Episode 20: Your FAQs Just Stopped Being Optional

FAQs have sat on product pages for years, and for most of that time they were an SEO exercise. A block of extra content at the bottom of the page, written for a crawler, skimmed by nobody.

That calculation has quietly changed.

AI models are answer engines. FAQs are answers. Which means the block at the bottom of your page has stopped being decoration and started being the thing that determines whether your product gets recommended at all.

The old FAQ was decoration

You can tell the old ones on sight, because they answer questions nobody asked.

The specification says size medium. The FAQ underneath asks whether the product is available in a size medium. It restates a value already sitting three inches up the page, adds a keyword, and helps precisely nobody.

That was fine when the audience was a search engine counting content on a page. It is useless now, because an answer engine already has the specification. What it does not have is everything the specification does not say, and that is where buying decisions actually get made.

A real example

During the UK heatwave a few weeks ago, one of us did what half the country did and went out to buy a paddling pool. That part went fine.

The problem was the filter pump. It was a large pool, so it needed one, and working out which pump fitted meant turning to answer engines. Google's AI search first, then the marketplace's own AI assistant on Amazon.

The pump that arrived was the wrong one and went straight back.

Two things caused that. The product data on the listing was thin, which is common enough on a third party marketplace seller's record. And there were no indexed FAQs attached to it.

For a filter pump, compatibility is obviously the number one question anybody asks. The compatibility information was vague, and it was vague for a reason: the brand made two different ranges of paddling pool, some that supported filters and some that did not. That distinction existed. It was never written down anywhere a machine could read it.

Nobody in that chain did anything unreasonable. A return got processed, a customer was lost, and the single fact that would have prevented both was missing from the page.

The FAQ problem is circular

Here is what makes this genuinely hard rather than merely neglected.

How do you predict every important question for a given product before anybody has asked it? You cannot, which is why FAQ sections tend to get written once at launch, from the inside out, by somebody guessing at what a customer might want to know.

Break the circle by capturing the questions rather than inventing them. If your product page lets customers submit questions and you have a route to review them, you get visibility of what people actually want to know. You can then answer internally, or let socially harvested content do it.

The irony of the current moment is that AI closes the loop from the other end too. You can now have a model suggest the questions likely to be asked for a product, and generate the answers as well. But only if you hold structured data about that product, because otherwise there is nothing for it to answer from.

Where the real questions already are

Most businesses are sitting on this information and not using it.

Return reason codes are the clearest example. Historically the only team looking at those was logistics, and only to establish whether the return was their fault and whose P&L it landed in. Nobody treated them as content.

They are a treasure trove. Major marketplaces require a written reason on every return, which means somebody has already told you, in their own words, exactly what the page failed to explain. Harvest that and feed it back into the listing and you are answering a question that has demonstrably cost you money at least once.

Customer service and sales sit on the same material, gathered from the other direction: what buyers ask before they commit rather than after they regret it.

Reviews are a third seam, and one being reworked right now. Businesses are starting to feed review content back onto product pages specifically to answer questions, rather than simply display a star rating above a wall of comments.

The work is reverse engineering. Start from the customer's view and move inwards, category by category and product type by product type, rather than starting from the attribute list you happen to hold and hoping it covers the ground.

If you do not answer it, somebody else gets cited

This is the part with a direct commercial cost, and most teams have not priced it yet.

Answer engine traffic, including the high-intent questions like what is the best product for a given job, largely does not land on manufacturer or retailer product pages. It lands on blogs, social platforms and content-heavy sites whose entire business model is harvesting social proof and getting cited for it.

When we run AI readiness audits and look at which sources are being treated as trusted, Reddit comes up constantly, because the models are hunting for social proof as a signal of trust. An entire industry of platforms has appeared just to tell brands where they are being mentioned inside LLMs and which sources are getting the credit.

The mechanism is simple enough. If the question is not answered on your page, the citation goes to whoever did answer it. You do not get a second look, and none of that traffic is recoverable after the fact.

Which makes authority the thing to build, and authority does not come from answering the obvious questions that have been answered a thousand times already. It comes from answering the ones nobody else has bothered with.

The takeaway

There is a neat piece of history in this. The best part of fifteen years ago, the reviews and ratings platforms were selling the first versions of social question and answer tools for product pages. At the time the value was real but modest, because a human still had to read the questions and write the answers.

The brands that bought in then are sitting on a backlog of exactly the content answer engines now reward. They solved a problem years before anybody could see it coming, and are being paid for it now.

The rest of the market has to build that library deliberately, and quickly.

So stop thinking of the FAQ block as content marketing bolted onto a product page. Your product page is becoming an answer engine hub, and the questions you leave unanswered are not neutral. They are directions to a competitor.

Listen to the full episode

Episode 20 of Product Data Weekly is available now. For more episodes and the weekly newsletter on operational issues inside product data and ecommerce teams, visit productdataweekly.com.