The new front door does not send traffic

Something quietly broke in how people find things online. New reporting this month puts it in numbers: around 10% of people worldwide now use AI chatbots for news every week, up from 7% a year ago. The figure that should make any founder sit up is the second one. Only about 4% of those users regularly click through to the original source.

People are getting the answer and staying put. The chatbot reads your page, summarises it, and the reader never arrives. For a decade the deal was simple. You made good content, you ranked, you got the visit. That deal is being rewritten in real time, and not in your favour.

Why this hits more than publishers

It is easy to file this under "news industry problem" and move on. That would be a mistake. The same mechanic applies to anyone whose growth depends on someone discovering them through search or a feed.

Your product comparison page. Your pricing explainer. Your "best X for Y" guide. Your store category copy. All of it is now retrieval fuel for an assistant that answers on your behalf and rarely names you. If the model can satisfy the question without the visit, the visit stops happening. Sessions fall. Top of funnel dries up. And the dashboard shows a slow leak you cannot fully explain, because the referrer is simply gone.

Stop optimising only for the click. Optimise for the citation.

Here is the shift in plain terms. For years the goal was to be ranked. Now the goal is to be quoted. Those are not the same job.

Being quoted by an AI system rewards different things:

Clean, machine-readable structure. Clear headings, real answers near the top, structured data, and FAQs the model can lift cleanly. Walls of marketing prose get skipped.
Specific, verifiable claims. Models lean on content with concrete numbers, dates, and named facts. Vague copy gets paraphrased into nothing.
Being the primary source. If you publish the original data, the original benchmark, the original how-to, you become the thing other answers point back to.
Presence where retrieval happens. Feeds, APIs, and structured catalogues matter more than a pretty landing page the crawler has to fight through.

For ecommerce this is sharper still. If shoppers start their search inside an assistant, your product data, availability, and specs need to be clean and exposed, not locked inside a theme that only renders for a human browser. The store that wins the agent recommendation is the one whose data is easiest to read and trust.

Measure what the dashboard is hiding

Most teams still grade themselves on sessions and bounce rate. Add a second scoreboard. Test the assistants directly. Ask ChatGPT, Gemini, and Claude the questions your customers ask, then check whether you show up, how you are described, and whether the facts are right. That is your new search ranking, and almost nobody is checking it weekly yet.

Treating this as a content problem misses it. It is an architecture problem. How your information is structured, exposed, and verified now decides whether a machine can represent you faithfully.

We are here to help founders and teams design and build digital products that are built to scale with you, not slow you down. If you are looking to build something, get in contact with us today!

The takeaway

The click was never the point. The customer was. AI just removed the click and kept the customer in its own window. The teams who adapt will stop fighting for the visit and start building to be the source the answer cannot leave out.