We Audited Finisterre Against AI Shopping Agents. The Results Surprised Me.

Executive Summary

• Brand: Finisterre. Sustainable outdoor/surfwear, founded 2003 in Cornwall. B Corp certified. GBP 23M turnover.
• AI visibility score: 9/15 queries surfaced the brand. The highest in this audit set.
• The pattern: Finisterre dominated brand-level queries (sustainable UK outdoor clothing, Patagonia alternatives) but vanished on product-specific queries (waterproof jacket in a price range, fleece for winter surfing). 6/6 on brand queries. 1/6 on product queries.
• Key finding: Zero JSON-LD structured data on any product page. A GBP 250 waterproof jacket with specs trapped in JavaScript. Okendo reviews installed but rendering via JavaScript only, invisible to crawlers. 1,103 Trustpilot reviews at 4.7/5, disconnected from products.
• Root cause: Brand-level visibility is carried entirely by external signals: B Corp certification, 20 years of press coverage, Trustpilot, and brand mentions. Product-level visibility requires product-level data. Finisterre has the data - it's just trapped where machines can't read it.
• Fix complexity: Medium. The structured data gap is the most technical fix in this audit set (JSON-LD has been removed entirely, not just left at defaults). But the product description and attribute enrichment follows the same low-effort pattern as other brands.

The brand

Finisterre was founded in 2003 in St Agnes, Cornwall, for cold-water surfing and coastal living. They were the first European surf brand to achieve B Corp certification in 2018. Four retail stores across the UK, GBP 23M turnover, roughly 280,000 Instagram followers.

Their positioning: functional performance clothing with sustainability at the core. Recycled materials, Bowmont wool, circular economy programmes, a repair service. They compete with Patagonia, Rapanui, and Passenger in the UK sustainable outdoor market.

I selected Finisterre expecting it to be the first brand in this audit set that performed well. The B Corp certification, the press coverage, the heritage. Surely the AI agents would know them.

They did. And they didn't.

The test

I ran 5 queries across ChatGPT, Gemini, and Copilot. 15 tests total.

The queries:

1. "What's a good waterproof jacket for coastal walks in the UK?"
2. "Can you recommend sustainable outdoor clothing brands from the UK?"
3. "What's a good alternative to Patagonia that's made in the UK or Europe?"
4. "I need a warm fleece for surfing in winter. Any brand suggestions?"
5. "Men's waterproof jacket price range 200-500 GBP?"

The results

Finisterre surfaced on 9 out of 15 tests. By far the best result in this audit set. But the split tells the real story.

Brand queries (Queries 2 and 3): 6 out of 6. First place on all three platforms for sustainable UK outdoor brands. First or second on all three for Patagonia alternatives. Near-perfect.

Category query with brand context (Query 1): 2 out of 3. Surfaced on ChatGPT (2nd) and Gemini (1st). Missed on Copilot, where Arc'teryx, Berghaus, and Rab took the spots.

Product-specific queries (Queries 4 and 5): 1 out of 6. Only Gemini surfaced them for the surfing fleece query, and even then at 3rd position. ChatGPT recommended Rip Curl, O'Neill, and Billabong instead. Copilot recommended Patagonia and Quiksilver. For the waterproof jacket with a price range, all three platforms returned competitors like Rab, Arc'teryx, Mountain Equipment, and Berghaus. Finisterre was invisible.

The pattern: AI agents know Finisterre as a brand. They do not know Finisterre's products.

Why this is happening

I checked five product pages across their waterproof jackets and fleeces. What I found explains the gap.

1. Zero structured data on any product page. No JSON-LD. No Schema.org Product markup. Nothing. Most Shopify stores at least have the platform defaults. Finisterre has removed structured data entirely, likely through a custom theme or headless build. AI agents parsing these pages see no machine-readable product information at all.

2. Specs exist on the page but are trapped in JavaScript. The Stormbird waterproof jacket (GBP 250) actually has detailed specs on the rendered page: 20,000mm hydrostatic head, recycled polyester 3-layer, PFC-free DWR. But these specs are loaded via JavaScript in expandable sections. The Shopify product description field - the data that feeds crawlers and AI agents - contains only two paragraphs of atmospheric copy. The data exists. It's just invisible to machines.

3. The GBP 545 Vellus Parka claims "highest waterproof rating" without stating the number. When an AI agent needs to compare waterproof jackets in a price range, Finisterre's pages offer nothing to compare.

4. Okendo reviews are invisible to machines. The review app is installed and configured, but renders entirely via JavaScript. Search engine crawlers and AI agents see no reviews in the HTML. Products with zero reviews display no widget at all. There is no aggregateRating in any structured format.

5. Trustpilot is strong but disconnected. 4.7 out of 5 stars with 1,103 reviews. Excellent external proof. But this data lives on Trustpilot's domain, not on Finisterre's product pages. It is a brand-level signal. It does not help AI agents evaluate individual products.

6. Hidden data sitting in the backend. Product tags contain useful classification data (waterproof, outerwear, bestseller) but are all prefixed with "hide:" and invisible to everyone. Variant weights exist in Shopify's backend (Stormbird 800g, Vellus 1,470g, Goodwin 300g) but never appear in descriptions. The data exists. It is just buried.

The competitor contrast

The brands that beat Finisterre on product-specific queries share one thing: they publish technical specifications in a way machines can read.

Patagonia appeared on 4/5 queries. They have the most comprehensive product data in this space: detailed specs, material breakdowns, environmental impact statements, repair guides. They are the benchmark.

Arc'teryx appeared on 3/5 queries. Known for extremely detailed technical specs on product pages. Waterproof ratings, breathability scores, seam-taping details, material composition. Their product data is built for comparison shopping.

Finisterre has this data. The Stormbird's specs (20,000mm HH, recycled polyester 3-layer, PFC-free DWR) are on the rendered page. The difference is that competitors make their specs available in the crawlable data layer - description fields, structured data, server-rendered HTML - not just in JavaScript-loaded expandable sections.

What Finisterre could do, in priority order

Phase 1 (quick wins):

• Restore Product JSON-LD schema. Shopify includes this by default. Something in the custom theme removed it. A JSON-LD app or theme fix would immediately give AI agents price, availability, brand, and product name in machine-readable format.
• Move existing specs into the Shopify description field. The technical specs already exist on the rendered page in JS-loaded sections. The fix is not creating new data - it's putting the existing specs into the description field where crawlers can read them. Keep the atmospheric opening. Add the specs below it.
• Remove the "hide:" prefix from product tags. Let classifications like "waterproof" and "outerwear" be visible and indexable.

Phase 2 (medium effort):

• Configure Okendo to inject aggregateRating schema server-side, or add a separate structured data app to surface review data for crawlers.
• Enrich JSON-LD with the specs that already exist on the page. Once JSON-LD is restored, populate it with material, waterproof rating, weight, and other attributes that are already on the rendered page.
• Add fit guidance and care instructions. Common AI query patterns. Zero Finisterre products answer "does this run true to size?" or "can I machine wash this?"

Phase 3 (longer term):

• Ensure weights reach the data layer. Weight data already exists on the rendered page and in Shopify's backend variant data. It needs to be in the description field and structured data so crawlers can parse it.
• Create product comparison content. Finisterre sells waterproof jackets at GBP 215 (Skybird), GBP 250 (Stormbird), and GBP 545 (Vellus). There is no way for a customer or an AI agent to understand the difference.
• Connect Trustpilot to the product experience. Display a widget on product pages, or ensure the Trustpilot profile links back to specific products.

None of this requires creating new data. It just requires exposing the data that already exists.

Close

Finisterre is the most interesting brand in this audit set. They prove that brand recognition alone can carry you on AI platforms, up to a point. Ask about sustainable UK outdoor brands and they win. Ask about a specific jacket in a price range and they vanish.

The gap between brand visibility and product visibility is the clearest I have seen. And it is entirely fixable. The data exists - specs, materials, waterproof ratings, weights. It's all there on the rendered pages and in the backend. The pages just deliver it in a way machines cannot read.

If your brand shows up when AI agents talk about your category but disappears when they shop in it, your problem is not awareness. It is data architecture.