How to start a cosmetics brand: the sequence that works

A cosmetics brand is almost never bought for its chemistry. It is bought because someone trusts the person selling it, recognises themselves in the positioning, or has a problem the category has not addressed well.

So the first question is not "what should the product be" but "who is already listening to me, and what do they ask for". Founders who have an audience before they have a product survive the first year at a rate the ones who did it the other way round do not. If you have no audience, your first real task is distribution, and the product decision can wait a quarter without costing you anything.

Categories differ enormously in what they demand of a new brand. Shampoo, shower gel and body lotion are high-volume, habit-driven and forgiving: the formulation space is mature, the minimums are reachable, and repeat purchase is built into the product.

Serums and actives-led skincare carry better margins and far more risk. The claim has to be substantiated, the testing bill is higher, and the customer compares you against brands with laboratories.

Hair styling and men's grooming sit in between, and are where a small brand can still differentiate on performance — hold, finish, feel are judged immediately and remembered. Fragranced body products and home scent are the easiest categories to make gift-worthy, which matters more to a small brand's fourth quarter than most founders plan for.

Choose the one where your advantage compounds, then check that the minimum order for a credible range in that category is a number you can write a cheque for twice.

White label puts an existing product in your pack. Fastest, cheapest, least defensible — useful for filling a gap or testing whether a category sells at all.

Private label adapts a manufacturer's base to your brief. This is where most new brands belong, and where the economics actually work at launch volumes.

Contract manufacturing develops a formula that belongs to you. Correct when the formula is the advantage; expensive and slow when it is not.

The honest test: if your product disappeared and a competitor's appeared in your packaging, would your customers notice? If the answer is no, do not pay for bespoke development — put the money into the reason they buy from you instead.

Founders typically design a range, then go looking for someone to make it, and discover that the range implies minimums they cannot meet. Doing it the other way round costs nothing and saves a quarter.

Send the same brief to four factories. Ask for the minimum per SKU, not per order. Ask what they do not do in-house. Ask what their lead time is from approved sample to delivered goods, and whether that includes the testing.

The replies will sort themselves. The ones that quote immediately without asking a question are quoting a template. The ones that ask about your destination market before pricing understand that compliance is part of the cost.

Every new or modified formula needs stability data and preservative efficacy testing before it can be sold. A challenge test to ISO 11930 runs 28 days. Real-time stability runs longer; accelerated protocols shorten the wait but do not eliminate it. These weeks are not negotiable and they are the most common reason a launch date moves.

If you are selling into the EU you also need a Responsible Person established in the EU, a Product Information File, a Cosmetic Product Safety Report and a CPNP notification before the first unit reaches a shelf. None of this is difficult; all of it takes calendar time, and a founder who starts it after the goods are produced will have stock sitting in a warehouse it is illegal to sell from.

Put these on the Gantt chart at the same time as the production slot, not after it.

Two or three SKUs. Not six, not a full routine. A short range tells you which product people actually come back for, and that single piece of information is worth more than the breadth you think a range needs to look credible.

You can add the fourth scent in four months, funded by the first three. You cannot un-buy five thousand units of a variant nobody wanted.

Over-ranging. Covered above, and it is the big one. The capital tied up in variants nobody asked for is capital not spent on reaching the people who would have bought the hero.

Packaging decided before volume. A custom bottle carries tooling cost and a minimum of its own, often larger than the formula's. Stock packs are not a compromise at launch; they are the correct decision until a single SKU earns its own mould.

Paying for development you did not need. Bespoke formulation on a first product, before the market has validated anything, is the most expensive way to learn that the positioning was wrong.