I Audited P.F. Candle Co. Against AI Shopping Agents. 23 Words. 909 Reviews. The Highest AI Visibility in the Audit Set.

Executive Summary

• Brand: P.F. Candle Co. (pfcandleco.com). Founded in 2008 in Los Angeles by Kristen Pumphrey. $24 soy wax candles with cotton wicks. No phthalates, no animal testing. Available at Target, Nordstrom, Urban Outfitters, and Anthropologie. Estimated $20-40M annual revenue. Shopify store.
• AI visibility score: 33/150 tests (22%). ChatGPT 0%, Copilot 18%, Gemini 48%. The highest overall visibility in the candle audit set.
• The pattern: The audit set's great paradox — worst descriptions, best visibility. 23-word descriptions (2.0/10) but 909 reviews on one product and strong editorial presence compensate. Until you check ChatGPT.
• Key finding: Every product differentiator — soy wax, vegan, cruelty-free, phthalate-free — lives on the About page, not on product pages. Descriptions are 23-36 words of lifestyle copy with scent notes only. The word "soy" does not appear in any product description. Tags score 1.8/10 with misspellings. But structured data scores 10/10 thanks to 134-909 reviews per product with aggregateRating.
• Root cause: Review volume and editorial presence drive visibility on Gemini and Copilot. But the complete absence of product data in descriptions means ChatGPT cannot surface the brand — 0/50 across all queries.
• Fix complexity: Low — adding basic product attributes to existing descriptions would fill the only gap.

The brand

P.F. Candle Co. was founded in 2008 in Los Angeles by Kristen Pumphrey, originally starting at a farmers' market. They make $24 soy wax candles with cotton wicks, no phthalates, no animal testing. Available at Target, Nordstrom, Urban Outfitters, and Anthropologie. This is not a niche brand. This is mainstream retail distribution with a craft positioning. Strong wholesale business. Amazon presence. Estimated $20-40M annual revenue.

They have the strongest review infrastructure in the audit set by an order of magnitude — 909 reviews for Amber & Moss alone at 4.49 out of 5. More reviews on one product than all other audited candle brands combined.

The test

I ran five queries across ChatGPT, Gemini, and Copilot, each repeated 10 times per platform — 150 total tests. All tests were run in incognito mode via Playwright with anti-detection measures. No authentication, no history.

The queries:

1. "What's the best scented candle brand that uses natural ingredients?"
2. "Can you recommend a soy wax candle with a woody scent?"
3. "What's a good candle for relaxation under $40?"
4. "I need a candle gift set for a housewarming. Suggestions?"
5. "What are the best California-made candle brands?"

The results

Overall: 33 out of 150 tests (22%) — the highest in the candle audit set

The California provenance query produced the strongest single-query result in the entire audit set: 19/30 (63%). Gemini surfaced P.F. Candle Co. in every single test — 10/10, always at position #1-2. Copilot surfaced it at #1 in 9/10 tests.

Gemini is the clear winner for this brand: 24/50 (48%), the highest Gemini score in the candle audit set. Gemini surfaces P.F. Candle Co. for budget queries (70%), gifting (50%), and provenance (100%). The $24 price point and massive review volume appear to be exactly what Gemini's recommendation engine values.

And then there is ChatGPT: 0/50. A complete shutout.

A brand sold at Target and Nordstrom. 909 reviews on a single product. $24 price point. Strong editorial presence. And ChatGPT cannot surface it once across 50 tests. This is the most surprising 0/50 in the audit programme.

The description problem

P.F. Candle Co. has the thinnest descriptions in the entire audit set. Automated score: 2.0/10.

Amber & Moss — their flagship product with 909 reviews:

"A weekend in the mountains, sun gleaming through the canopy. Sage, moss, lavender. TOP moss, lavender MID sage, orange, lavandin BASE amber, musk"

23 words. The scent note pyramid (TOP/MID/BASE) is genuinely useful structured information. But that is all there is.

No mention of soy wax. No burn time. No wick material. No candle weight or dimensions. No mention that it is vegan. No mention that it is cruelty-free. No mention that it is phthalate-free.

Teakwood & Tobacco — the founding product, "the O.G.," "the boyfriend scent":

"The one that started it all. Some call it the boyfriend scent, we call it the O.G. Leather, teak, orange. TOP orange, leather MID black tea, pepper, tobacco leaf BASE sandalwood, teak, patchouli, musk, amber"

35 words. Same pattern. Lifestyle sentence, scent notes, nothing else.

The invisible differentiator

Every single differentiator that would make AI agents recommend P.F. Candle Co. — soy wax, vegan, cruelty-free, phthalate-free, lead-free cotton wicks — exists on the About page. It is all there.

But AI agents do not crawl your About page when someone asks for a product recommendation. They look at the product page. And on the product page, there is nothing to find.

The word "soy" does not appear in a single product description. P.F. Candle Co. makes soy wax candles. They never say so on their product pages. They are invisible for "best soy candle" queries — surfacing just 3% of the time for the woody soy candle query.

The tag mess

Tags score 1.8/10. Two to five tags per product with extreme inconsistency. Sandalwood Rose has 2 tags (Classic, Shop P.F.). Teakwood & Tobacco has 5 tags — including misspellings: "tabac" and "tabacco." No scent family classification. No wax type. No certification tags.

The review advantage

What P.F. Candle Co. does have is the strongest review infrastructure in the audit set. Structured data score: 10/10.

• Amber & Moss: 909 reviews, 4.49/5
• Teakwood & Tobacco: 350 reviews, 4.53/5
• Golden Coast: 207 reviews, 4.26/5
• Sandalwood Rose: 134 reviews, 4.29/5

All with aggregateRating in JSON-LD structured data. This review volume is what drives Gemini's 48% surfacing rate. Reviews are the currency that matters most on Gemini.

But the Trustpilot profile is empty — 0/5 with 0 reviews. The massive on-site review volume has no external counterpart.

Why this is happening

P.F. Candle Co. proves that review volume and editorial presence can outweigh description quality — but only on platforms that weight those signals.

Gemini surfaces the brand 48% of the time. Gemini appears to weight review volume, price data, and provenance signals heavily. With 909 reviews and a $24 price point, P.F. Candle Co. has exactly what Gemini values.

Copilot surfaces the brand for California queries (90%) but nothing else. This suggests Copilot's Bing Shopping data associates the brand with California but lacks the broader product data to recommend it for other query types.

ChatGPT returns 0/50. This is the most instructive result. Despite massive review volume, strong editorial presence, and mainstream retail distribution, ChatGPT cannot recommend the brand. One hypothesis: ChatGPT may associate the products with retailer pages (Target, Nordstrom) rather than pfcandleco.com, or its recommendation engine may require richer product descriptions than the 23 words currently available.

The audit set's paradox in full: Apotheke has the best descriptions (9/10) and 14% visibility. Earl of East has the best structured data (10/10) and 11% visibility. P.F. Candle Co. has the worst descriptions (2/10) and the highest visibility (22%). Description quality is not the most important variable. Social proof is — at least on the platforms that can read it.

What P.F. Candle Co. could do, in priority order

Phase 1 (quick wins):

• Add "soy wax" to every product description. This is a one-line fix that could unlock an entire query category.
• Add product details to descriptions. Even 50 additional words per product would transform visibility: burn time, weight, dimensions, wax type, wick material. The vibes copy can stay. Just add the facts.
• Standardise and expand tags. Fix misspellings (tabacco to tobacco). Add scent family, room size, occasion, and season tags to every product.

Phase 2 (medium effort):

• Add additionalProperty fields to JSON-LD. Burn time, wax type, scent notes, and fragrance family as structured properties.
• Create a Trustpilot profile. The brand has massive on-site review volume but zero external review presence. Even redirecting a fraction of review requests to Trustpilot would build a strong external signal.
• Move sustainability claims from the About page to every product page. Soy wax, vegan, cruelty-free, phthalate-free, lead-free cotton wicks — on every product.

Phase 3 (longer term):

• Investigate the ChatGPT blind spot. The 0/50 result for a brand of this size is anomalous. Check whether ChatGPT's crawler can access the Shopify store. Consider whether the brand's wholesale focus means ChatGPT associates products with retailer pages rather than pfcandleco.com.
• Build a buying guide page comparing all scent families. "Amber & Moss for the cabin; Teakwood & Tobacco for the living room; Golden Coast for the bathroom" — structured content that helps AI agents recommend the right product.

Close

P.F. Candle Co. is the audit set's great paradox. 23-word descriptions. 909 reviews. The highest AI visibility score in the candle set.

But that visibility is fragile. It depends entirely on review volume and editorial presence — signals that work on Gemini (48%) and partially on Copilot (18%), but not at all on ChatGPT (0%).

The fix is not a strategic overhaul. It is adding the word "soy" to a product page. It is putting burn time next to the scent notes. It is moving the About page claims to where they actually matter.

Sometimes the difference between being recommended and being invisible is not a strategy. It is a single word.