How AI Search is Reshaping How Customers Find Businesses
Here is a number worth sitting with: ChatGPT has roughly 900 million weekly active users as of early 2026. That is more than the population of Europe.
These are not all using it for coding help. A growing share are using it the way they used to use Google: to find things. Hotels, restaurants, software tools, accountants. The full range of commercial discovery.
The Old Discovery Funnel is Fracturing
For two decades, the customer discovery path was stable. Someone had a need, typed it into Google, scanned results, clicked a few links, and made a decision. Your job was to rank in those results.
The path is changing. Not disappearing, but fracturing into multiple surfaces, and one of those surfaces is growing faster than any other in the history of web traffic.
AI referral traffic grew 357% year-over-year in June 2025, according to Similarweb. In the travel sector specifically, AI-driven traffic grew 539% during the 2025 holiday season, according to Adobe. Retail grew 693%.
Gartner projects a 50% decline in organic search traffic by 2028. Their 2026 prediction: a 25% drop in traditional search volume, attributed directly to AI chatbots and virtual agents absorbing queries.
How AI Recommendations Actually Work
AI platforms like ChatGPT, Gemini, and Perplexity do not return a list of ten results for you to scan. They return one synthesized answer. They have already done the evaluation and reached a conclusion before you see it.
Training data. AI models are trained on large corpora of web content. If your business was well-documented and mentioned across credible sources before the model's training cutoff, the model has a representation of you in its weights.
Real-time web retrieval. ChatGPT with web search, Perplexity by default, and Gemini with Google Search grounding actively pull current information from the web when generating answers.
Third-party signals. Reviews, mentions in editorial content, listings in directories. AI systems cross-reference these signals to evaluate trust. A business with 500 detailed reviews and mentions in travel publications is more likely to be cited than one with identical services but a thin online footprint.
Two Things Make This Different
First, the AI collapses the research process. Profound's research found that 79.7% of buyers rely on answer engines for at least half of their decision-making process. The consideration phase, which used to involve clicking through multiple websites, is increasingly happening inside a single AI conversation. If you are not part of that conversation, you are not part of the consideration set.
Second, the AI introduces an intermediary. In traditional search, the customer saw your page title and chose to visit you. In AI search, the AI summarizes you. It writes the description. It decides what context to provide. If the AI has wrong information about you, the version of you that appears in the answer is not the real you.
The Asymmetry Nobody Talks About
Large brands benefit from years of web presence, press coverage, and consistent entity data. AI models have seen them many times, from many credible sources, and have a well-calibrated representation of them.
Smaller businesses, even excellent ones with strong local reputations, may be largely absent from the data an AI model was trained on. The AI fills gaps with uncertainty, and uncertain entities get left out of answers.
This creates a version of the zero-click problem that is harder to solve than the original. In traditional SEO, you could invest in content and earn traffic. In AI search, you need to invest in building an evidence base that AI systems find credible.
What This Means for Businesses
Your AI presence is now part of your business infrastructure, whether you manage it or not.
Every week, hundreds of millions of people ask AI platforms questions that could be answered by recommending your business. Whether you appear in those answers depends on the quality of your real-time web presence today.
The businesses that will be well-positioned by 2028 are the ones that started treating AI visibility as a managed asset in 2026. Not because the tools are mature yet. Because the compounding effects of reputation, citation, and consistent entity data take time to build.