Can You Test a New Style Without Committing to a Large Order?
A design team has a new hoodie shape for the autumn range, the trend evidence is thin, and the opening buy would normally be the same size as a proven core style. How do you test a new style without carrying the stock and the markdown exposure of a full launch if it does not sell? The answer is to decide, in writing and before the PO, what the test must prove, and to size the buy so that a wrong guess is affordable.
For this guide, a test is a small production buy of a finished, sampled style, sold to real customers at the intended price in a channel you can read. It is not a sample, which is made to check fit and construction, and it is not a soft launch with no exit rule. A decision gate is the meeting, set in advance, at which the buyer compares results with thresholds and chooses scale, hold or exit. The constraint is nearly always time: the test has to read early enough that a winner can still be replenished inside its selling season.
A test reduces risk. It does not remove it: a small buy can still miss, and a strong result can still be a one-off. The rest of this guide covers how to set the test up so that the result is readable, which is the real difficulty. If you are choosing which styles to scale afterwards, our guide on which styles deserve a bigger buy picks up where this one ends.
How to Test a New Style: The Four-Step Method
Most failed tests fail before the first unit is made, because nobody wrote down what success meant or because the test was too thin to read. The four steps below are the order in which the decisions need to be made. Each is covered in its own section.
Four steps to test a new style
- 1Set success criteriaChoose KPIs and thresholds, with one owner, before the PO is raised.
- 2Choose the channelPick stores or an online route that represents the customer you want to win.
- 3Control quantity and sizesFix the units per style and colour and a size curve you can defend.
- 4Plan replenishmentSettle fabric, trims and factory capacity so a winner is not stranded.
The order matters: thresholds first, because they decide which channel and quantity can answer the question.
Notice what is missing: there is no step called "forecast demand". A test exists because the forecast is weak. What you are building is a cheap way of replacing a forecast with evidence. If the same question can be answered from existing sales data, for example by reading a close relative of the new style, do that first and save the test for the questions data cannot answer.
Step 1: Set Commercial Success Criteria Before the PO
The first job when you test a new style is to turn "see how it goes" into numbers. Choose three or four KPIs, set a threshold for each, and record who owns the call. Thresholds set after the results arrive are almost always bent towards whatever the team already wanted, so they must be written before the PO and kept in the PO file.
The four KPIs for a style test, defined
| KPI | Formula | What to state up front | Common trap |
|---|---|---|---|
| Sell-through | Units sold divided by units received, over the same period | The denominator (received into the test channel, or available to sell) and the read period; definitions vary by retailer | Counting marked-down sales as success; quoting a week-one figure against a week-six threshold |
| Margin | Gross margin = (selling price excluding VAT minus cost) divided by selling price excluding VAT | Whether cost is FOB or landed, and whether the margin is at full price or after expected markdown | Confusing gross margin with markup, which divides by cost and always reads higher |
| Conversion | Orders divided by sessions on the product page (online); definitions differ for stores | How traffic reached the page and whether it was promoted | Reading conversion on a style that had almost no traffic |
| Returns | Units returned divided by units sold | Reason codes, especially fit, size and quality, and how long the returns window runs | Reading returns too early, before the window has closed |
Definitions vary by retailer. State which version your business uses in the test brief.
Each KPI answers a different question. Sell-through says whether customers want the style at the price. Conversion says whether those who see it buy it, which separates a weak product from weak exposure. Margin says whether winning would be worth it. Returns says whether the sale will stick. A style can pass any one and fail the others, which is why the gate uses all four.
With a selling price of £40 excluding VAT and a cost of £16 (both illustrative), gross margin is (40 minus 16) divided by 40, which is 60 per cent. Markup is (40 minus 16) divided by 16, which is 150 per cent. Write which one your threshold uses.
What to write in the test brief
- The style, colourway and intended selling price, so that price is not a hidden variable.
- The read window and the date of the decision gate, agreed with the factory and the planning team.
- The threshold for each KPI, and the action tied to each outcome: scale, hold or exit.
- The single owner who makes the call, and the evidence that would make them change it.
Step 2: Choose Representative Stores or an Online Channel
A test only transfers to the main business if the channel resembles it. A style tested in the one store with the most fashion-forward shoppers will over-read, and a style tested only through one promoted email will measure the email. Decide what you need to learn, then choose the channel that can show it.
Choosing a test channel
| Option | Strengths | Limits | Suits |
|---|---|---|---|
| Online only | Fast read; sessions and conversion are measurable; easy to hold price and placement constant | Traffic can be bought, which flatters results; return reasons depend on your data | Fit-sensitive styles where returns data matters; fast fashion cycles |
| A small group of stores | Shows how the style looks next to the range; allows size-curve reading by location | Needs store selection rules; conversion is harder to measure; less stock available to sell per store | Styles that depend on touch, drape or visual merchandising |
| Online plus a store group | Cross-checks two readings of demand | More set-up and more stock to place; two sets of operations | Larger tests where the answer drives a significant volume decision |
| Wholesale or pre-order | Early demand signal before production, or a small committed order from trade customers | Customers may order what the sample shows rather than what finally sells; slower to read | Brands selling through trade accounts |
Qualitative comparison. Choose the route your own customer actually uses.
Three rules keep the channel honest. First, select stores or traffic sources by a rule you can write down, such as a spread of store grades or a defined email segment, not by who asks for the stock. Second, keep price, placement and promotion the same as the style would have if it went to scale; otherwise you will measure the promotion. Third, make sure the stock is actually available. A style that is invisible for the first fortnight, or sold out in the best-selling size, has not been tested at all.
Before you commit to the channel
- The customer in the test channel resembles the customer you want at scale.
- Traffic or store selection follows a written rule, not a favour.
- Price, placement and promotion are the same as planned for scale.
- Stock is live and visible from day one of the read window.
- Returns and customer comments can be traced back to the test style.
Step 3: Control Test Quantities and Size Curves
The quantity of a test is a trade-off between exposure and readability. Too few units and one large order or one stockout decides the result; too many and the test is a launch. The right reading unit depends on your sales rate and channel, and the proposed test quantity in our test and scale model is 500 pieces per style and colour, where commercially and technically feasible. It varies with fabric, dye, trims and factory, and it is a proposal, not a guaranteed minimum order quantity. For background on how minimums are set, see what MOQ means in clothing manufacturing, and for brands working at smaller scale, low MOQ clothing manufacturers in the UK.
Colours multiply the quantity, so decide deliberately which variable you are testing. If you test a new style in three colours at once, you have 1,500 units and three questions. Change one variable per option wherever you can, as the table below shows.
How the test cell grows with options
| Test design | Options | Units at 500 per option | What you learn |
|---|---|---|---|
| One style, one colour | 1 | 1 × 500 = 500 | Whether the silhouette sells at this price |
| One style, three colours | 3 | 3 × 500 = 1,500 | Colour preference, but the style signal is blurred by colour |
| Three styles, one colour | 3 | 3 × 500 = 1,500 | Which style wins, with colour held constant |
| Six options, one variable each | 6 | 6 × 500 = 3,000 | A broader read; only worthwhile if the gate and the replenishment plan can handle six outcomes |
Arithmetic only. 500 is a proposed test quantity per style and colour, not a fixed minimum.
The size curve needs the same care. A curve is the share of units in each size, and a test curve is a hypothesis about who your customer is. If the curve is too narrow, you hide demand in the sizes you did not make. If it is copied from a core style without thought, you may stock the wrong ends. The illustrative split below is a mid-weighted curve that puts the largest share in the middle sizes; use your own sales data, and see our grading and size chart guide for how sizes are graded.
Illustrative size-curve split for a 500-piece test
Units per size
Illustrative split that sums to 500 units. Not a recommended curve; build yours from your own size sales and your customer.
Two further points. Do not read small sizes on small numbers: nine units sold in XS tells you very little. And record when each size sold out, because a size that sells out in week two has had its demand cut short, and its true rate of sale is higher than the numbers suggest.
Step 4: Plan Replenishment Materials and Capacity Upfront
The cruellest outcome of a successful test is a winner you cannot reorder in time. If the style sells through in four weeks but fabric takes longer than the remaining season to arrive, the win is wasted. For this reason replenishment is planned before the test PO, not after the result, and it should be agreed with the factory as part of the test brief. Production timing differs by product and programme; our guide to clothing production lead times explains the stages that drive it.
What to settle before the test PO
| Item | Settle in advance | Risk if left open |
|---|---|---|
| Fabric | Whether the fabric can be reordered or held, the dye lot approach, and any mill minimum for a repeat | A repeat needs a fresh fabric booking and the style misses its window |
| Colour | Approved colour standard and how shade will be matched on a repeat | The reorder arrives in a visibly different shade from the test stock |
| Trims and labels | Which trims are common and available, and which need a long lead | A single custom trim delays the whole repeat |
| Decoration or wash | Whether the process is repeatable at volume and who performs it | The test looks good but the process cannot be scaled |
| Factory capacity | An indicative slot for a repeat if the gate is passed, and the date by which it must be confirmed | Capacity is given to other programmes while you read the result |
| Specification lock | Which elements are frozen between test and bulk | Changes made after the test mean the winner is no longer the thing that won |
A planning checklist, not a statement about any particular factory. Confirm each item with your supplier.
Holding materials costs money, so do not hold everything for every style. A sensible approach is to hold or reserve for the options where the test brief already gives a credible chance of winning, and to leave the long shots uncommitted. Ask the factory what can be reserved without a firm purchase, and what the cost of holding is. Flexible terms for this may be part of a wider operating model; see our guide on building flexible sourcing for the two-lane version used by larger retailers.
Worked Example: Three Styles at 500 Pieces Each, With a Decision Gate
The example below is illustrative and uses invented numbers throughout. It is not a benchmark, a market result or a supplier price. It shows how a buyer would test a new style alongside two others, read the results at a single gate and reach a decision. The assumptions are listed first so that every figure can be checked.
- Three new hoodie styles (A, B, C), one colour each, 500 units received per style, all sold at the same assumed full price of £40 excluding VAT.
- Assumed landed cost of £16 per unit, so the test exposure is 500 × £16 = £8,000 per style and £24,000 for the three.
- A four-week read window. Any unit unsold at the gate is assumed to clear later at a markdown price of £24 (40 per cent off £40), purely to test margin.
- Conversion is orders divided by sessions on the product page, assuming one unit per order. Returns are units returned divided by units sold.
Illustrative results at the four-week read
| Measure | Style A | Style B | Style C |
|---|---|---|---|
| Units received | 500 | 500 | 500 |
| Full-price units sold (week 4) | 310 | 190 | 95 |
| Sell-through (sold ÷ received) | 62.0% | 38.0% | 19.0% |
| Product-page sessions | 9,300 | 7,600 | 8,900 |
| Conversion (sold ÷ sessions) | 3.33% | 2.50% | 1.07% |
| Units returned | 22 | 25 | 14 |
| Return rate (returned ÷ sold) | 7.1% | 13.2% | 14.7% |
| Revenue if remainder clears at £24 | £16,960 | £15,040 | £13,520 |
| Gross margin on that revenue | 52.8% | 46.8% | 40.8% |
Illustrative. Revenue = sold × £40 + unsold × £24. Gross margin = (revenue minus 500 × £16) divided by revenue. Returns are tracked separately and not netted off here.
Style A sold more than half of its stock in four weeks, converts well and has the lowest return rate. Style B converts reasonably, so customers who saw it did buy, but it sold fewer units and its return rate is higher. Style C had comparable traffic to A and B yet barely converted. Traffic is not the problem for C; the product, price or presentation is.
Illustrative four-week sell-through by style
% of units received sold in four weeks
Illustrative sell-through (units sold divided by 500 received) from the worked example above. Not a benchmark.
The gate then turns the numbers into actions. The thresholds below are the buyer’s own and are illustrative; your business should set its own for the price point, season and customer.
Illustrative decision gate
| Metric | Threshold (set by the buyer, illustrative) | A | B | C |
|---|---|---|---|---|
| Sell-through at week 4 | At least 55% to scale; 30–55% to hold; under 30% to exit | 62.0%: scale | 38.0%: hold | 19.0%: exit |
| Conversion | At least 2.0% (buyer’s assumed category baseline) | 3.33%: pass | 2.50%: pass | 1.07%: fail |
| Return rate | No more than 10% | 7.1%: pass | 13.2%: fail | 14.7%: fail |
| Gross margin after clearing | At least 50% | 52.8%: pass | 46.8%: fail | 40.8%: fail |
| Action | All four pass to scale; mixed to hold with a named fix; majority fail to exit | Scale | Hold and fix | Exit |
Thresholds are illustrative placeholders and are not benchmarks. Set your own, in writing, before the PO.
Style A is the scale candidate, but the gate is not finished yet. The size split matters, because it decides what the repeat should look like and how urgent it is.
Style A: planned and actual sizes at week 4 (illustrative)
| Size | Planned units | Sold | Share of sales | Sell-through | Left | Weeks of cover |
|---|---|---|---|---|---|---|
| XS | 25 | 9 | 2.9% | 36.0% | 16 | 7.1 |
| S | 75 | 41 | 13.2% | 54.7% | 34 | 3.3 |
| M | 150 | 105 | 33.9% | 70.0% | 45 | 1.7 |
| L | 125 | 98 | 31.6% | 78.4% | 27 | 1.1 |
| XL | 75 | 36 | 11.6% | 48.0% | 39 | 4.3 |
| XXL | 50 | 21 | 6.8% | 42.0% | 29 | 5.5 |
| Total | 500 | 310 | 100% | 62.0% | 190 | 2.5 |
Illustrative. Weeks of cover = units left divided by average weekly sales over the four weeks (sold ÷ 4). Total: 190 ÷ 77.5 = 2.5 weeks.
Sizes M and L produce 65.5 per cent of the sales against 55 per cent of the plan, and L has about one week of cover left. If a repeat takes longer than that to arrive, those sizes will be out of stock while the others still hold stock. This is the reason the replenishment plan in step 4 is agreed before the test: the repeat size curve should lean toward M and L, and the capacity and materials should already be waiting.
An illustrative test timeline (relative, not a lead-time claim)
- Before the POCriteria, thresholds, channel and replenishment plan written into the test brief.
- PO to landingProduction of the test quantity. Duration is set by the product and the factory; confirm it for your programme.
- Launch dayAll sizes live and visible; price, placement and promotion as planned for scale.
- Week 2Early-signal check only: sessions, conversion, first return reasons. No scale decision yet.
- Week 4Read window closes. Compile the four KPIs and the size read.
- Decision gateScale, hold or exit, with owner, action and date recorded.
Illustrative sequence. The read window should be long enough to see the sales rate settle and short enough to replenish inside the season.
Reading the Gate: Scale, Hold or Exit
A gate is only useful if each outcome has an action and a named owner. Scale means placing a repeat or larger order against the replenishment plan agreed earlier. Hold means the evidence is mixed and the buyer extends the read or runs a defined fix. Exit means the remaining stock is cleared on a plan and the style closes. The table sets out the signals that tend to sit behind each outcome.
Mixed signals: what to check and what to do
| Signal | Possible cause | Check | Action |
|---|---|---|---|
| High conversion, low sell-through | Not enough traffic or visibility | Placement, stock availability, channel reach | Hold; extend the read with better exposure |
| Low conversion, high traffic | Price, product or imagery is off | Price against comparable styles, page content, size availability | Exit, or one controlled change then retest |
| Good sell-through, high returns | Fit, size or quality problem | Return reason codes by size; measure returned garments | Hold; fix the fit or quality and recheck before scaling |
| Strong sales in a few sizes only | Size curve mismatch or stockout | Sell-through and cover by size | Scale with a revised size curve |
| Good full-price sell-through, thin margin | Cost too high for the price | Landed cost, price architecture, value engineering options | Hold; revise cost before the repeat |
Qualitative decision aid. Thresholds and actions are for the buyer to set.
The most valuable discipline is deciding in advance what evidence would change your mind. If Style B’s returns fall inside the threshold after a fit adjustment and a re-read, it can move from hold to scale. If a repeat of Style A would land after the point at which the season’s demand falls away, a strong gate result may still end in a smaller repeat or none. The KPI to watch is the one that triggers your action, so name it in the brief. If you want to understand the reorder timing side, our guide on when to reorder a bestseller covers weeks of cover and lead time in detail.
What a Small Test Costs, and When It Is Not Worth It
Testing is not free. A small buy usually carries a higher cost per unit than a volume order because fixed costs such as pattern work, setup and fabric minimums are spread over fewer pieces. The question is whether that premium is smaller than the cost of being wrong at scale. In the worked example, the exposure at assumed landed cost was £8,000 per style. A single speculative buy of ten times that quantity would be 5,000 × £16 = £80,000 at risk on a style with no evidence behind it.
Is a style test worth it?
When a test pays
- Caps the cost of a wrong guess and gives real evidence on price, fit and returns.
- Lets design try more ideas and gives planning a stronger case for the winners.
- Surfaces fit and quality issues before volume is committed.
When it may not
- Higher unit cost on a small run, and the same development work for each style.
- Slower to market than an immediate volume buy if the style would have won anyway.
- A small sample can mislead if the channel or size curve is unrepresentative.
Qualitative judgement. The balance depends on the style, the season and your tolerance for stock risk.
A test is least useful where the risk is already low or the learning is already available: a repeat colour of a proven style, a late-season buy with no time to replenish, or a style with a long fabric lead time that makes any repeat arrive too late. In those cases decide the buy size with your existing data. Where the question is truly open, a test earns its premium. Our guide on reducing overproduction without reducing sales shows how staged commitments fit alongside a protected core range.
Mistakes to Avoid When You Test a New Style
Most problems come from the set-up rather than the product. These are the failures that appear again and again, and each can be prevented in the test brief.
Pre-flight checks for the test brief
- Thresholds are written down, with an owner, before the PO is raised.
- Only one variable changes per option, or the confounding is acknowledged.
- Sell-through is quoted with its denominator and its read period.
- Sizes are live from launch and stockouts are recorded by size.
- Returns are read after the returns window has run, not before.
- Replenishment materials and capacity are agreed before the test, not after.
- The specification is frozen between test and repeat, so the winner is the thing that repeats.
One more mistake sits outside the brief: treating a test win as a forecast. A style that sells 62 per cent in four weeks in a small, well-chosen channel will not necessarily sell at the same rate across the whole business. Use the result to justify a staged scale-up, with a second gate, not a single leap to volume.
Which single number would make you reorder this style, and have you written it down, with its owner, before the PO is raised?
How Ruhrose Approaches Testing a New Style
Ruhrose develops hoodies, jersey, denim and woven casualwear for UK buyers, with development and production in Bangladesh. Our test and scale model is built around the steps above: sample the style first, agree the replenishment plan, then run a small production quantity. We propose 500 pieces per style and colour as a test quantity where it is commercially and technically feasible, which depends on the fabric, dye, trims and factory. It is a proposal, not a guaranteed minimum order, and the model reduces risk without making it disappear. We do not quote prices, minimums or lead times in guides: those are agreed per programme. If you are planning a test, the test and scale model explains how we work through it.