What it does. Replaces "let's give it another week" with a rule: when to take the first markdown, how deep, and what to do with the money. A rule is not a formula for every business — it is a written trigger you calibrate and keep.
Inputs. Sell-through by week of life; weeks of cover; weeks left in the season; the category's expected curve; the margin at full price and at each markdown depth; the cost of holding stock (space, cash, handling).
Steps.
1. Set the trigger. Define "behind" for the category: for example, ST% below 70% of the expected value for the week, two weeks running. One miss is noise; two is a pattern.
2. Set the first depth by distance. Slightly behind → 20–25%; clearly behind → 30–40%; far behind with long cover → 50%. Depth follows how far behind the style is, not how much stock sits on it.
3. Set the calendar. The first markdown is always cheaper than the second: a planned date for the first review (for example week 6 of 16) beats waiting for the sale. Late season, the rule flips: if weeks of cover are shorter than weeks left, hold — the stock will clear at full price.
4. Protect what works. Never mark down a style whose full-price sell-through is on track to sell one size that is stuck; fix the size, not the price.
5. Record. Every markdown gets a line: style, week, depth, reason (depth / timing / price), and the margin it cost. This log is the raw material for the next range plan.
6. Review. At season end, compute the markdown rate (share of sales below full price) and split it into planned and reactive. Reactive above the planned level is the buy telling you where it went wrong.
Common errors. Marking down on units on hand rather than on the curve. Uniform depth for every style ("everything −30%"). Waiting for the seasonal sale to clear what week 6 already announced. Forgetting that markdown trains the customer: frequent small cuts teach waiting faster than one decisive one.
What a good output looks like. A written rule of no more than a page: trigger, depth ladder, calendar, protections, log format. Applied for one season, it produces the two numbers the next plan needs — planned markdown rate and reactive markdown rate — and the list of styles that caused the second.