Is AI Replacing Google, or Changing What People Ask It to Do?

Every few months a headline declares that Google is finished. Then Google keeps posting record usage, and the cycle restarts. For a founder with a limited marketing budget, the question is practical: do you need to change what you do? The balanced answer is that AI search is changing how people look for information and recommendations, but it has not replaced Google, and the right response is to adapt, not to panic. Our GEO services are built on that view.

How much of Google's traffic is shifting to AI search

No one outside the platforms has a complete picture, so treat precise figures from any source with caution. What we can observe is directional:

• AI assistants such as ChatGPT, Perplexity, Gemini and Copilot now handle a large and growing volume of questions.
• Google has added AI Overviews and AI-generated answers to many results, which changes how often people click through.
• Many sites report a small but increasing share of referrals from AI assistants, while classic search still delivers most organic traffic.

The most reliable data about your market is your own: check your analytics for referral sources from AI platforms and watch for changes in click-through rates in Google Search Console.

How user behaviour is changing

People are not abandoning search. They are splitting tasks. Quick factual lookups, navigation and local needs still go to Google. Multi-step research, comparisons, summaries and "recommend something for me" requests increasingly go to AI assistants. Buyers also use both in sequence: ask an assistant for a shortlist, then search each name on Google to check reviews and pricing.

For businesses, that second step matters. Even if the first mention comes from an AI answer, your Google presence decides whether a prospect trusts you.

What this means for your marketing strategy

Three implications stand out.

1. Visibility is now multi-channel. You need to appear in blue links, in AI Overviews and in assistant answers.
2. Clear, structured, quotable content wins in all of them. Good SEO fundamentals still apply, and they carry into AI search.
3. Brand signals matter more. Consistent descriptions, reviews and third-party mentions help AI systems recommend you with confidence.

None of this requires abandoning existing work. It adds a layer on top.

Should you prioritise Google or AI search?

Both, but differently. For most businesses:

• Keep Google as the base channel. It still drives the majority of organic demand, and strong rankings support AI visibility.
• Add AI-specific improvements such as answer-first passages, clear entity information, structured data and checks that AI crawlers can read your pages.
• Measure both. Track classic rankings and traffic alongside periodic checks of what AI assistants say about your category. Our post on GEO vs SEO explains how the two fit together.

If your buyers are heavy AI users, such as developers and technology buyers, weight the AI work more heavily.

Signals worth watching

Rather than reacting to headlines, track a short list of signals:

• The share of your referral traffic coming from AI assistants.
• The click-through rate on your main Google queries in Search Console.
• How often AI Overviews appear for your target keywords.
• How AI assistants describe your category and your brand when asked.

If those change materially, adjust your mix. If they do not, keep building the fundamentals.

A worked example: two founders, two responses

Founder A reads a headline, pauses SEO spending and moves the whole budget into experimental AI tactics. Six months later, rankings have slipped, and the AI work has no baseline to measure against. Founder B keeps core SEO running, spends a small slice of time on answer-first rewrites of three key pages, and sets a monthly routine of asking 20 buyer questions in three assistants. When a competitor starts appearing in assistant answers, Founder B sees it within weeks and can respond. Neither approach guarantees a result, but B stays informed and keeps the channels that already work.

A simple way to split effort

Treat this as a starting point, not a rule. Your analytics and customer conversations should adjust it.

What not to do

Do not rebuild your entire strategy around a single platform, rely on one forecast or stop investing in search because of a headline. Do not publish low-quality pages in bulk to "feed" AI either. The durable approach is useful content, clear facts and consistent brand information, because those help in every version of search.

The most likely search landscape in 2027

Predictions are uncertain, but a plausible picture is a mixed one: Google with more AI-generated answers, AI assistants with more live search and more sources shown, and advertising formats appearing in both. In that world, the winners are brands that publish clear, trustworthy information and are consistently described across the web. Preparing now is inexpensive compared with catching up later.

Questions to ask before changing your budget

Before shifting spend, answer these honestly: Where do your best customers say they found you? Do your analytics show any referrals from AI assistants, and are they converting? When you ask an assistant for recommendations in your category, are you named, and is the description accurate? Do your competitors appear? If your buyers are not using assistants yet and your Google traffic is healthy, a modest AI layer is enough. If buyers say they "asked ChatGPT", raise the priority.

What to do this quarter

1. Check your analytics for AI referral traffic and note the baseline.
2. Ask ChatGPT, Perplexity and Gemini the 20 questions your buyers ask, and record which brands appear.
3. Improve your three most important pages with answer-first structure.
4. Keep investing in your core SEO.

For a faster baseline, run our free AI visibility audit.

Information Gain

Ask new customers which resources they used and what each helped them decide. Keep those observations separate from analytics attribution. Distinguish discovering options, comparing them and confirming a choice. Use the role to guide content without treating a small customer sample as a market-share measurement.