The Death of the Click-Based Funnel

Traditional search made customers do the research. Anyone looking for a mattress might search “best mattress for back pain,” then open several websites, compare materials and reviews, and have a preference. Whereas now they can ask an AI assistant the same question and receive a short list with explanations and recommendations in seconds.

And this is already showing up in commerce. Adobe¹ reported that AI-driven traffic to U.S. retail websites grew 393% year over year in Q1 2026. Even more interestingly, AI-referred visitors converted 42% better than non-AI traffic in March 2026.

This tells us that AI is changing the quality of intent behind that traffic.

Let’s take an example: You have an acne prone skin and need a moisturiser for yourself. Instead of browsing ten brands, you will most probably ask an AI assistant for the best options. The AI recommends Brand A because it is fragrance-free, explains why it suits your skin type and compares it with two alternatives.

When you land on Brand A’s product page, you aren’t starting from zero as you already have a reason to prefer the product. Now your question has changed from “Should I buy this?” to “Is this really what I was told it is?”

That distinction is crucial for CRO.

If the AI highlighted “fragrance-free” as the reason to choose the product, that information should be immediately visible. If the recommendation was based on customer reviews, relevant proof should be easy to find.

If the customer was told the product is ideal for sensitive skin, the page should make that claim credible. Otherwise, the website creates doubts in the customer’s mind.

Because of this, the subtle shift in CRO is that high-converting pages may increasingly need to confirm decisions.

A new problem has emerged. Now, we may not know what influenced the conversion. For example, A customer discovers your brand through an AI conversation on Monday, searches your brand directly on Friday and purchases on Saturday.

Analytics may simply report: Direct → Purchase
But the AI interaction that created the preference is invisible. This is the growing gap between conversion attribution and decision attribution.

CROs use tools to investigate customer intent, such as qualitative research, surveys, session recordings, behavioural analytics, segmentation and experimentation.

The first step is identifying AI-referred traffic. When a customer clicks through to your website directly from platforms such as ChatGPT, Perplexity or Gemini, the referral source may be visible in your analytics.

Let’s say you discovered that the AI-referred visitors converted but repeatedly visited the returns page before purchasing. That gives you a hypothesis: “These customers are ready to buy, but return uncertainty is creating the final hesitation.” You can test stronger returns messaging on the PDP and measure the impact on add-to-cart rate, checkout initiation and revenue per visitor.

The next evolution of CRO is about understanding that the website is no longer the starting point of every decision. As AI takes over more of the research, CRO needs to become better at understanding what happens before the click.

This is where the research-led, experimentation-driven approach ConvertPolo has always taken becomes even more relevant: understanding intent, identifying friction, and testing what actually moves customers closer to revenue. The first half of the decision may belong to AI. The second half is where great CRO earns its keep.

1. https://business.adobe.com/blog/ai-traffic-surge-retail-sites-not-machine-readable