How to Appear in AI Assistant Answers When People Ask What to Buy

To get recommended by AI assistants, make your products easy for machines to read and easy for humans to trust: publish clear, factual product information marked up with structured data, collect genuine reviews, and earn mentions on the independent sources that AI models already read. There's no button to press and no ad to buy — you win by being the most legible, most corroborated answer to a shopper's question.

The behaviour is already shifting. Instead of typing "best waterproof hiking boots" into a search box and scrolling ten blue links, a growing number of people now ask an assistant — ChatGPT, Google's Gemini, Microsoft Copilot, Perplexity — a full question: "What are good waterproof walking boots under £120 for wide feet?" The assistant replies with a short list of specific recommendations. If your product is on that list, you get the sale. If it isn't, you're invisible in a way that's harder to fix than a poor Google ranking.

This new discipline has a name: Generative Engine Optimisation (GEO). It's the cousin of SEO, and the good news is that most of the groundwork overlaps. Do the fundamentals well and you show up in both places.

How AI assistants decide what to recommend

You can't optimise for something you don't understand, so start with the mechanics. When someone asks an assistant what to buy, the model does roughly three things:

1. It draws on what it learned during training — a compressed memory of the public web, including product pages, reviews, forum threads, and "best of" articles.

2. It often searches the live web to ground its answer in current information, then reads a handful of pages it retrieves.

3. It synthesises a recommendation, favouring products it can describe confidently with specific, corroborated facts — a name, a price, key attributes, and evidence that real people rate it well.

Notice what that rewards. The assistant isn't dazzled by your brand video or your clever homepage animation. It rewards clarity and corroboration: information it can extract without ambiguity, confirmed across more than one source. That's the whole game, and everything below is a way of playing it.

1. Give machines clean, structured product data

The single highest-leverage move is structured data — machine-readable markup (usually Schema.org, embedded as JSON-LD) that labels exactly what's on a page. Instead of hoping a model correctly infers that "£89.99" is your price and "4.6 out of 5" is your rating, you state it explicitly in a format built for machines.

For a product page, the Product schema lets you declare the name, description, brand, price, currency, availability, GTIN/SKU, and aggregate review score. This is the same markup that powers rich results in Google Search, and it's increasingly what grounds AI shopping answers too. A page that spells out its facts is far more "quotable" to an assistant than one where the price is baked into an image and the specs live in a PDF.

You shouldn't have to hand-write JSON-LD for every product. On Dirora, product pages emit structured data automatically as part of the built-in SEO Tools, so your catalogue is machine-readable from day one — and our SEO for online stores guide covers the wider fundamentals that feed the same signals.

2. Write clear, factual, genuinely useful content

Structured data tells a machine what a product is; your words tell it why the product is right for a specific person. Assistants pull from prose, so vague marketing language actively works against you.

• Lead with facts, not adjectives. "Merino wool, 200gsm, machine-washable, made in Portugal" is extractable. "Luxuriously soft, effortlessly stylish" is noise a model can't turn into a recommendation.

• Answer real buying questions on the page. Sizing, materials, care, compatibility, who it's for, who it isn't. If a shopper would ask it, an assistant will look for the answer.

• Use specifics that match how people search. "Wide-fit", "under £50", "vegan", "for sensitive skin" — the attributes shoppers put in their questions are the ones you want written plainly on the page.

Longer-form content helps too. A well-written buying guide or comparison on your blog gives assistants context they can cite, and it ranks in ordinary search at the same time. Dirora includes a Professional Blog Engine for exactly this, and our product descriptions guide goes deeper on writing listings that read well to both people and machines.

3. Collect genuine reviews and ratings

Reviews are the corroboration layer. When an assistant weighs two similar products, social proof — a solid average rating across a decent number of reviews — is a strong tie-breaker, because it's independent evidence that the product delivers. Review content is also a rich source of the natural, specific language shoppers actually use ("runs small", "battery lasts all weekend"), which helps the model match your product to nuanced questions.

Make collecting reviews routine rather than occasional: ask after delivery, keep it low-friction, and display ratings prominently. Dirora's Product Reviews & Ratings feed the aggregate rating straight into your product schema, so the same reviews that reassure a human shopper also become a machine-readable signal. If you're starting from zero, our guide to collecting customer testimonials has practical tactics.

4. Get cited by the sources AI models already read

Here's the uncomfortable truth: AI assistants often trust what other people say about you more than what you say about yourself. Their training and live retrieval lean heavily on third-party sources — editorial "best of" round-ups, comparison sites, Reddit and niche forums, YouTube reviews, and reputable publications. Being mentioned there is one of the strongest ways to enter an assistant's consideration set.

This is old-fashioned digital PR and community presence, refocused:

• Pitch relevant round-ups and gift guides. Getting into "the best X for Y" articles pays off twice — human readers and the models that ingest those articles. Our gift guides guide walks through the outreach.

• Be genuinely present in communities. Honest participation where your audience already discusses products builds the mentions models notice — see marketing your store on Reddit.

• Encourage independent reviews. A creator's hands-on video or a blogger's write-up is corroboration you can't manufacture on your own site.

• Keep your facts consistent everywhere. Same product name, same key specs across your store, your listings and any profiles. Contradictory information makes a model less confident, and a less-confident model recommends someone else.

5. Make sure assistants can actually reach you

None of this matters if the machines can't find or fetch your pages. The plumbing still counts:

• Submit a clean sitemap and verify with search tools so your catalogue is discoverable. Dirora handles Google Merchant & Sitemap Sync and Verified Webmaster Tools for you.

• Keep pages fast and technically sound. Slow, broken, or JavaScript-locked pages get skipped by crawlers and retrieval systems alike.

• Set good social and Open Graph metadata so links to your products render with the right title, image and description wherever they surface. That's built into Dirora's Social Sharing & OG Metadata.

Then measure. You won't see "AI assistant" as a tidy line in most analytics yet, but you can watch for referral traffic from assistant domains and track whether branded searches rise as your visibility grows. Dirora's Real-Time Analytics help you spot those shifts early.

The honest bottom line

GEO isn't a trick, and there's no shortcut to buy. Appearing in AI answers is what happens when your product information is clean and structured, your content is factual and genuinely helpful, your reviews are real, and independent sources vouch for you. That's also, not coincidentally, what good SEO and good marketing have always looked like — the audience reading it has simply expanded to include the machines. Build for clarity and trust, and you'll be the answer whether the question is typed into a search box or asked out loud to an assistant.

If you're setting up a store with these foundations in place from the start, our getting started guide is the place to begin, and you can see how the pieces fit together on the features page.