Understand what AI Overviews cite for product queries
For buying questions like 'best X for Y', 'is X worth it' or 'what size X do I need', the AI answer stitches together a few sources it trusts. Some of that comes from your product page directly. A lot of it comes from a supporting article, a review site or a comparison page that mentions your product. Both routes matter, and both start with you already ranking near the top, because the box mostly draws from page one.
So the honest first step is unglamorous: the product page has to rank for the query before it can be quoted in the answer above the results. If it's on page three, fix that first, then worry about the answer box.
Lead the description with the buying answer
Most Shopify descriptions open with brand mood-setting: the story, the craft, the vibe. The AI has nothing quotable to grab, so it skips you. The pages that get lifted answer the buying question in the first two or three sentences: who it's for, what problem it solves, the one spec that decides the purchase.
Put the plain answer at the top, then expand into the story underneath. 'This is the best option for [use case] because [reason], available in [sizes], from [price].' That self-contained sentence is exactly the kind of passage an AI Overview pulls and attributes to you.
Put the real specs in a table, not a paragraph
AI Overviews lift structured facts far more readily than prose. Dimensions, materials, compatibility, weight, care, what's in the box: put them in a tight spec table or a clean bullet list, not buried in a flowing paragraph. A machine reads a table without guessing, and product answers lean heavily on those concrete facts.
This is also what wins the 'what size do I need' and 'does it fit X' style questions, which are pure AI-Overview territory because they have a single correct answer the box can state.
Add Product, Review and FAQ schema, and keep it valid
Structured data is how you tell Google exactly what the page is without it interpreting your layout. Product schema with real offers and price, aggregateRating and genuine Review nodes, plus an FAQ block covering the actual questions buyers ask about that product. That combination makes the page trivially easy to parse and quote.
Validate it, don't just add it. Broken or mismatched schema (a rating with no reviews, a price that doesn't match the page) gets ignored at best. On Shopify, most themes and review apps can output this, but check what's actually rendering rather than assuming the app did it right.
Build the supporting article that does the citing
Here's the part most stores miss: a big share of product citations in AI Overviews come from a 'best [product] for [use case]' article, not the product page itself. The AI trusts a page that compares and recommends more than a page that's obviously trying to sell one thing. So write that article, recommend your product honestly within it, and link straight to the product page.
Now you have two shots at the box: the product page as a direct source, and the article as the recommendation that names it. The blog earns the citation, the product page takes the sale.
Make sure reviews and off-site mentions back you up
AI Overviews synthesise sentiment, so thin star-only reviews add nothing. Reviews with actual sentences, specific and varied, give the answer something to summarise and a reason to treat your product as a safe recommendation. Quantity plus substance.
Off the page, being named in roundups, comparisons and forum threads makes the AI more confident recommending you. On-site clarity plus off-site mentions is the same combination that surfaces a store in AI answers, applied at the product level.
Don't block the AI crawlers
None of this matters if the bots can't read the page. Check your robots.txt isn't blocking Google-Extended, GPTBot, ClaudeBot or PerplexityBot, and confirm your product copy renders server-side. Shopify serves content server-side by default, but some review, upsell or 'read more' apps inject the important text with JavaScript, which the AI may never see.
If the buying answer only appears after a script runs or a tab is clicked, treat it as invisible and move it into the base HTML.
What to do first
Pick the two or three product queries where an AI Overview is already eating your clicks. Confirm you rank for them (fix that before anything else). Rewrite those descriptions to lead with the buying answer, add a spec table, get Product plus Review plus FAQ schema validated, and write one supporting 'best for' article that links back. Then re-check the box in a few weeks.
You're not gaming the AI. You're making your product page the easiest, clearest, best-backed answer to the question, which is exactly what it wants to cite.