The infinite FAQ: the part of your site that knows your customers best.

FAQs are the part of your site that knows your customers best. And almost nobody listens to them. It is where the user arrives with the specific doubt, the one standing between them and a purchase or a quote request. So is support. And yet we almost always write them blind: five questions marketing THINKS the customer asks, placed at the bottom of the page.

This week I saw a very smart solution on Foglamp's site. Five fixed questions and a sixth row that says "Ask anything else". You type your doubt and the AI answers right there, inside the FAQ, as one more question. Voilà! An FAQ that never ends.

What an infinite FAQ changes for the business

It looks like a design detail, but for the business it changes three things:

1. You ALWAYS answer. This does not need the most powerful model: a small, cheap one that only uses that page's information and the business documentation costs so little per answer that you can reply at any hour and with no queue.
2. You replace a good part of the chat. A chat with people costs money and does not answer at 11 p.m.; the classic bot is a bubble that does not know which page you are on. The infinite FAQ handles what repeats and hands over to a person what it does not know.
3. The question arrives attached to its page. "Can I cancel?" on the pricing page is not a technical doubt, it is fear of commitment. The model gets that context for free and answers better. And you know which page is holding back the sale and why.

What we already did with reviews, but before the purchase

At iSocialWeb we have spent years doing something similar with e-commerce reviews: read them all, group them by topic and hand them back to the client to improve the product page (and sometimes the product itself). We used to do it by hand. Now with AI. Amazon plays the same game when it summarises reviews with AI and flags items that get returned often.

But a review arrives after the purchase, and only from whoever bought. The FAQ question arrives before. And also from people who leave. It is the only way to hear the customer you are losing.

How I would build an infinite FAQ step by step

1. A cheap model with closed sources: the page content and the verified documentation. If the answer is not there, it says so and hands the doubt to a person.
2. Store every question with the page, the element where it was asked and whether that session converted.
3. Group them weekly by page, the same way we do with reviews.
4. Move the most repeated ones to the fixed FAQ, hand-written and in the HTML. Nobody indexes the answer the AI generates on the fly; the fixed question is read by Google and by AI assistants.
5. Send the ones the AI cannot answer to product and copy. They are not an FAQ failure, they are gaps in the business.

One non-negotiable condition: the AI only answers with verified information. If it makes up a price or a delivery time, you do not have an FAQ. You have a sales promise nobody approved.

And a bonus: what the user types in that box is almost word for word what they ask ChatGPT before reaching your site. They are the best prompts you can have to measure whether AI assistants mention you, which is exactly what we track with Mencoro.

My takeaway: the FAQ stops being a section and becomes a sensor

The FAQ was the highest-intent spot on the whole page and we were writing it blind. With an open question and a cheap model it always answers, covers a good part of the chat and tells you, page by page, what is holding your customer back. The question is no longer whether to put AI in your FAQ. It is what you are going to do with what people type.