66-word flagship. 9-word corrector. Subscription brands need consistency most.

Executive Summary

• Brand: Global men's skincare brand built on a subscription model, with a range spanning daily essentials through to LED therapy devices
• AI visibility score: 0/50 ChatGPT tests surfaced the brand
• The pattern: The problem isn't what's missing — it's what's inconsistent. Some products have rich data, others have almost nothing. The relationships between products become invisible
• Key competitor gap: Subscription competitors that present a coherent system to AI agents win the "men's skincare routine" recommendation
• Root cause: 9-66 word descriptions (7x variance), mixed consumer-facing and internal tags, aggregateRating present but without review counts
• Fix complexity: Medium — requires levelling up every product to the flagship standard, not averaging out

The brand

Lumin is a global men's skincare brand built on a subscription model. The product range spans daily essentials like face washes and moisturisers through to LED therapy devices. It's positioned as a complete men's skincare system — the kind of brand that should own AI queries about men's skincare routines.

The test

We ran 50 automated ChatGPT tests using Playwright — 10 repeats × 5 queries. Queries targeted Lumin's positioning: best men's skincare subscription, best men's face wash for acne scars, men's complete skincare routine, best LED therapy device for skin, and affordable men's moisturiser.

The results

0% ChatGPT visibility. Lumin appeared zero times. The data tells us why, and the answer is not about what is missing — it's about what is inconsistent.

Why this is happening

The widest content variance in grooming. Lumin's descriptions range from 9 to 66 words. A 7x gap between the most described product and the least. The flagship face wash gets 66 words — enough to cover basics, though still thin by AI visibility standards. The Dark Circle Corrector gets 9 words. Nine words cannot describe what the product does, let alone who it's for or how it fits into a routine.

This is the widest variance in the grooming audit group. Dr. Squatch ranges 11-21 (thin but consistent). Huron ranges 18-23 (thin but very consistent). Supply ranges 36-89 (broadest after Lumin, but starting from a higher floor).

AI agents don't average data quality. When an agent evaluates a brand's catalogue, it evaluates each product individually. A well-described face wash can be recommended. A 9-word corrector cannot. The brand's range looks fragmented — some products are describable, others are not.

The tagging inconsistency mirrors the description problem. Some Lumin products have consumer-facing tags — Acne Scars, Moisturiser — that map to how customers search. Others have only internal tags. The mix is unpredictable. Consumer-facing tags create direct matches with conversational query patterns. But if only some products have them, the AI agent gets an incomplete picture. It might recommend the face wash for acne but not know the corrector exists.

The rating gap. Lumin's structured data includes aggregateRating with scores of 4.6-4.7. Strong ratings signalling customer satisfaction. But the JSON-LD does not include review counts. A 4.7 rating with 500 reviews is meaningfully different from a 4.7 rating with 5 reviews. Without the count, AI agents must treat the rating as lower confidence. Huron includes review counts from 313 to 869 per product — giving AI agents both quality and confidence signals.

Subscription brands need consistency most. A single-product brand can survive with one strong product page. A subscription brand cannot. The value proposition is the range. A customer considering a men's skincare subscription needs the AI agent to understand the face wash, the moisturiser, the serum, the corrector, and how they work together. If some products have rich data and others have almost nothing, the agent cannot construct a coherent recommendation for the system. It's not that some products are invisible — though they are. It's that the relationships between products are invisible.

What Lumin could do, in priority order

Phase 1 (quick wins):

• Level every product up to the 66-word flagship standard as the floor, not the ceiling
• Add review counts to the existing aggregateRating markup
• Apply the consumer-facing tag pattern (Acne Scars, Moisturiser) consistently across every relevant product

Phase 2 (medium effort):

• Expand all descriptions to 80+ words with consistent attribute coverage: what it does, who it's for, key ingredients, how it fits in a routine, what skin concerns it addresses
• Build a structured "routine" data model that connects products into a system — so AI agents can see the relationships, not just the parts
• Add additionalProperty fields for skin concern, routine step, and skin type across every product

Phase 3 (longer term):

• Develop content explicitly built around "complete men's skincare routine" queries — the subscription value proposition needs to be discoverable
• Pursue editorial inclusion in "best men's skincare subscription" roundups
• Build the LED therapy device into product range visibility — currently the most differentiated product but also least visible

Close

Lumin already has pieces of the answer scattered across its catalogue. The problem is that they are scattered, not systematic. The brand that needs the most consistency has the least. The 66-word flagship description is not the standard to match — it is the floor to build from. The subscription model requires AI agents to see a connected system. Inconsistent data shows them disconnected fragments.