Subscription beauty was supposed to be the future. Birchbox launched in 2010 and the market exploded - discovery boxes, replenishment subscriptions, curated kits. By 2019, beauty subscription revenue hit £2.5 billion globally.
Then churn happened. And happened. And kept happening.
The category has matured considerably since those early days, and the picture is now much clearer about what works, what doesn't, and what technology can fix.
What's working: replenishment subscriptions
If a customer uses your moisturiser every day and it runs out every 60 days, a subscription that delivers a fresh jar on day 55 is genuinely valuable. The customer never runs out. The brand gets predictable revenue. Everyone wins.
Replenishment subscriptions work best for:
- Skincare staples (cleansers, moisturisers, SPF)
- Haircare basics (shampoo, conditioner)
- Personal care essentials (deodorant, body wash)
The common thread: products with predictable usage rates that consumers have already committed to repurchasing. The subscription removes friction, not choice.
Brands like Dollar Shave Club (now in beauty) and Hairstory have built strong subscription businesses because their products are consumed at predictable rates and the convenience value is obvious.
What's struggling: discovery boxes
The discovery box model - receive 5-6 sample-sized products monthly for £10-15 - has largely plateaued. The novelty wears off. Consumers accumulate drawers full of tiny products they'll never finish. And the unit economics are challenging: acquiring a £12/month subscriber costs nearly as much as acquiring a full-price customer, but the revenue per subscriber is far lower.
Birchbox's decline told this story clearly. The model works as a customer acquisition tool for the brands inside the box, but as a standalone business, the margins are thin and churn rates run 8-12% monthly.
The exceptions are premium discovery services with strong curation - boxes that feel like a personal shopper rather than a random assortment. These command higher price points (£30-50/month) and retain better because the curation itself is the value.
The churn problem - and the AI opportunity
Beauty subscriptions face a fundamental challenge: consumer needs change. Skin changes seasonally. People get bored of fragrances. A product that was perfect in winter feels too heavy in summer.
Traditional subscription models don't account for this. They send the same product on the same schedule regardless of changing needs. The result: cancellation rates of 40-60% within the first six months.
This is where AI-driven approaches are changing the economics. By analysing customer behaviour patterns - engagement with emails, browsing behaviour on site, customer service interactions, review activity - models can now predict churn risk 30-45 days before it happens.
That prediction window is valuable. It's enough time to intervene: offer a product swap, adjust delivery frequency, send a personalised offer, or simply ask what's changed. Early intervention reduces churn by 25-35% in the programmes I've seen implement it well.
The hybrid model
The most promising subscription approaches in beauty combine replenishment with flexibility:
- A core subscription product (the staple they always need) paired with a rotating discovery element (a new product each quarter)
- Flexible frequency - let customers set their own delivery schedule and change it easily
- Easy swaps - if their skin has changed, let them switch to a different variant without cancelling
- Pause without guilt - brands that make pausing easy actually retain better long-term than those that make it difficult
The investor perspective
For investors evaluating subscription beauty businesses, the key metrics are:
Cohort retention curves. Not average churn - month-by-month retention for each acquisition cohort. Healthy businesses show retention curves that flatten after month 4-6, meaning the customers who make it past the initial period stick around.
Gross margin per subscriber. Including product cost, shipping, and packaging. Many subscription businesses are margin-negative at the unit level and relying on volume to reach profitability - a risky bet.
Churn prediction accuracy. Businesses using predictive models to manage churn are structurally more valuable than those reacting to cancellations after they happen.
Subscription as percentage of total revenue. The healthiest beauty brands use subscriptions as one revenue stream alongside one-time purchases, retail, and bundles. Over-dependence on subscription revenue creates fragility.
The subscription model in beauty isn't dead - but the naive version of it is. The winners are the ones using data to make subscriptions adaptive, personal, and genuinely convenient.