Perplexity is the AI engine most brands are sleeping on. ChatGPT has the volume, AI Overviews has the SERP real estate, and Claude has the prestige. Perplexity has something more useful for merchants: it actually names you.
In our State of AI Shopping Citations 2026 pilot, Perplexity Sonar mentioned monitored brands in 86% of successful answers, returned 5–8 named citations per response, and ran at $0.022 per response - the cheapest engine on the panel. The headline finding: on shopping-intent prompts, Perplexity is the lowest-cost, highest-yield, most-attributable AI surface in the market.
Buy with Pro - Perplexity's integrated shopping experience for Pro subscribers - is the surface where that citation behavior translates into commerce. It is a small market today but a high-intent one, and the optimization playbook is concrete enough that brands can act on it now.
What Buy with Pro actually is
Buy with Pro is the in-product shopping flow inside Perplexity that, for eligible products, lets Pro subscribers complete a purchase without leaving the Perplexity surface. The user asks a product-intent question, Perplexity returns a sourced recommendation set with embedded merchant cards, and (where supported) the Pro user can transact directly. For products that don't support in-context checkout, the cards still surface - with a click-out to the merchant site - making them functionally equivalent to a high-CTR ad slot with editorial framing.
The pipeline underneath is Sonar - Perplexity's retrieval-augmented model - plus a merchant-discovery layer that prioritizes:
- Products with valid Schema.org Product data on the live PDP.
- Merchants with clean, fetchable PDPs (no aggressive blocking of PerplexityBot).
- Editorial corroboration in the cited sources (reviews, comparison content, expert roundups that name the product).
- Price and availability data that matches what the engine sees on the live page.
That third bullet is what most brands miss. Perplexity does not just read your PDP. It reads the ecosystem of content that talks about your product- and weights that ecosystem when deciding which products to surface. A great PDP without editorial corroboration is a quiet PDP.
What Perplexity reads on your domain
Perplexity Sonar's on-domain crawler is friendly to standard SEO infrastructure. It reads:
- Schema.org Product markup, particularly the nested Brand object, Offer (with price + availability + priceValidUntil), and AggregateRating where genuine.
- Canonical URLs. Sonar dereferences canonicals aggressively - if your variant URLs canonicalize to the master PDP, Sonar will read the master.
- Clean HTML. Single-page-app PDPs that require JS execution to render content are read inconsistently. Server-side rendering or progressive enhancement is the durable choice.
- llms.txt. Of the major engines, Perplexity is the one we see most consistently dereferencing the lightweight Markdown index file. A well-curated llms.txt pointing at your category pages and flagship PDPs is the cheapest cross-engine lift available, and it lands hardest on Perplexity.
The full technical playbook lives on our Perplexity AI SEO services page. The summary: standard Google-friendly technical SEO does most of the work, with two specific exceptions covered below.
The Perplexity-specific exceptions
Exception 1: editorial corroboration is non-negotiable
Perplexity returns 5–8 citations per answer. The brand-named-in-the-answer slot is almost always a product whose name appears in at least one of those cited sources. Practically: if your product is not mentioned in any of the editorial coverage Perplexity surfaces for your category, you are not in the candidate set, no matter how clean your schema is.
This makes off-domain citation work - the slow, compounding kind - the most undervalued lever for Perplexity. We work with eCommerce clients on placing case studies, comparison content, and expert-byline pieces in the publications Perplexity is already citing. The signal of success is straightforward: query the panel before and after placement, watch your brand's mention rate move.
Exception 2: Sonar rewards passage clarity
Sonar paraphrases at paragraph granularity. A page that buries the buyer-relevant attributes in a 1,200-word marketing essay will get paraphrased poorly or not at all; a page that opens with a tight, factual, attribute-dense paragraph will get cited verbatim.
The discipline:
- First paragraph after the H1: the elevator pitch. What it is, who it's for, the three attributes that matter.
- Second paragraph or section: the spec or attribute table. Sonar likes structured comparison data and will paraphrase it cleanly when asked attribute-shaped questions.
- Third section: trust signals. Where it's made, return policy, what's included, who else has reviewed it.
This is not "AI-optimized copy" in the gimmick sense. It is good ecommerce copy that happens to also be the format Sonar handles best.
Pricing, availability, and the feed cross-check
Perplexity's Buy with Pro flow leans heavily on price and availability matching. The engine appears to compare:
- The price in your Product schema's Offer block.
- The price visibly rendered on the PDP.
- The price in your Google Merchant Center feed (when accessible).
- The price in any cited comparison page.
When those four disagree, Perplexity hedges. Sometimes it shows the lowest. Sometimes it shows no price at all. Sometimes it suppresses the merchant from the Buy slot entirely while still surfacing the brand in the editorial citations. The fix is operational, not editorial: get your feed and your PDP into lockstep, validate weekly, treat schema price as a contractual claim, not a marketing input.
Availability of product reviews
Perplexity weights review presence and review recency more than any of the other engines we instrument. A flagship PDP with 30+ recent reviews (within 90 days), valid schema-level AggregateRating, and visible review widgets earns Perplexity citation at meaningfully higher rates than the same product without that signal. If you are running a brand-new SKU, the first thing to do post-launch is line up review velocity- sample programs, post-purchase email flows, anything that generates real reviews from real customers.
What about brand-comparison queries?
This is where Perplexity is the most useful citation surface for many brands. When a user asks "best [category] for [use case]," Perplexity is the engine most likely to return a named-brand recommendation set with sourced citations. That is the slot you want - and the optimization is the combination of:
- A category landing page with strong topical authority, clean schema, and a structured comparison block.
- Editorial corroboration: third-party comparison pieces that name your brand in the same context as the query attribute.
- PDP-level product data that the engine can verify on the live page.
If you operate in an agency-comparison vertical (the slot where we measured our own brand's performance in the report), Claude is actually a stronger citation surface than Perplexity in our pilot - a finding we covered in the State of AI Shopping Citations 2026 report. For brand-recommendation prompts in product categories, Perplexity wins.
Common questions
Does Perplexity use llms.txt?
Perplexity has not formally committed to reading llms.txt. We observe Perplexity behaving as if it does - the cleanest evidence is that brands with curated llms.txt files have surfaced consistently in Sonar queries about brand identity ("what does [brand] sell") faster than brands without. That's correlational, not causal, but the cost of shipping llms.txt is so low and the cross-engine benefit so broad that we recommend it as default.
Should I run Perplexity Ads if I'm already cited organically?
Different surface. Perplexity Ads (the sponsored slot inside the answer) is a separate paid channel from the organic Sonar citation flow. The two are not exclusive - brands cited organically can run ads and brands running ads can be cited organically. The question is the usual one: does the incremental click from the ad slot pay back better than the next unit of citation-share investment? In most cases we instrument, organic citation work pays back longer; ads accelerate top-of-funnel discovery while the organic work compounds.
Is this the same as ChatGPT Shopping?
No. ChatGPT Shopping is OpenAI's product surface; the optimization playbook is different- heavier on schema, heavier on feed consistency, less on editorial corroboration. Perplexity rewards the on-domain cleanliness too, but the editorial layer matters more on Perplexity than on ChatGPT.
Is my catalog too big for this?
If you have 50,000 SKUs, you cannot meaningfully optimize each one for Perplexity citation. Pick 30–50 flagship products and 15–25 category pages. Those become your Buy with Pro priorities; the long tail benefits from the same on-domain hygiene by default. We cover the prioritization framework as part of an AEO audit.
Key takeaways
- Perplexity Sonar is the highest-mention-rate AI engine in our pilot (86%) and the most-attributable surface for measuring AI-driven citation.
- Buy with Pro rewards merchants with valid Product schema, clean PDPs, fetchable HTML, and consistent feed-to-page price/availability data.
- Editorial corroboration in cited sources is the under-rated lever. If no third-party content names your product, no amount of schema will earn the slot.
- Front-load buyer-relevant attributes in the first paragraph. Sonar paraphrases at paragraph granularity; structured comparison blocks get cited verbatim.
- Review presence and recency carry unusual weight. Treat post-purchase review velocity as a Perplexity-specific KPI.
For a citation-monitoring program that tracks Perplexity Sonar performance against your priority queries weekly and ties the data to schema, content, and editorial-corroboration recommendations, that's what Workspace does. Start with our Perplexity shopping optimization page, or book a call.