AI visitors are worth more. AI tools are about to cost more.

Two stories ran side by side this week, and marketing leaders need to hold both at once. The visitors that AI sends you are becoming more valuable. And the AI tools your team relies on are getting more expensive to run, which usually means usage limits come next.

What happened

Adobe found that in September, traffic referred by AI tools generated 83% more revenue per visit than other traffic and converted 90% better. AI visitors also spent 73% more time on site and added products to their carts 53% more often. The catch: the average retail homepage scored just 63% on Adobe's measure of how easily AI can read it. Product specifications, setup guides and customer service pages did better.

The volume is growing too. In a recent interview, OpenAI said ChatGPT's search volume is about 20% of Google's, and that it added 200 million users in the past two months.

Email is quietly benefiting from the shift. An analysis of Similarweb data puts US email traffic up 16.3% in the first eight months of 2026 compared with the same period last year, with the median website's email traffic up 14.7%. That includes receipts and notifications, so it counts visits, not new customers. Still, the direction is clear.

Now the other side. The AI companies are spending heavily and struggling to turn it into profit. One report says OpenAI posted more than $20 billion in operating losses in the first half of 2026, and Anthropic's infrastructure costs rose 329% year over year. xAI has already cut Grok usage limits because operating costs were too high relative to revenue. The same report suggests other providers may tighten limits as pressure for returns grows. Even product changes point that way: ChatGPT and Codex will soon draw from the same usage allowance.

What it means if you lead marketing

First, AI visitors deserve their own line in your reporting. If they convert better, a small share of traffic can carry a large share of revenue, and averaging them in with everything else hides it. Make sure the pages AI tools read most, such as product details, specifications and support content, are clear and complete.

Second, be careful with the numbers. One piece this week made a point I wish more dashboards carried: with small samples, a month-to-month jump in AI search visibility may just be chance. Look for a trend across several months before you credit a new piece of content.

Third, plan for AI tool costs to rise or limits to tighten. If your team has built daily workflows on a single tool's generous allowance, that is a dependency worth knowing about now. Know which tasks matter most, and what you would do if the limits changed.

Finally, email is your most dependable channel right now. When search is less predictable, a list of people who want to hear from you is an asset no platform can take back.

What I would do

• Split AI-referred traffic out in your analytics and compare conversion and revenue per visit against other channels.
• Review your product, specs and support pages for AI readability. Clear structure and complete facts serve both people and AI.
• List the AI tools your team depends on daily, with what each costs and what happens if its limits change.
• Give people a reason to open your emails that is not a discount: useful updates, real advice, a person's voice.

Worth trying

A "Why did you cancel?" email. Send it automatically the moment a customer cancels. Keep it short and conversational, and ask one question. The answers tell you what to fix, and some replies turn into conversations that win the customer back. It costs nothing and puts a person back in a moment most companies leave to a confirmation screen.

AI is sending better visitors, and the tools themselves are getting more expensive to run. Measure the first carefully and plan for the second.