Isabella Vrana Against AI Shopping Agents

We audited Isabella Vrana: a brand with 175,000 Depop followers: against AI shopping agents. It didn't surface on a single query. Here's the full breakdown.

Executive Summary

• Brand: Isabella Vrana is an independent womenswear, ethically made in London. Team of 4. GBP 40-110.
• AI visibility score: 0/15 queries surfaced the brand. Total invisibility.
• The pattern: A brand with 175k Depop followers, 130k TikTok, and 50k Instagram has zero AI presence. Social-first growth has built an audience on platforms AI agents don't crawl.
• Key competitor gap: ADKN, with 1,300 Instagram followers, surfaced on Gemini and Copilot for the same ethical London fashion query. The difference: a free review app and a Trustpilot profile.
• Root cause: Two layers deep. The visible product descriptions are inconsistent (one strong page, the rest thin). Underneath, the invisible structured data is bare Shopify defaults across every product, no material, colour, reviews, or country of origin in machine-readable format, even on the strongest page.
• Fix complexity: Low. Phase 1 (free review app, JSON-LD enrichment) is invisible to customers and takes a day. Phase 2 (consistent descriptions using the Elle Trousers page as template) takes 1-2 days.

The brand

Isabella Vrana is an independent womenswear label founded in 2015, based in East London with a team of four women. They make 90s and 00s-inspired pieces: high-waisted flared trousers, slinky tops, tailored jackets, coordinated sets. Everything is ethically manufactured in personally-audited London factories using UK-sourced and deadstock fabrics.

Price point: GBP 40-110. Sizes 4-16. Shopify store.

Their social numbers are strong: 175,000+ Depop followers, 130k+ TikTok, 50k+ Instagram.

The test

I ran 5 queries across ChatGPT, Gemini, and Copilot, 15 tests total. Each query was designed to match Isabella Vrana's exact positioning and products.

Some of the example queries:

1. "Can you recommend an ethical womenswear brand that makes clothes in London?"
2. "I need high-waisted flared trousers in a stretchy material. What brands do good ones?"
3. "What's a good independent women's fashion brand with a 90s or Y2K aesthetic?"
4. "I need smart casual trousers for work that are comfortable and stretchy. Suggestions?"
5. "What are some good small UK fashion brands for women that aren't fast fashion?"

The results

Isabella Vrana surfaced on 0 in the tests. Zero.

Here's what showed up in these 5 example queries:

• Query 1 (ethical London-made): ChatGPT recommended Belles of London, Maimie London. Gemini recommended ADKN, Fanfare Label, Phoebe English. Copilot recommended ADKN.
• Query 2 (high-waisted flared trousers): All three platforms recommended M&S, ASOS, Zara, Bershka, Spanx, Lululemon, despite this being Isabella Vrana's bestselling product.
• Query 3 (90s/Y2K aesthetic): ChatGPT recommended House of Sunny, Bat & Boa. Gemini recommended Jaded London, Minga London.
• Query 4 (smart casual work trousers): Dominated by M&S, Uniqlo, Spanx, Lululemon.
• Query 5 (small UK brands, not fast fashion): Lucy & Yak appeared on all three platforms. Damson Madder, Aligne, Baukjen, Fanfare Label all surfaced. Isabella Vrana did not.

The standout finding: ADKN

ADKN, a brand with roughly 1,300 Instagram followers, surfaced on both Gemini and Copilot for the ethical London fashion query. Isabella Vrana, with 100x the social audience, didn't.

The difference: ADKN has on-site product reviews (via a free Shopify app), a Trustpilot profile with 35 reviews, and listings on ethical fashion directories like British Made Clothing and Ethical Globe. That's it.

AI agents don't count followers. They parse structured product data, review schema, and third-party web citations. 175,000 Depop followers are invisible to ChatGPT. Five Trustpilot reviews are not.

Why this is happening

I looked at the actual structured data on Isabella Vrana's product pages, the invisible machine-readable layer that AI agents parse. It's bare-minimum Shopify defaults: product name, price, availability, and size variants. That's it.

Here's what's missing from the structured data, even on the strongest product page (the Elle Trousers):

• No material property. The fibre composition (75% Polyester, 22% Viscose, 3% Spandex) is in the description text but not structured as a machine-readable field. An AI agent filtering by material can't find it.
• No color property. Even though every product comes in named colours.
• No aggregateRating or review data. Zero reviews anywhere on the site. No Trustpilot. No social proof that machines can read.
• No countryOfOrigin. "Made in London" is in the description copy on some pages but not structured as data. When an AI agent filters for London-made fashion, this brand doesn't register.
• No additionalProperty for use case, care instructions, season, or fit type.

The human-readable descriptions actually vary in quality. The Elle Trousers page is genuinely well-written, it names occasions ("work, school, date night"), gives full material composition, includes inline measurements, and cross-references other products. It reads well for humans AND contains the kind of detail AI agents need.

But the Chloe Trousers page, the bestseller, leads with "signature stretch fabric" instead of the actual composition. Other pages are thinner still.

The core problem is two layers deep. The visible descriptions are inconsistent, some good, some thin. And underneath, the invisible structured data is uniformly bare across every product, even the well-written ones. The Elle Trousers page proves you can write for humans and include the detail AI agents need. But even that page has nothing in the machine-readable layer beyond Shopify defaults.

What Isabella Vrana could do, in priority order

The invisible layer is where the biggest gains are, and none of it changes what customers see.

Phase 1 Invisible fixes (1 day, changes nothing on the page):

• Install Judge.me (free Shopify review app). It automatically adds aggregateRating schema to product pages. Send a post-purchase email to existing customers asking for reviews. This is likely why ADKN beats a brand with 100x the audience. They did this one thing.
• Install a JSON-LD structured data app (e.g. JSON-LD for SEO). Configure it to output material, color, countryOfOrigin, and additionalProperty fields for use case, care, and fit type. All pulled from Shopify meta fields. The customer sees the same page. The AI agent sees a completely different data profile.

Phase 2 Visible fixes (1-2 days, improves the page for humans too):

• Use the Elle Trousers page as the template for every product. Add use-case language, actual fibre percentages, care instructions, and cross-references. This isn't writing for machines: it's writing more completely in the same brand voice. The Elle page proves it works.
• Replace "signature stretch fabric" with real material data. The composition exists. It's just not on every page.
• Add "Ethically made in London" consistently to every product description.

Phase 3 External signals (1-2 weeks):

• Create a Trustpilot profile. ADKN's 35 Trustpilot reviews create a third-party citation AI agents can find. Isabella Vrana has none.
• Get listed on ethical fashion directories, British Made Clothing, Ethical Globe, Good On You. These are free listings that create the web authority AI agents use to validate brands.
• Set up Shopify meta fields for use case, season, fabric composition, and care instructions to feed the structured data layer long-term.

A team of four could complete Phases 1 and 2 in a week without spending anything on advertising.

Close

Isabella Vrana's products, ethics, and audience are all strong. The gap is purely data, both the visible descriptions and the invisible structured layer underneath.

If you run an e-commerce brand, try this: ask ChatGPT to recommend a product in your category. See who shows up. If it's not you, now you know where to start.