How to test if Perplexity recommends your product.

Why Perplexity is the engine to watch

Perplexity is the citation-forward AI engine. Its shopping answers typically include three to seven cited sources per query, based on eCommerce Insights's manual review of 200 shopping queries in Q1 2026. That citation density is roughly twice ChatGPT's. For a D2C brand, Perplexity is the highest-leverage place to win citations because each answer offers more slots, and Perplexity's growth in shopping share is the fastest among the engines eCommerce Insights tracks. The Buy with Pro agentic checkout (early access as of Q1 2026) adds a transaction layer that raises the stakes further.

Build a 10-prompt category set

The test only works with category-typical buyer prompts. Generic "best [category]" prompts produce uninformative results. Specific buyer prompts produce signal. A pattern that works:

• Three top-of-funnel category prompts ("best [category] for [use case]").
• Three feature-comparison prompts ("[feature] vs [feature] for [category]").
• Two competitor prompts ("[your brand] vs [competitor]").
• One price-band prompt ("best [category] under $X").
• One use-case-specific prompt ("[category] for [specific scenario]").

Write the prompts the way a real buyer types them. Lowercase, no brand-voice softening, no AI-prompt jargon. For a waterproof dry bag, "best waterproof dry bag for paddleboarding" beats "What are the best waterproof storage solutions for water sports enthusiasts?"

Run the test cleanly

Three controls to keep the test honest:

1. Private browser window. Avoids personalization based on your account history.
2. Logged out. Perplexity Pro answers differ from free-tier answers, particularly with Buy with Pro. Run both if you need both.
3. Stable prompt wording. Tiny rewordings can produce different answers. Use the same wording each week.

For each prompt, copy the full answer text, the citation list, and a screenshot. Manual tests take 30 to 45 minutes for ten prompts. The eCommerce Insights grader runs this in 90 seconds and stores history.

Score each prompt

Three statuses per prompt, applied to your SKU specifically:

• CITED — the product page is in the citation list, or the product is named in the answer with a link.
• BRAND — your brand is named in the answer or appears in a citation, but the specific product is not.
• MISSED — neither product nor brand appears.

Also record the citation position (which number in the citation list, 1 through 7) and the competitors that won the slots. Position-1 citations carry roughly twice the click weight of position-7 in eCommerce Insights's tracking through Q1 2026, though Perplexity does not publish per-position click data.

Read the result

A directional read for a mid-market D2C SKU in Q1 2026, based on eCommerce Insights's audit data (illustrative):

• Citation rate above 60% across ten category prompts. Strong. The SKU is on Perplexity's short list for its category.
• 40% to 60%. Working. Real optimization opportunity per missed prompt.
• Below 40%. Not registering. Entity recognition or PDP fundamentals are missing.
• 0%. Crawl access blocked, schema gap, or competitive entity dominance. Diagnose the root cause first.

What to fix per failure mode

Perplexity's failure modes are diagnosable. The common ones:

• MISSED on every prompt. Check robots.txt and crawl access first. Confirm PerplexityBot is allowed. Re-verify llms.txt. See verify llms.txt.
• BRAND-only on most prompts. The engine recognizes the brand but not this SKU. Strengthen Product schema, add an FAQPage block, rewrite the first 300 characters with quotable facts. See write PDP copy for AI search.
• CITED but at position 5+. Add third-party grounding. Perplexity weights independent review-site corroboration heavily for position. Get cited by Outside, Wirecutter, or category-specific publications.
• CITED with wrong product details. The retrieval layer is pulling from outdated content. Audit the PDP for stale specs, refresh schema, and confirm Perplexity has recrawled.

Why position matters

Perplexity's UI surfaces citations as a numbered list above the answer body. The first three citations carry visible prominence; citations beyond position five are progressively less likely to be clicked. eCommerce Insights does not have access to Perplexity's first-party click data, but inferred click weight from inbound referral traffic suggests roughly:

• Position 1: ~35% of citation clicks.
• Position 2: ~25%.
• Position 3: ~15%.
• Positions 4-7 combined: ~25%.

These percentages are directional and vary by category. The actionable conclusion: optimizing from "cited at position 7" to "cited at position 2" is typically larger lift than optimizing from "missed" to "cited at position 7." Run the test with position in mind, not just CITED/MISSED.

Cadence and stability

Test weekly. Daily produces too much noise (Perplexity's session variance is meaningful); monthly misses real shifts. After three weekly runs, you have a trend line; after eight, you have stability metrics on each prompt. A SKU that fluctuates between 30% and 60% citation rate week to week is real signal of an unstable position; a SKU stable at 50% is real signal of a workable platform to improve from.

Perplexity cites three to seven sources. ChatGPT cites one to three. The brand that wins Perplexity tends to win the engine its buyers trust most.

Frequently asked questions

Related jobs

Related tools

• Shopify SKU Visibility Grader — free, includes Perplexity checks.
• Perplexity coverage — eCommerce Insights's Perplexity-specific tracking.