Your Customers Are Asking AI Where to Buy. Is Your Business in the Answer?

The question has changed

For twenty years, the shopping question on Google looked the same: a few keywords typed into a box, a page of blue links, and the customer doing the comparing. That is no longer how a growing share of your customers shop.

They now ask full questions — "which ceiling fan brand is most reliable for daily use", "best mixer grinder for a family of five" — inside Google's AI Mode, and the AI answers directly: named products, where to buy them, what other buyers say. It draws those answers from the Shopping Graph, product feeds, reviews, and business profiles. If your products and your business details are not in the data the AI reads, you are not in the answer. There is no page two to fall back on.

This is not about running ads or hiring an SEO agency. It is about the boring accuracy of your business information — product names, prices, stock status, reviews, opening hours — because the AI can only recommend what it can see clearly.

Two September announcements worth your attention

Google spent September turning this shift from a prediction into infrastructure.

On September 16, the company rolled out agentic commerce updates for merchants ahead of the holiday shopping season. The headline for small businesses: "AI performance insights" in Google Merchant Center are now generally available in Australia, Canada, India, New Zealand, and the US. The report shows how your brand compares with other brands across AI Mode and AI Overviews — your share of voice in AI-generated shopping answers. Google's own line on the launch: "To succeed in this new era of commerce, businesses need to know how customers are discovering them through conversational experiences."

The same update pushed retailers to submit richer product information: conversational attributes and video attributes for their products. During testing, Google said conversational attributes supplied by Lululemon appeared in relevant AI Mode product recommendations about half the time. In plain terms, the extra plain-language details you write about your products — who they suit, what conditions they handle, what they do well — are now directly feeding the AI's recommendations. Google also said merchants adopting core Merchant Center feed best practices saw a 5% average increase in conversions the following month.

Then, on September 28, Robby Stein, Google's VP of Product for Search, announced on X that AI Mode's information monitoring is rolling out globally to all users. Until then it had been limited to paying AI Pro and Ultra subscribers through the summer. The feature is simple: you tell AI Mode what you want it to watch, and Search keeps checking for changes — websites, forums, social posts, real-time data, and the Shopping Graph — and tells you when something new appears. Stein's own examples were deliberately mundane: new restaurants or pop-ups opening nearby, local activities for families, and — the one that should interest every retailer — back-in-stock and price-drop updates.

Put the two together and the direction is unmistakable. Shopping is moving from "search and click" to "ask, get watched, and buy". And in both moves, the AI needs the same thing: clean, complete, current business data. The feed you neglect is now the recommendation you lose.

What "being in the answer" actually requires

Forget ranking tricks. An AI answer is a synthesis, not a list, and it is built from a handful of sources:

Your Merchant Center product feed — the free product listings most stores have never set up. Your Google Business Profile — hours, categories, photos, service details. Your website content — product pages written the way customers actually talk. Reviews and ratings — what real buyers say, and whether you respond. Product attributes — title, description, images, price, availability, and now conversational details like "handles voltage fluctuations well".

None of this is new data. It is data you already have, made complete and consistent. A ceiling fan listed with a real product name, an accurate price, in-stock status, and three clear photos will beat a "better" fan whose listing is a model number and a grey square.

What to ignore: anyone selling you an "AI SEO package" they cannot explain in plain language. If a tactic cannot be described without jargon, it is not a tactic — it is an invoice waiting to happen. The work below is unglamorous and it works.

The practical use case: one month at a home-appliances store

Picture a family-run home-appliances store — fans, coolers, mixer-grinders, small kitchen appliances — trading for fifteen years, steady walk-in traffic, a trickle of online orders. Nobody on the team has touched the store's Google listings since they were set up years ago. Here is what a month of AI-visibility work looks like, one week at a time.

Week one is claiming the basics. Verify the Google Business Profile: correct hours, correct categories, real photos of the shop floor. Create the Merchant Center account if there isn't one, and connect the product feed with free listings. Most stores discover two things in this week: the profile is half-complete, and the product feed does not exist at all. That gap is the whole opportunity.

Week two is fixing the feed. Every product gets a real title — what a customer would say out loud, not the distributor's SKU. Accurate price, honest in-stock status, two or three clear photos, and the attributes that actually matter: capacity, wattage, warranty period. The description gets plain-language details: who the product suits, what it does well, what it is not for. Discontinued models come out. This is slow, tedious work, and it is the single highest-value week of the month.

Week three is reviews and answers. Ask recent happy customers for a Google review at the billing counter — a small printed card with a QR code does the job. Answer every existing review, including the grumpy ones; AI reads the responses too, and so do humans. Fill out the Business Profile Q&A with the questions customers actually ask on the phone: delivery areas, installation, exchange policy.

Week four is looking at the data. Open Merchant Center's AI performance insights and read your share of voice across AI Mode and AI Overviews against similar brands. Note which products and attributes appear, and which gaps the report flags. Fix the biggest gap first. Then repeat monthly — this is bookkeeping now, not a project.

Two honest notes. First, this is a month of boring administration, not magic; if your feed was empty, the improvement is real but it compounds over quarters, not days. Second, the insights report only measures what Google measures. It is a compass, not a scoreboard — use it to find the next gap, not to declare victory.

The mundane work this creates (and how to stop doing it by hand)

Here is the catch the announcements skip: feeds rot. Prices change, stock moves, new models arrive, discontinued ones linger. A 200-product feed maintained by hand gets updated for a month and then quietly abandoned, because updating it is exactly the kind of repetitive chore that loses to every urgent task in the shop.

This is mundane work, and mundane work is what automation is for. A scheduled sync from your billing or inventory system — or even from a carefully maintained Google Sheet — into Merchant Center keeps prices and availability honest without anyone remembering to do it. New products can be added to the feed the same way new stock enters the shop. Out-of-stock items can be paused automatically instead of embarrassing you in an AI answer. Review requests can go out after every sale without the billing counter remembering.

This is the kind of plumbing we build in our workflow automation projects: nothing flashy, just the boring connections that keep your public data trustworthy while you run the business. The AI visibility work above is worth doing once; automating its upkeep is what makes it stick.

Three honest caveats

First, being in the answer does not guarantee the sale. AI recommendations cite their sources and customers still compare. Good data gets you considered; it does not get you chosen. Your prices, your service, and your reviews still do the closing.

Second, feed quality beats feed size. Forty accurate products will outperform four hundred stale ones every time. Google's own testing rewarded attribute completeness — the Lululemon example again — not catalogue volume.

Third, AI answers keep changing. Google made two significant moves in a single September, and the features, reports, and even the vocabulary will keep evolving. Treat AI visibility as a routine like bookkeeping, not a one-time project. The stores that win are the ones whose data is still accurate in March, not the ones who sprinted in October.