The least glamorous file in your business may have just become one of your best salespeople.

I mean the product feed. The spreadsheet-shaped, attribute-heavy thing that probably gets attention only when a price is wrong or Google Merchant Center starts sending stern little warnings.

That is no longer enough. As shoppers start asking fuller questions—Which linen shirt will travel well? Which waterproof boot is actually comfortable?—the details you give a platform are helping it decide whether your product belongs in the answer. Google has been unusually blunt about this: its AI-driven shopping experiences are powered by the product data retailers provide, and a messy or incomplete feed can make products harder for customers to find (Google, April 2026).

So, yes: your product data is marketing copy now. Not the romantic, late-night, headline-writing kind. The useful kind. The kind that helps a real buyer understand what they are looking at before they lose patience and go elsewhere.

A product title cannot carry the whole conversation

For years, plenty of product listings got by on the basics: a tidy title, one flattering photo, a short description and a price. That worked reasonably well when shoppers searched in neat little fragments such as “black tote bag” or “ceramic mug Australia.”

Conversational shopping is messier in a good way. People ask for an item that suits a particular problem, person, occasion or constraint. They want the tote that fits a 14-inch laptop without looking like office luggage. They want a mug that survives a dishwasher and does not have a fussy handle. It is less a keyword hunt than a conversation with the most patient shop assistant imaginable.

Google’s newer Shopping tools lean on Merchant Center details such as fabric softness, durability and fit to match products to conversational queries (Google, April 2026). That does not mean you should stuff every product description with adjectives until it reads like a perfume ad. It means the quiet, factual details need to be present.

If you sell clothing, that might be fibre content, weight, stretch, cut, lining, care and a properly explained size guide. If you sell tools, it could be compatible models, dimensions, materials, power source and what is included in the box. If you sell services or digital products, the same principle applies on the page itself: who it is for, what happens next, what is included and what it will not solve.

The easiest test is this: could a stranger answer three sensible customer questions from your listing without emailing you? If not, the data is not finished. It is merely uploaded.

Treat the feed and the page as the same promise

Here is where smaller businesses get caught out. Someone updates a price on the website but not the feed. A product is sold out, but the listing still says it is available. The returns policy lives on a page nobody can find. Shipping time is described as “fast,” which is not a timeframe in any known calendar.

Those gaps were always annoying. In a shopping environment built around richer product answers, they are actively unhelpful. Google’s own documentation says merchant listings can surface price, availability, shipping and return information, and recommends supplying both on-page structured data and a Merchant Center feed so the information can be understood and verified (Google Search Central).

That is not an argument for turning yourself into a part-time data engineer. It is an argument for choosing one source of truth and giving it a weekly check. Your website, feed and customer-service reality should agree on the things a buyer cares about five minutes before checkout.

Start with the products that already earn attention. Do not try to perform a heroic clean-up of 4,000 SKUs in one Friday afternoon. Pick your top 20 by revenue, traffic or strategic importance, then compare each listing against the actual customer questions your team receives. You will probably find the same missing facts again and again.

Specificity is not boring when it saves a purchase

Marketing people can be suspicious of detail because detail looks so very… detailed. We want the lovely lifestyle image. We want the brand story. We want a headline with a bit of swagger.

Keep all of that. But a precise product fact is often more persuasive than another paragraph of mood.

“Soft, premium knit” is pleasant. “Merino wool, 18.5 micron, machine-washable on wool cycle” tells a buyer something they can use. “Designed for busy families” is fog. “Fits a standard 600mm dishwasher opening; installation kit included” is a decision aid.

The point is not to sound clinical. The point is to remove the little doubts that send people back to search. Your best copy should make someone feel understood; your best product data should make them feel safe to act. Different jobs, same sale.

Google is adding more Merchant Center attributes aimed at conversational commerce, including information that can answer common product questions and identify compatible accessories or substitutes (Google, January 2026). You do not need to chase every new field on day one. But you should notice the direction of travel: platforms want products described in ways humans actually use them.

Write for the question behind the question

This is the part I find genuinely interesting. Better product data is not just an SEO task in a slightly different hat. It is customer research.

Open your inbox, chat transcripts, reviews and returns notes. Look for the questions that recur:

• “Will this fit?”
• “Is it suitable for sensitive skin?”
• “Does it work with the older version?”
• “Can I get it before Friday?”
• “What happens if I change my mind?”

Each one is a prompt for a listing improvement. Add the answer where a shopper can see it: in the product specification, the description, the FAQ, the feed attribute or all of the above where appropriate. Use plain language first; technical terms can follow when they genuinely help.

This is also where a smaller business can beat a bigger, vaguer competitor. You know why a customer returns a certain item. You know the odd use case that makes one product a quiet favourite. You know which colour looks different in winter light. That knowledge is marketing material. It just needs to stop living only in your head or your customer-service team’s memory.

Your practical reset for this month

If this has made you look nervously at a product spreadsheet, fair. Here is the manageable version.

Choose five best-selling products. For each one, check the title, images, price, availability, shipping estimate, returns information, variants and the five questions customers ask most often. Make the on-page information and feed agree. Then ask someone who does not work in your business to find the answer to one of those questions. Watch where they get stuck.

That small exercise will show you more than a generic “optimise for AI” checklist ever could.

The businesses that show up well in AI-assisted shopping will not necessarily be the ones with the flashiest automation. They will be the ones whose products are described honestly, consistently and with enough care that a platform—and a person—can understand what is being sold. Not glamorous, perhaps. But neither is losing a sale because your stock status said one thing and reality said another.