveller adds another requirement.
Search engines traditionally required consumers to perform much of that synthesis themselves.
This may help explain why AI is gaining ground in research-heavy journeys.
Adobe’s August 2026 AI Traffic Trends research, which drew partly on more than one trillion visits to US retail websites and a survey of more than 5,000 consumers, found that 95% of AI users considered responses from AI platforms at least as trustworthy as traditional search engines. Among regular AI users, that figure increased to 97%.
Trust does not necessarily mean accuracy, and people who already use AI regularly are likely to view the technology differently from non-users. But the numbers indicate that AI assistants have moved beyond novelty for a substantial group of consumers.
Behaviour inside conventional search is changing as well.
Similarweb’s 2026 Generative AI Landscape analysis found that more than 40% of US Google searches were triggering an AI Overview. It also reported that average Google query length had increased by 5.4% following the arrival of AI Mode.
Ethan Smith, CEO of Graphite, described the shift as consumers adopting “a new more natural way to search and discover.”
That may be AI’s most immediate challenge to the traditional search model.
Keyword search trained people to shorten their intentions into machine-friendly phrases such as “best laptop student India” or “cheap flight Delhi London.”
Conversational interfaces encourage them to do the opposite.
Consumers can explain what they want in detail, ask another question and refine the answer without beginning the search again.
This makes AI particularly relevant earlier in the purchase funnel. Search has historically been powerful at capturing intent that already exists. An AI assistant can potentially participate while that intent is still being formed.
A consumer may not know which running shoe to search for, for example. They may only know that they want something for daily walking, suitable for wide feet, below a certain price and available in India.
AI can turn that problem into a shortlist.
Shopping is becoming an important test
Commerce is where the difference between AI discovery and conventional search becomes particularly visible.
McKinsey’s 2026 State of the Consumer research, based on 4,863 consumers across Brazil, France, Germany, the UK and the US, found that around one-quarter were already using generative AI for shopping. Among Gen Z respondents, the figure reached 28%.
The same research found that 60% of Gen Z respondents regularly used AI Overviews at the top of traditional search experiences.
That distinction matters. AI discovery is not occurring only on standalone services such as ChatGPT or Perplexity. It is also being inserted into the search engine itself.
For consumers, the practical difference may eventually matter less than it does for marketers.
A shopper researching televisions can now ask about OLED versus Mini LED, compare models for gaming, narrow the selection by budget and ask about reliability before reaching Samsung, LG, Sony or a retailer’s website.
The brand website is no longer guaranteed to be the place where consideration begins.
Adobe’s 2026 retail traffic data provides an indication of what happens when AI users eventually do click.
In July, visitors referred to retail websites by AI services converted at a rate 60% higher than non-AI traffic and generated 53% more revenue per visit. Adobe also reported that these visitors spent more time on retail websites and were less likely to leave immediately.
Among consumers surveyed by Adobe, 42% said they had used AI assistants for online shopping, while 83% of those users believed AI had improved the shopping experience.
These numbers should be interpreted carefully. AI-referred traffic is still relatively new, and consumers who use an assistant before visiting a retailer may already be unusually research-oriented or closer to making a decision.
The findings therefore do not establish that an AI referral will always be more valuable than a search referral.
They do, however, suggest a different path to the website.
A consumer coming through search may still be investigating a broad category. Someone arriving after an extended conversation with an AI assistant may have already compared products, eliminated alternatives and established a budget.
AI may therefore reduce part of the research that traditionally happened after the click.
The same pattern is appearing in travel. Adobe reported that AI-driven visits to travel websites increased 119% year on year in July 2026. Among surveyed travellers using AI assistants, 84% said the tools had improved their planning experience.
That creates a new measurement problem.
If an AI assistant recommends a hotel, explains why it fits the traveller’s requirements and influences the eventual booking, the hotel may have benefited from AI even if the final visit comes through Google, an online travel agency or a direct URL.
Rand Fishkin, founder of SparkToro, summarised the problem: “AI influence is happening. What marketers need now is a new way to measure and attribute that impact.”
The answer is easier, but someone else chooses the sources
AI’s ability to remove work from the discovery process is also what makes it complicated.
Search presents choices.
AI interprets those choices before presenting an answer.
A 2026 University of Toronto study compared Google Search with GPT-4o, Claude 4.5 Sonnet, Gemini 2.5 Flash and Perplexity Sonar Pro across 1,000 consumer ranking queries.
The information environments were markedly different.
GPT-4o’s cited domains overlapped with Google’s top ten results by an average of only 4%. The overlap reached 11.1% for Gemini, 12.6% for Claude and 15.2% for Perplexity. For more than half of GPT-4o queries, none of the cited domains appeared among Google’s top ten.
This does not establish which system provided the better answer.
It demonstrates that asking AI and searching Google can expose consumers to substantially different sources.
That has consequences for brands.
A company ranked sixth on a conventional results page can still be discovered when someone scrolls. If an AI assistant generates three recommendations and leaves that company out, there may be nothing for the consumer to scroll to.
AI therefore changes visibility from ranking among options to being selected for the answer.
McKinsey’s consumer research provides another warning for marketers. In its analysis of consumer goods, only about 1% of sources cited by large language models came from brand-owned websites.
AI systems were also drawing information from publishers, forums, retailers, reviews, videos and other parts of the open web.
For marketers, this means optimising a corporate website may no longer be enough. Product descriptions across retailers, independent reviews, earned media, community discussions and consistent brand information can all influence what an AI system understands about a product.
Troy Treangen, NIQ’s Chief AI and Product Officer, has described the wider shift simply: “Consumers are making purchasing decisions earlier and through more channels than ever before.”
There is also a consumer risk.
AI-generated responses can be fluent and decisive even when the underlying information is incomplete or incorrect. Search results expose competing sources relatively visibly. An AI assistant can synthesise several sources into one polished response, making uncertainty less obvious.
For high-stakes questions, verification, source diversity or rapidly changing information, the ability to inspect primary sources can therefore remain more valuable than receiving the fastest answer.
Search may survive by becoming more like AI
Perhaps the biggest problem with asking whether AI will beat search is that search engines are adopting AI themselves.
Google said in 2026 that AI Mode had surpassed one billion monthly users within a year of launch and that queries were more than doubling each quarter. At the same time, the company said overall search queries had reached an all-time high.
These are company-reported figures, but they complicate the idea that conversational AI is simply taking search volume away from Google.
Google’s stated strategy is to “bring together the best of a search engine with the best of AI.”
AI Mode allows follow-up questions, longer prompts and multimodal inputs while continuing to provide links to external websites. Search is therefore evolving from a system that retrieves pages into one that can increasingly interpret questions, synthesise information and assist with tasks.
For marketers, that means SEO is unlikely to disappear.
Its role is broadening.
Brands still need websites that search engines can crawl, useful content, accurate product information and conventional visibility. But they increasingly also need to understand whether AI systems recognise the brand, describe it accurately and include it when consumers ask category-level questions.
This has produced new labels such as generative engine optimisation and answer engine optimisation, although the underlying practices remain closely connected to traditional SEO.
Lily Ray, VP of SEO and AI Search at Amsive, has called generative engine optimisation “a new layer on top of traditional SEO.”
That may be a more useful way of understanding the transition than treating AI and search as opposing channels.
Consumers are unlikely to abandon search for every task. They may use Google to find a particular website, an AI assistant to compare several products, Maps to find a nearby store and another AI-powered search experience to research reviews.
The winner may depend on the question.
AI has a natural advantage when consumers want synthesis: compare these options, explain the difference, remember my preferences and narrow the choices.
Traditional search remains useful when people want to navigate directly, inspect several sources, independently verify information or control what evidence they consider.
Increasingly, the consumer may not even know which system they are using because AI answers and conventional search results will appear within the same interface.
For marketers, the more important change is therefore happening behind the search box.
The old discovery model asked whether a brand ranked highly enough to earn a click.
The emerging model introduces another question before that click ever happens: when AI turns thousands of webpages, reviews and recommendations into one answer, does the brand make it into that answer at all?
AI channels are not universally better than search engines.
But for consumers who want an answer rather than a list of places to find one, they are increasingly changing the definition of what a good search experience looks like.
Disclaimer: All data points and statistics are attributed to published research studies and verified market research. All quotes are either sourced directly or attributed to public statements.