I Audited Brooklyn Candle Studio Against AI Shopping Agents. Dominant on Two Platforms, Invisible on the Third.

Executive Summary

• Brand: Brooklyn Candle Studio (brooklyncandlestudio.com). Founded in Brooklyn, operating from Industry City, Sunset Park. $33 soy wax candles with lead-free cotton wicks. Vegan, cruelty-free, phthalate-free, petroleum-free. Travel-inspired scent collections. Estimated $5-15M annual revenue. Primarily DTC via Shopify.
• AI visibility score: 23/150 tests (15%). ChatGPT 0%, Copilot 18%, Gemini 28%.
• The pattern: The most polarised visibility profile in the audit set — dominant at #1 for Brooklyn-specific queries, invisible for everything else. 74% of all surfacings come from a single query.
• Key finding: ChatGPT returns 0/50 despite frequently recommending the brand as a competitor when testing other candle brands. Trustpilot shows 0.0/5 with 742 reviews. ~40 words of every 82-105 word description are identical boilerplate. Description score: 6.0/10. Tag score: 3.2/10.
• Root cause: Boilerplate descriptions dilute product differentiation. Navigational tags mask sparse product classification. Trustpilot at 0.0/5 is actively toxic. No on-site reviews. No aggregateRating.
• Fix complexity: Medium — urgent Trustpilot investigation, description rewriting, and tag restructuring needed.

The brand

Brooklyn Candle Studio was founded in Brooklyn and operates out of a workshop in Industry City, Sunset Park. They make $33 hand-poured 100% soy wax candles with lead-free cotton wicks, premium fragrance oils, and a clean ingredients list — vegan, cruelty-free, phthalate-free, petroleum-free. The brand positioning is travel-inspired botanical fragrances, with scent collections named after destinations: Tulum, Santa Fe, Catskills, Brooklyn.

They are frequently cited as the go-to Brooklyn candle brand in editorial roundups. A Shopify success story. On paper, this is a brand that should be surfacing in every "best clean candle" and "best soy candle" query an AI agent handles.

They are not. Except when the query says "Brooklyn."

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 a good vegan soy candle brand?"
2. "Can you recommend a travel-inspired scented candle?"
3. "What's the best candle for a botanical, earthy scent?"
4. "I need sustainable candles as a gift. Suggestions?"
5. "What are the best eco-friendly candle brands in Brooklyn?"

The results

Overall: 23 out of 150 tests (15%)

The polarisation is extreme. For the Brooklyn-specific query, Gemini surfaces the brand at position #1 in 90% of tests. Copilot surfaces it at #1 in 80% of tests. This is dominant, reliable visibility. Brooklyn Candle Studio owns this query on two platforms.

For every other query type, the brand is nearly invisible. 6 surfacings across 120 non-Brooklyn tests. That is 5%.

The ChatGPT anomaly

The most surprising result in the audit set: ChatGPT returns 0/50. A complete shutout across all five query types and 50 tests.

This is not because ChatGPT does not know the brand. When I tested Apotheke's Brooklyn query, ChatGPT recommended Brooklyn Candle Studio as a competitor. It "knows" the brand. It does not "recommend" it when asked directly — even when "Brooklyn" is in the query.

This is the audit set's biggest anomaly and warrants further investigation.

The boilerplate problem

Here is the Santa Fe Jar Candle description (82 words):

"TOP NOTES: Orchid Cactus, Piñon MIDDLE NOTES: Palo Santo, Spruce BASE NOTE: Woodsmoke, Sage, Cedar A spontaneous February trip inspired this earthy, calming scent... Made in Brooklyn, New York with 100% soy wax for an eco-friendly clean burn. We use lead-free cotton wicks and premium fragrance oils infused with essential oils. Our candles are vegan, cruelty-free, phthalate-free, and petroleum-free."

The scent pyramid is good. The travel story is evocative. Those parts work.

But that last paragraph — the sustainability statement — is identical on every single product. Word for word. Across the entire catalogue. Roughly 40 of every 82-105 words are shared boilerplate. That leaves 42-65 unique words per product. That is all an AI agent gets to differentiate your Catskills candle from your Tulum candle from your Brooklyn candle.

No burn time. No weight. No dimensions. No room size guidance. Description quality score: 6.0/10.

The tag overload

Brooklyn Candle Studio uses 17-18 tags per product — the highest count in the audit set. But most are navigational: active, all, candle, candles, shop, shop all, singlecandle, minimalist-single.

Buried in there are some genuinely useful scent tags — woody, smoky, floral, gourmand — and seasonal tags like SPRING, SUMMER, winter scent. But they are diluted by the noise of catalogue management tags that add zero value for AI agent consumption.

An AI agent filtering by "woody" would find Santa Fe and Catskills but miss Brooklyn (tagged "spice" not "woody" despite having patchouli and sage). High tag count masks low tag quality. Tag score: 3.2/10.

The Trustpilot crisis

This is the finding that needs immediate attention. Brooklyn Candle Studio's Trustpilot profile shows 0.0 out of 5 with 742 reviews.

Zero stars. Seven hundred and forty-two reviews.

When an AI agent evaluates a brand, it checks multiple signals. On-site reviews are one. External review platforms — particularly Trustpilot — are another. A 0.0 rating with 742 reviews is a catastrophic signal.

The brand has zero on-site reviews. No aggregateRating in their JSON-LD. So the only review signal an AI agent can find externally is a 0.0 Trustpilot score.

This could be a data anomaly, a review bombing incident, or a genuine customer service crisis. Regardless of the cause, the effect on AI visibility is the same: it is actively toxic.

The structured data gap

Their Shopify implementation includes basic JSON-LD (6.0/10), but without aggregateRating. No on-site reviews means no star ratings in structured data. Combined with the Trustpilot situation, there is effectively no positive review signal anywhere an AI agent would look.

Why this is happening

Brooklyn Candle Studio has a strong editorial footprint in one niche — Brooklyn candle roundups and indie candle lists. Gemini and Copilot have learned to associate the brand with "Brooklyn candle" queries so reliably that they surface it at #1 almost every time.

But that editorial presence is narrow. For generic queries — vegan soy candle, sustainable gift candle, botanical scent — the brand does not appear in the roundups and listicles that AI agents draw from. The eco credentials are real (vegan, cruelty-free, phthalate-free, petroleum-free) but they are buried in boilerplate copy, not surfaced as structured data or tags.

The Trustpilot crisis compounds the problem. Even if the brand had better product data, a 0.0/5 Trustpilot score would undermine any trust signal an AI agent tries to build.

What Brooklyn Candle Studio could do, in priority order

Phase 1 (urgent):

• Claim and manage the Trustpilot profile immediately. A 0.0/5 with 742 reviews is an emergency. This is likely the single most damaging data signal for the brand. Claim it, respond to reviews, and begin requesting reviews from satisfied customers.
• Install an on-site review system. Implement Judge.me, Klaviyo Reviews, or similar. Feed aggregateRating into JSON-LD. The brand clearly has customers — capture their feedback.
• Add burn time, weight, and dimensions to every product description. Keep the scent pyramids and travel narratives, but add a "Details" section with structured product data.

Phase 2 (medium effort):

• Replace navigational tags with descriptive tags. Remove tags like "active", "all", "shop", "candles" that serve no classification purpose. Add: fragrance family, room size, occasion, season (systematically, not sporadically).
• Deduplicate the boilerplate. Move the "Made in Brooklyn... vegan, cruelty-free, phthalate-free" copy to a brand-level element. Use the freed word count for unique product data — burn time, room size, comparison context.
• Add comparison context between products. When Santa Fe and Catskills both tag as "woody," the descriptions should help an AI agent recommend one over the other.

Phase 3 (longer term):

• Investigate the ChatGPT blind spot. The 0/50 result despite strong editorial presence is unusual. Consider whether the brand's Shopify implementation has technical issues that prevent ChatGPT's crawler from accessing product data.
• Expand editorial presence beyond Brooklyn-specific queries. Build content and roundup presence for "sustainable candle gift," "travel-inspired candle," and "best soy candle" queries.
• Submit product feed to Bing Merchant Center to strengthen Copilot presence beyond Brooklyn queries.

Close

Brooklyn Candle Studio is the most polarised brand in this audit programme. Position #1 on two platforms for one query. Invisible everywhere else.

The travel-inspired storytelling and scent pyramids can stay. They are part of the brand identity and they are genuinely useful data. But they cannot carry the entire description when 40% of the surrounding text is identical boilerplate and the only external review signal is a 0.0 Trustpilot score.

The brand owns "Brooklyn candle." Now it needs to own "candle."