You typed your own category into Perplexity last week, watched it recommend three competitors, and none of them were you. That stings. It's also more fixable than it feels, because Perplexity Shopping builds its recommendations from product data it can parse, reviews it can weigh, and coverage it can trust. You control all three.
This guide walks you through eight steps of ecommerce Perplexity AEO, in the order you'd actually do them. By the end you'll have a source map, clean structured data, a live product feed, and a way to check whether any of it moved.
What you need: access to your product page templates, your review platform, your product feed or Shopify admin, and about two weeks of calendar time.
Let's start where the leverage is.
Step 1: Map the shopping prompts your buyers actually type
Perplexity Shopping optimization is scored against real buyer questions, so those questions are your measuring stick for everything that follows.
Write out 20 to 30 queries a shopper would type when they're close to buying. Not your category head term. The specific, attribute-loaded ones: "lightweight running shoes for flat feet," "non-toxic cookware for glass-top stoves," "organic face serum for combination skin." Sort them into three buckets: category queries, use-case queries, and comparison queries.
Then run each one in Perplexity and screenshot what comes back. You're looking at three things: which products appear in the card set, which sources Perplexity cited to build that answer, and where you sit if you appear at all.
You're done with this step when you have 20 to 30 prompts, each mapped to a bucket and a baseline screenshot.
Where people go wrong: tracking only the big category terms. Those are the hardest queries to win and the least likely to convert. The long-tail, attribute-rich questions are where a smaller brand actually has a shot, because the shopper has narrowed the field for you.
If staring at a blank prompt list feels like a wall, this is one place software genuinely helps. DeepSmith's AI Visibility module includes Discover Prompts, which generates a starter prompt set from your product, persona, and buyer-stage context, so you're editing a populated list instead of inventing one from scratch.
Step 2: Find out where Perplexity is getting your product data today
Before you fix anything, find out what Perplexity already knows about you and where it learned it.
Perplexity Shopping sources product data from six streams, and each one gives you a different amount of control:
- Merchant Program feeds. Product feeds submitted directly by enrolled merchants. The highest-fidelity signal, because you control completeness and accuracy.
- Shopify Catalog syndication. Shopify merchants who opt in have their product data synced automatically to connected AI platforms, Perplexity included.
- Public web crawl. Perplexity's crawler reads product detail pages across the open web, parsing schema.org Product markup in JSON-LD, Open Graph product tags, and visible on-page data.
- Editorial and third-party sources. Buying guides, review sites, and roundups, which supply both product data and citation authority.
- Marketplace listings. Products on Amazon, Walmart, Best Buy, Target, and similar marketplaces.
- Direct retailer integrations. Brands selling through an integrated retailer or merchant network get carried along without separate feed work.
Go back to your screenshots from Step 1 and label the cited source for every product that appeared, yours and your competitors'. Was it a product page? A marketplace listing? A roundup on a publication you've never pitched?
You're done when every prompt has a source label and you can see the pattern in one glance. That map tells you which of the six Perplexity Shopping sources is actually speaking for your brand right now, and which ones you've left empty.
Common mistake: assuming your own site is the source. For plenty of brands it isn't. Perplexity is quoting an Amazon listing with a two-year-old price, or a "best of" article that ranked you fourth. You can't fix a source you haven't identified.
Step 3: Complete your Product schema on every product page
If you only fix one thing this month, make it this. Complete structured data is the highest-leverage technical move you can make to get products cited in Perplexity, and it's the one most catalogs half-finish. Observational analysis of Perplexity product recommendations points the same direction every time: products with complete, valid Product schema get cited noticeably more often than products with partial markup.
Put JSON-LD on every product detail page with four blocks working together:
- Product: name, image, description, sku, gtin (gtin8, gtin13, or gtin14), brand, color, size, material, weight.
- Offer: price, priceCurrency, availability (InStock, OutOfStock, or PreOrder), url, priceValidUntil.
- AggregateRating: ratingValue, reviewCount, bestRating, worstRating.
- Review: individual review entries where you have them.
Add the supporting layer too. BreadcrumbList for navigation, Organization at the site root, sameAs links pointing to your social profiles, Wikidata, and Crunchbase to reinforce your brand as a recognizable entity, and FAQPage schema wherever product education helps someone buy.
Three implementation rules that decide whether any of it counts. Use JSON-LD, not microdata. Place it in the page head or at the top of the body. Make sure every value in your schema matches what a human sees on the page, because mismatches read as an untrustworthy source.
You're done when every product template passes the Google Rich Results Test and the Schema.org validator, and you've re-validated after your most recent template change.
Where people go wrong: shipping schema that only exists after JavaScript runs. If the markup isn't in the raw HTML, it may as well not be there. Curl one of your product URLs and read what comes back before you call this finished.
Step 4: Send Perplexity your product feed directly
Crawling is Perplexity guessing at your catalog. A feed is you telling it. If you sell and ship to US customers, the Merchant Program is the strongest lever available to you, and it costs nothing.
The program is free to join. No listing fees, no commissions, no per-transaction charges under current terms. To qualify you need to sell and ship to US customers, provide a product feed as CSV following the Google Merchant Center specification (or an equivalent structured feed), have Product structured data on your product pages, and support a recognized checkout integration.
What you get back: inclusion in Perplexity's product index with priority placement over non-enrolled products, eligibility for product card recommendations inside answers, a merchant dashboard with search and shopping trend data, free API access to build Perplexity-powered search into your own properties, and eligibility for Buy with Pro checkout.
On Shopify? Your path is short. Opt in to Catalog syndication in your Shopify settings, then confirm your theme actually emits Product JSON-LD. Default themes do. Third-party themes need auditing. Make sure titles, descriptions, prices, GTINs, variants, and inventory are accurate in Shopify itself, because that's what gets syndicated.
Not on Shopify? Build the feed. Required attributes are id, title, description, link, image_link, price, availability, gtin, brand, mpn, condition, and google_product_category. Add the recommended ones: sale_price, sale_price_effective_date, additional_image_link, item_group_id, color, size, gender, age_group, product_type, plus shipping and tax attributes. Host it on SFTP or a secure endpoint Perplexity's ingestion can reach.
Pro tip: Perplexity hasn't published a required refresh cadence. Daily is the safe default, and if your prices or inventory move fast, go hourly. A feed that disagrees with your live page is worse than no feed, because pricing discrepancies can disqualify a product from cards entirely.
You're done when your feed is accepted and refreshing on schedule without errors.
One honest note. The Merchant Program and Buy with Pro are US-only right now. If you sell elsewhere, you can still be recommended through structured data, marketplaces, and editorial coverage. You just can't enroll yet, and that's worth knowing before you spend a sprint on it.
Step 5: Rewrite product copy so it answers a specific question
Your product page has to earn the recommendation on relevance, not just exist in the index. Perplexity product recommendations skew toward products whose data clearly matches the shopper's actual question, which means generic copy loses to specific copy every time.
Start with titles. The format that works: Brand, product type, key differentiator, size or quantity. "Brand X Merino Wool Running Socks, Lightweight Cushion, 3-Pack" beats "Brand X Performance Socks" because it hands the retrieval layer four attributes instead of one.
Then the description. Front-load the first 100 words with the answer to the query you mapped in Step 1, in natural language a person would say out loud. Attributes, use cases, and buyer-relevant detail come after. Keep paragraphs to two to four sentences. Put specifications in scannable bullets. Add a short question-and-answer block covering fit, sizing, care, and shipping, marked up with FAQPage schema.
Two rules that quietly matter. Write unique copy for every SKU instead of pasting manufacturer boilerplate, since duplicated text gives the retrieval layer nothing to distinguish you by. And give every image descriptive alt text, with a high-resolution main shot on a clean background showing the whole product.
You're done when your top 20 SKUs are rewritten and each one maps to at least one prompt from Step 1.
Where people go wrong: writing for the category when shoppers ask by use case. "Premium running socks" answers nothing. "Built for flat feet" answers a question someone is typing right now.
Step 6: Build review depth, not just review count
Reviews do double duty here. They feed your AggregateRating schema, and star ratings show up on the product card a shopper sees.
Here's the part that should take pressure off: volume is not the game. Depth, specificity, sentiment, recency, and use-case context carry more weight than raw star count. A product with 200 detailed reviews can outrank one with 2,000 shallow ones. You don't need to catch up to the market leader's review count. You need reviews that say something.
So build the habit. Send a soft review prompt seven days after delivery and a final reminder at day 21. Use a verified-purchase platform such as Trustpilot, Yotpo, Judge.me, Reviews.io, or Loox so the reviews are credible and feed both your on-site display and third-party aggregation. Pull customer photos and use-case context onto the product page. Make sure your AggregateRating values update dynamically instead of sitting frozen at whatever they were the day you launched.
You're done when review velocity is steady and your schema ratings move on their own.
One line worth reading twice: never gate reviews, offer incentives, or selectively filter out the critical ones. That runs against FTC endorsement guidelines, and it erodes exactly the trust signal you're trying to build.
Step 7: Earn the third-party coverage Perplexity trusts
Perplexity weighs what other people say about your product, not only what you say. Ecommerce Perplexity AEO stalls here more than anywhere else, because this is the step you can't ship in an afternoon.
The sources that carry weight on shopping queries are consistent: Reddit threads where people ask buying questions, YouTube product reviews and demos, manufacturer pages, top-tier news outlets, and vertical review sites in your category.
Four moves, in rough order of effort:
- Get onto roundups. Find the "best X for Y" listicles already being cited for your Step 1 prompts and pitch inclusion. You already know which ones matter, because you labeled them in Step 2.
- Seed products to reviewers. Send units to YouTube creators in your niche and let the review be honest.
- Pitch trade press. Vertical publications and trade outlets punch above their traffic on category queries.
- Show up on Reddit as a person. Answer real buying questions in relevant subreddits, disclose who you are, and never astroturf. Fake enthusiasm gets caught and costs more than it earns.
Marketplace presence belongs in this step too. Listings on Amazon, Walmart, Target, Best Buy, and category-specific marketplaces like Etsy, Newegg, or Sephora work as citation sources and as indirect authority signals. Keep your seller name, address, and phone consistent across every profile so the entity behind them is unmistakably one brand.
You're done when you've landed three new high-authority placements in a quarter. Authority compounds, so one placement on a genuinely authoritative site beats a dozen scattered mentions.
Step 8: Track citations and fix what slips
You've changed a lot. Now find out whether any of it moved, because AI answers shift and a single check tells you almost nothing.
Track five things: citation rate for your Step 1 prompts, your position within the product card set when you do appear, referral traffic from perplexity.ai in GA4 and your server logs, share of voice against named competitors, and your catalog-wide review volume and rating distribution.
The free version works. Run your prompt list weekly, log which products appear and which sources Perplexity cited, and watch the trend rather than any single answer. It takes an hour and it beats guessing.
When that hour a week stops scaling, that's the point to bring in software. DeepSmith's AI Visibility module tracks mention rate, citation rate, and share of voice with a per-platform breakdown, plus a Pages view showing which of your own pages AI actually cites and a competitor view showing who wins your prompts and on which exact pages. Perplexity tracking starts on the Grow plan at $199 per month, or $160 per month billed annually, and there's a 7-day free trial with no long-term contract.
You're done when you have four weeks of data on the same prompt set and can name your three weakest queries.
Where people go wrong: checking once, seeing nothing, and concluding the work failed. Feeds re-ingest, crawlers re-visit, and roundups get re-crawled. Give it four weeks of consistent measurement before you judge it. Perplexity Shopping optimization is a loop, not a launch.
What to do next
Go back through the eight steps and pick the two you haven't done.
For most brands those two are schema completeness and the feed, in that order. Schema is the cheapest and it's usually the most broken. The feed is the one that turns guessing into telling.
If the list feels long, that's normal. Almost nobody has all eight running when they start, and the first two close fast. Take it one step at a time. Momentum matters more than a perfect rollout.
Good news about the order: everything you do to get products cited in Perplexity compounds. Clean schema makes your feed easier to trust. A trusted feed makes your product easier to match. Better matching earns the roundup mentions that raise your authority. Nothing you do here is wasted on a single query.
Then re-run your prompt list in four weeks and compare it to those Step 1 screenshots. That comparison is the whole scoreboard.
If you'd rather not run the measurement half by hand every week, that's what a platform is for. DeepSmith tracks where you show up across AI engines and produces the on-brand content that closes the gaps it finds, from the same data. Start a free trial and see your real numbers before you decide anything.
You're closer than that first screenshot made you feel.
