Claim vocabulary is defined as the specific set of words a brand uses to describe what its products do, which indicates how much evidence it holds, because a measurable claim must be substantiated while an atmospheric one carries no such requirement. That makes it the first layer to read when you research competitors in beauty. Published research puts average cart abandonment near 70% across eCommerce [Baymard Institute, 2025], so a competitor's visible range tells you less than its reorder behaviour does. This guide sets out five observation layers and what each one reveals.
- Sort competitor claims into measurable and atmospheric to see what evidence they bought.
- Read range depth for positioning and persistent stockouts for unserved demand.
- Use bundles to identify the acquisition product and the margin product.
- Track reviews added per month, because a lifetime total describes history.
- Judge creator relationships by recurrence rather than by follower count.
- Benchmark your own store first so the research targets an area you are behind on.
Omniconvert has measured store performance across the CROBenchmark dataset of 7,000+ websites in 15+ industries, against 248+ audit criteria, over 13 years in eCommerce [CROBenchmark Report 2026, Omniconvert]. In beauty the finding that travels is that brands differ far more in what they are willing to promise than in what they actually sell.
Why you can research competitors in beauty from the outside
The structural reason is regulation. What a cosmetic product may be said to do is governed, and a brand making a specific claim about a result has to be able to stand behind it. That turns product copy into a signal: the brand that says a measurable thing has usually paid for the evidence, and the brand that says an evocative thing has not had to.
The second reason is the subscription and replenishment model that dominates the category. Repeat purchase is where the money is, so bundles, minis and sampling are not merchandising decoration. They are the acquisition machine, and its shape is visible on any product page.
What follows is a five-layer method built on repeated observation. The wider structure for the exercise is in the DTC competitor research stack; this is the beauty-specific version of its middle layers.
Layer one: claim vocabulary
A measurable claim names a result, a timescale or a proportion. An atmospheric claim names a feeling, a ritual or an ingredient story. Both sell; they cost different amounts to make. The first requires substantiation and often a clinical or consumer study, which is a real line in a budget.
So count them. A competitor whose range leans heavily on measurable claims has invested in evidence and will win a factual comparison, which means competing on breadth, service or price rather than on proof. A competitor leaning on atmosphere has invested in brand, and a well-evidenced measurable claim is the thing it cannot easily answer.
Note also where a brand declines to claim. A product with warm copy and no stated result in a category where rivals state one is usually a formulation the brand cannot evidence, and that is the gap worth testing into.
The trap here is reading the ingredient list as the claim. A team sees a competitor leading with a fashionable active, matches the ingredient at a similar concentration, and launches against them, only to find that the competitor's advantage was never the molecule. It was a consumer study it could cite and a claim it was therefore allowed to print, which the new product cannot say even though the formula is comparable.
Layers two and three: range depth and bundle design
Count variants and shades per line, then check availability repeatedly over a month. A shade that is simply not offered may be a deliberate positioning decision. A shade that is listed and perpetually unavailable is demand the brand has chosen not to fund, which is a different and far more actionable finding.
Bundle design is read the same way. An acquisition product is defined as the item a brand is willing to discount or give away in order to start a customer relationship, and in beauty it is almost always visible: it appears in every bundle, in every sampling offer and in every first-order incentive, while one other product never discounts at all. That second product is where the margin lives.
| Layer | What it reveals | Cadence |
|---|---|---|
| Claim vocabulary | What evidence the brand paid for | Quarterly |
| Range and shade depth | Audience served, demand unmet | Monthly |
| Bundle and sampling design | Acquisition product and margin product | Monthly |
| Review velocity | Which products sell now | Monthly |
| Creator mix | Which partnerships keep working | Fortnightly |
One qualification on bundles. A brand sometimes bundles its margin product to clear a formulation it is about to discontinue, which looks identical from outside. The way to tell is persistence: an acquisition product is bundled every month, a clearance is bundled hard and then disappears.
Layer four: review velocity
This is the cheapest comparative signal in the category and almost nobody collects it. Record the review count on a competitor's top ten products once a month and keep the differences. Within a quarter you have a ranked picture of which of their products are actually moving, which no total can give you.
Velocity also exposes a launch that is not working. A product given a large push whose review count barely moves has a conversion or a repeat problem, and the brand usually knows it before the market does. Conversely a quiet product gaining reviews steadily is the one to look at closely, because somebody is buying it without being told to.
Treat the absolute numbers carefully. Review solicitation practices differ between brands, so velocity is comparable within one competitor's range and only roughly comparable between brands.
AliveCor used Omniconvert to run a structured A/B testing programme and achieved +21% conversion rate, +5% revenue per visitor, and 94% statistical relevance across their experiments. [Omniconvert, AliveCor case study] The standard carries over: a competitor-derived idea is a hypothesis until it has been measured on your own traffic to relevance.
Layer five: creator mix
The public advertising transparency tools cover the paid side of this. Meta's Ad Library shows which creatives are currently live for a page and when they began, which makes duration observable even though spend is not. For the organic side, the useful record is which creators appear more than once over several months.
What you are extracting is the angle rather than the asset. A brand running the same argument across many creators for a quarter has evidence for that argument, and the transferable finding is the argument. The method for turning it into a test is in turn findings into creative tests.
What to do this week
- Classify one competitor's claims. Ten hero products, each claim sorted into measurable or atmospheric. An hour, and it reframes how you compete with them.
- Start the monthly review snapshot. Review counts on their top ten products, recorded once a month. Fifteen minutes.
- Log bundle composition. Which product appears in every bundle and which never discounts. That is their acquisition and margin structure in one note.
Before committing a quarter to this, it is worth knowing which part of your own store is behind. A free eCommerceBenchmark leaderboard score grades a Shopify store across six dimensions, benchmarked against real competitors in your category and country, and the paid audit report sets out what to fix and in what order. Once the gap is identified, Omniconvert Explore is the platform for testing into it, with A/B and multivariate testing, on-site surveys and session recordings; Explore averages a 23.2% conversion uplift across 70,000+ experiments. Further upstream, Omniconvert Commander (formerly Nexus) identifies which metrics to prioritise and queues the experiments most likely to close the gap, and you approve what goes live.
A closing caution that matters more in beauty than in most categories. Everything above is observable and none of it is explained, and the regulated nature of the claims makes the temptation to copy language actively risky. Read a competitor's claim as evidence that they hold evidence, never as permission to make the same claim yourself. For category context on the numbers, read what is a good conversion rate as context rather than as a target.
This now has a sibling worth the detour: research competitors in fashion.
Frequently asked questions
How do you research competitors in beauty?
In five layers of public evidence: claim vocabulary, range and shade depth, bundle and sampling design, review velocity and creator mix. Beauty brands disclose more than they realise, because a measurable claim has to be substantiated and a bundle shows which product is the acquisition tool. Collect the layers on a schedule, since several of them are only meaningful as a change over time.
Why does claim vocabulary matter so much?
Because the words divide into two classes with different costs. A measurable claim about a result has to be substantiated, so a brand making one has usually paid for testing. An atmospheric claim about how a product feels requires no such evidence. Counting which class a competitor leans on tells you whether you are facing an evidence-led brand or a positioning-led one.
Is review velocity better than review count?
For competitive reading, yes. A total review count records a brand history and keeps rising even as a product declines. Reviews added per month tracks current sales volume, so comparing velocity across a competitor range shows you which products are actually selling now rather than which ones sold best at some point in the past.
What do bundles tell you?
Which product acquires the customer and which one earns the margin. A brand that discounts a cleanser into every bundle and never discounts its serum is telling you where it makes money and which product it is willing to lose on to start a relationship. Sampling placement says the same thing one step earlier in the funnel.
How do you spot an unserved segment?
Through persistent stockouts rather than through absent products. A shade or variant a competitor does not sell may be a deliberate decision. A shade that is listed and repeatedly unavailable across several months is demand the competitor has decided not to fund, and those customers are actively looking for an alternative.
Where does benchmarking fit alongside this?
Benchmarking tells you which part of your own store is behind; competitor research tells you what a specific rival is doing about that part. Running them in that order saves a great deal of effort, because a quarter spent studying a competitor strength in an area where you are already ahead produces interesting reading and no decisions.
Read what they will not claim
The most useful hour in beauty competitor research is spent sorting a rival's claims into the measurable and the atmospheric, and then noticing where they go quiet. A product with warm copy and no stated result, in a category where competitors state one, is a formulation they cannot evidence.
Then start the monthly review snapshot. Velocity tells you what is selling now, and no lifetime total ever will.