Drop cadence is defined as the rhythm at which a fashion retailer introduces new styles to its catalogue, measured by how often new products appear and in what quantity. It is the first thing to record when you research competitors in fashion, because it tells you how a rival funds inventory and where it expects margin to come from. Published research puts average cart abandonment near 70% across eCommerce [Baymard Institute, 2025], which is a reminder that a competitor's traffic is not its revenue and a visible range is not a selling one. This guide gives five observation layers and what each one reveals.
- Log new-in counts weekly, because the rhythm of introduction reveals the inventory model.
- Map the price ladder by category instead of calculating an average price.
- Watch when the first markdown lands, which separates planned pricing from buying errors.
- Treat persistently out-of-stock sizes as unmet demand you could serve.
- Read creative rotation in the public ad libraries rather than judging single ads.
- Benchmark your own store first, then research competitors in the area you are behind.
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]. The pattern worth carrying into this work is that apparel assortments differ far more in their timing than in their taste, and timing is observable from outside.
Why you can research competitors in fashion from outside
This is not true of most categories. A software company's pricing experiments are hidden behind sales calls and a grocer's margin structure is invisible. A fashion retailer has to put the whole range on a page, with prices, in sizes, and then has to discount in public when it got the buy wrong.
What follows from that is a method based on repeated observation rather than on one-off analysis. The signal is in the sequence: what appeared, when, at what price, and when it first came down. A single snapshot of a competitor's site tells you what they sell. Four weeks of snapshots tell you how they operate.
The discipline is the hard part. This work fails because nobody logs it weekly, not because it is difficult. The wider structure for the whole exercise is in the DTC competitor research stack.
Layer one: drop cadence
A retailer introducing a handful of styles every week is competing on freshness. It accepts that a proportion of each drop will miss and will be marked down, and it funds that with the full-price sell-through of the hits. Its site is organised around newness because newness is the product.
A retailer introducing larger ranges a few times a season is buying depth. It expects to hold full price for longer, carries more units per style and markdowns are an event rather than a routine. Its risk is concentrated in fewer decisions.
Both models work. The error is reading a competitor's tactics without reading its model, then importing a newness playbook into a depth business and wondering why the markdown budget blew out. The count tells you which conversation you are in.
Where this goes wrong in practice: a team watches a fast-newness rival for a quarter, concludes that it wins on drop frequency, and pushes its own buying team to triple the number of styles. The buying team has neither the supplier terms nor the markdown budget the model requires, so the extra styles arrive late, miss their window and discount immediately. The observation was right and the inference skipped the inventory model entirely.
Layers two and three: price architecture and markdown timing
A price ladder is defined as the set of entry, core and top price points a retailer offers within one product category, and it is far more informative than an average price because it shows which tier the business actually earns on. A deep entry tier with a thin core is a traffic-buying structure. A thick core with a short ladder is a margin structure.
Markdown timing then tells you how well the ladder is working. Log the date a style first discounts and the size of the cut. A style holding full price for two months and then taking a shallow reduction was planned. A style cut deeply in its second week was a buying mistake, and a range full of those is a competitor under pressure regardless of how its site looks.
Collect this on a schedule, not retrospectively. Historic prices are not reliably observable, so the only version of this data you will ever have is the one you started collecting before you needed it.
Layer four: size range and persistent stockouts
| Layer | What it reveals | Cadence |
|---|---|---|
| Drop cadence | Inventory model and margin source | Weekly count |
| Price architecture | Which tier earns the money | Monthly |
| Markdown timing | Which buys worked | Weekly |
| Size range | Audience served and demand unmet | Monthly |
| Creative rotation | Which concepts have evidence behind them | Fortnightly |
The stockout read works because it is a repeated measurement rather than a single one. One size missing on one style is noise. The same two sizes missing across thirty styles for six weeks is a competitor consistently underbuying a part of the curve, and those customers are looking for somewhere to shop.
This is also the cheapest gap-finding exercise available in apparel. It needs no tooling, no spend and no access, and it produces a buying recommendation rather than an observation.
Layer five: creative rotation
The public ad transparency tools make this observable. Meta's Ad Library shows which creatives are currently running for a page and when they started, and Google's Ads Transparency Center covers the same ground for its platforms. Neither shows spend, and neither needs to: duration is the signal.
Record the concept rather than the asset. A competitor running the same angle across six executions for three months has evidence that the angle works, and that is the transferable finding. Copying the execution gets you a worse version of their ad; understanding the angle gets you a test worth running. The method for that is in turn findings into creative tests.
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] That is the standard a competitor-derived hypothesis should be held to: it is a hypothesis until you have measured it on your own traffic to relevance.
What to do this week
- Start the weekly log. Three competitors, two numbers each: new styles added, markdowns started. Twenty minutes a week.
- Map one price ladder. Pick your most competitive category and record entry, core and top for each rival. Compare it with your own.
- Run one stockout sweep. Thirty styles per competitor, note which sizes are unavailable. Repeat in a month before concluding anything.
Before you spend a quarter on competitor research, it is worth knowing which part of your own store is actually 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 in which order. Once you know where the gap is, 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.
One caution on interpretation. Everything in this method is observable and none of it is explained. You can see that a rival discounted a style in week two; you cannot see whether that was a buying error, a supplier problem or a deliberate traffic play. Hold the finding and the explanation apart, and where the category context matters, read what is a good conversion rate as context rather than as a target.
Since this was published, research competitors in beauty.
Frequently asked questions
How do you research competitors in fashion?
In five layers, all from publicly observable evidence: drop cadence, price architecture, markdown timing, size range and creative rotation. Fashion retailers reveal their strategy through what they put on the site and when they discount it, so a month of structured observation tells you more than any single teardown. Record it on a fixed schedule rather than looking when you feel curious.
What does drop cadence tell you?
How the business funds inventory and where it expects margin. A retailer adding a few styles weekly is competing on newness and accepting higher markdown risk per style. A retailer adding larger seasonal ranges is buying depth and expects to sell at full price for longer. Those are different businesses, and copying the tactics of one while running the other is how a range gets stuck.
Why track markdown timing rather than discount depth?
Because timing reveals intention and depth only reveals desperation. A style that holds full price for eight weeks and then takes a shallow cut was planned. A style discounted in week two at half price was a buying error. Watching when the first cut lands across a range shows you which parts of a competitor assortment are actually working.
What can the size range tell you?
Both who the retailer is serving and where demand is going unmet. A range offered in a narrow band of sizes is a positioning statement. More usefully, sizes that are persistently out of stock across many styles indicate demand the competitor is not fulfilling, which is the clearest gap-finding signal available in the category and costs nothing to collect.
How often should you do this?
Weekly for a light observation and quarterly for a full review. The weekly pass is a few minutes of logging new-in counts and any markdown starts, which is what makes the quarterly review possible at all. A single quarterly deep dive with no observation behind it describes a moment rather than a pattern, and patterns are the point.
Is competitor research a substitute for benchmarking?
No, they answer different questions. Competitor research tells you what a specific rival is doing and where its assortment is weak. Benchmarking tells you where your own store sits against your category and country on measures you can act on. The useful sequence is to benchmark first, to find which area is behind, then research competitors in that area specifically.
Start the log before you need it
Everything useful in fashion competitor research is a sequence rather than a snapshot, and sequences cannot be collected retrospectively. Twenty minutes a week on three rivals, recording new styles added and markdowns started, gives you in one quarter something no teardown can produce.
Then read the stockouts. Persistent gaps in a competitor's size curve are the one finding in this method that converts directly into a buying decision.