Collection gates (stage-gate): a funnel of hypotheses, the price of a hit, the saving against “everything into a trial” and the stop rule of the pilot batch
From the number of ideas, the cost of the stages per model and the share of models passing each gate (idea, tech pack, sample, pilot batch) it counts how many models reach a hit, how much is spent in all, what one hit costs, what part of the spend went to rejected models and how much the gates save against “all ideas go to the end”. Separately it counts the stop rule of the pilot batch: the sales threshold at the control week, the loss on stopping and the loss avoided by not scaling blindly. Take the pass shares from past seasons; here illustrative.
At a glance
- What question does it answer?
- How many models will reach a hit, what one hit costs, how much the gates save and at what pilot result the batch is stopped.
- What data is needed?
- The number of ideas and the profit of a hit for the season from past seasons.
- How to read the result?
- Example on the default values (20 ideas, four gates with pass shares 60, 50, 60 and 35%): hits 1.3, spent RUB 1.64 million, the price of a hit RUB 1.3 million; without gates the spending RUB 6.83 million, the saving RUB 5.19 million. The pilot stop threshold 30 units by week 3; on stopping we lose RUB 87,000, on a failed scale without a pilot RUB 1.31 million. Example on the default values: Hits from the funnel: 1.3; Share of ideas that became hits: 6.3%; Total spent on the funnel: RUB 1.64 million; The price of one hit: RUB 1.3 million. The figures are illustrative: enter your own to see your result.
Why and how to reason
Expensive collection mistakes are made at the late stages: the pilot batch and the scale cost tens of times more than a sketch. Gates put a check after each stage with a criterion set in advance, and a weak model is stopped while it is cheap. The funnel is counted simply: ideas × the product of the pass shares of the previous gates enter a stage, the stage spending equals the entered × the stage cost per model, hits equal ideas × the product of all shares. The price of a hit includes the rejected models too: it is all the spending divided by hits. On the default values of 20 ideas 1.3 models reach a hit, RUB 1.64 million is spent in all, one hit costs RUB 1.3 million. Without gates, if all ideas go to the pilot batch, the spending is RUB 6.83 million: the gates save RUB 5.19 million. The pilot stop rule: sell 30 units by the control week; otherwise stopping costs RUB 87,000.
When to use it
Apply it when planning a season and when launching a new line: when there are more ideas than you can bring to a batch, when you want to remove decisions “by feel” and when you need to explain to partners or an investor how you select models. It does not suit basic items with known demand: a re-order is enough for them. Take the pass shares from past seasons; a new brand does not have them, take cautious ones and recalculate after the first season. For footwear, accessories and children’s a sample and a pilot cost more and the pass shares differ: count from your category.
How to read the result
Look at four numbers. Hits and their share show whether the funnel is enough: if there are fewer than one hit, the season is built on luck. The price of a hit is compared with the profit of a hit: 0.7 hits are needed to pay back the funnel, and the expected profit is RUB 1.13 million. The share of spending on rejected models, 74%, is normal: this is how selection works; it is alarming if it grows at the expense of late stages. The stop rule forces you to name the threshold in advance: 30 units by the control week, a pace of 10 units a week. The difference in losses, RUB 1.22 million, shows that the pilot protects against blind scaling.
Common mistakes
They set the stop threshold after sales start and move it to fit the desired result. They let weak ideas through “out of friendship”, and the funnel turns into a formality. They count only the spending of the pilot and not the price of rejected models. They scale on the first days of sales without waiting for the control week. They take the pass shares “from the ceiling” instead of past seasons. They treat a fractional number of hits as exact: on a small number of models one hit changes the picture.
What data is needed?
- The number of ideas and the profit of a hit for the season from past seasons.
- The funnel stages: the cost per model and the share passing the gate.
- The pilot batch, the control week, the sell-through threshold, the recovery of leftovers and the scale batch.
Inputs
| Field | Unit | Value in the example | Meaning |
|---|---|---|---|
| Ideas at the entry | units | 20 | How many model hypotheses you put into the funnel for the season. The example value is illustrative. |
| Profit of one hit for the season | ₽ | 2,200,000 | How much a model that passed all gates and went to scale earns: profit after all costs for the season. The example value is illustrative. |
| Cost of a unit | ₽ | 1,450 | The cost of a unit in the warehouse without VAT: the leftovers of the pilot and the scale are counted by it. The example value is illustrative. |
| The pilot batch | units | 150 | How many units of each model go into the pilot: enough to see sales and little enough not to risk. The example value is illustrative. |
| Control week of the pilot | weeks | 3 | After how many weeks of sales you check against the threshold and decide: scale or stop. The example value is illustrative. |
| Sell-through threshold at the control week | % | 20 | What share of the pilot batch must be sold by the control week to go to scale; below the threshold the model is stopped. The example value is illustrative. |
| Recovery of leftovers as a share of cost | % | 50 | What part of the cost can be recovered by selling the leftovers at a discount. The example value is illustrative. |
| The scale batch | units | 3,000 | How many units would be ordered after a successful pilot: needed to estimate the loss if scaled without a pilot. The example value is illustrative. |
| Unsold share of the scale batch in a failure | % | 60 | What part of the scale batch stays unsold if the model turned out weak. The example value is illustrative. |
How it is calculated
| Step | Formula |
|---|---|
| The funnel | ideas × the product of the pass shares of the previous gates enter a stage |
| Spending | Σ entered × the stage cost per model |
| Hits and the price of a hit | hits = ideas × Π shares; price = total spend ÷ hits |
| The gates’ saving | ideas × Σ stage costs − spending with gates |
| The stop rule | threshold = pilot × sell-through threshold; the loss on stopping = leftover × cost × (1 − recovery of leftovers) |
Worked numbers
Default values from the app; in the interactive version you enter your own.
- Hits from the funnel
- 1.3
- Share of ideas that became hits
- 6.3 %
- Total spent on the funnel
- 1.64 m ₽
- The price of one hit
- 1.3 m ₽
- Spending on models that did not reach a hit
- 1.21 m ₽
- Share of spending on rejected models
- 74 %
- Spending without gates: all ideas go to the end
- 6.83 m ₽
- Saving from the gates
- 5.19 m ₽
- Expected profit of the funnel
- 1.13 m ₽
- Hits needed to pay back the funnel
- 0.7
- Stop threshold: units to sell by the control week
- 30
- Pilot pace needed for scale, units a week
- 10
- Loss on stopping at the threshold
- 87,000 ₽
- Loss on a failed scale without a pilot
- 1.31 m ₽
- Loss avoided by the pilot
- 1.22 m ₽
Data as a table
| Measure | Models |
|---|---|
| 1. Idea and hypothesis: brief, references, target calculation | 20 |
| 2. Sketch and tech pack | 12 |
| 3. Sample and fit test | 6 |
| 4. Pilot batch and sales | 3.6 |
| Hits | 1.3 |
- With gates
- Without gates
Data as a table
| Measure | With gates | Without gates |
|---|---|---|
| 1. Idea and hypothesis: brief, references, target calculation | 80k | 80k |
| 2. Sketch and tech pack | 300k | 500k |
| 3. Sample and fit test | 330k | 1.1m |
| 4. Pilot batch and sales | 927k | 5.15m |
The funnel by stage
| Stage | Models at the entry | Cost per model | Stage spend | Share passing | Reached the end of the stage |
|---|---|---|---|---|---|
| 1. Idea and hypothesis: brief, references, target calculation | 20 | 4,000 ₽ | 80,000 ₽ | 60 % | 12 |
| 2. Sketch and tech pack | 12 | 25,000 ₽ | 300,000 ₽ | 50 % | 6 |
| 3. Sample and fit test | 6 | 55,000 ₽ | 330,000 ₽ | 60 % | 3.6 |
| 4. Pilot batch and sales | 3.6 | 257,500 ₽ | 927,000 ₽ | 35 % | 1.3 |
Gates save money while the pass shares are honest: if weak ideas pass “out of friendship” the funnel turns into a formality. Set the pilot threshold before sales start and do not move it along the way. Estimate the pass shares and the profit of a hit from past seasons and hold them as a range: on a small number of models one hit changes the picture. For footwear, accessories and children’s the cost of a sample, of the pilot and the pass shares are different: count from your category.
Nuances and limits
- Setting the stop threshold after sales start: it is moved to fit the desired result.
- Letting weak ideas through “out of friendship”: the funnel turns into a formality.
- Counting only the spending of the pilot and not the price of rejected models: it is part of the price of a hit.
- Scaling a model on the first days of sales without waiting for the control week.