It is the second week of November. Your hero SKU has 180 units on hand, your factory needs 30 days, and marketing has just decided the Black Friday discount will be 35% instead of last year's 25%. Nothing you order today arrives in time. That is what bad BFCM inventory planning feels like from the inside — not a maths error, but a sequencing error. This guide gives you the sizing method instead of the platitudes: how much stock to order for Black Friday, worked out per SKU from your own velocity, your own uplift history, and the promotion you have actually committed to running.
Key takeaways
- Size BFCM per SKU from baseline daily velocity measured on in-stock days only — never from last year's total units.
- Apply that SKU's own uplift factor, then adjust for how much deeper your discount and ad spend are this year.
- Subtract the December trough. Demand pulled forward is not demand created, and a normal December stacked on a 4× BFCM is how you end up with dead stock.
- Two waves beat one: land 70% before the window, hold a booked slot for the December tail. Buffer only the SKUs carrying your ad spend and let the mid-tail sell out.
- The biggest BFCM inventory mistake is ordering to the forecast of a discount nobody has decided on yet.
Start from baseline velocity, not last year's totals
Most BFCM plans start with “we did 4,200 units last November.” That number is an outcome, not an input: it bundles a discount depth, an ad budget, a list size and a product mix that no longer exist.
The usable input is baseline daily velocity per SKU: how many units of that specific variant sell on a normal, fully in-stock day, measured over the trailing 45-60 days before the peak begins. Two rules make it honest:
- Exclude stockout days. A day the SKU was unavailable is not a day it sold zero — it is a day with no data. Averaging those zeros suppresses velocity on exactly the products that sold out because they were popular, which is why Honey Shelf measures velocity on in-stock days only. It is the most common reason a BFCM buy comes out too small.
- Use an October window, not an annual average. Your October catalogue is closer to your November catalogue than March's was. A SKU launched in September has six weeks of baseline and no uplift history — treat it as a category estimate and buy conservatively.
If that history is thin or noisy, our guide to demand forecasting methods for small brands covers how to smooth a short series without inventing signal.
The formula: how much stock to order for Black Friday
The BFCM buy is not one number. It is a peak window, a buffer, and a depressed December — depressed because a chunk of that December was already sold at 35% off during the window.
Each term, defined properly:
- Uplift Factor — last year's window units for this SKU divided by what baseline velocity would have produced over the same days. Per SKU where you have the history; per category where you do not.
- Promo Delta — the correction for the promotion you are running this year versus last. Last year's uplift already contains last year's discount and last year's ad spend, so do not count them twice. This term captures only the change.
- Trough Factor — post-window December velocity as a multiple of baseline. For most D2C brands it lands between 0.6× and 0.8×, offset upward by gifting demand in mid-December.
- Buffer % — a flat percentage of window demand, not a z-score. The usual safety stock formula protects a replenishment cycle; BFCM has no replenishment cycle. You get one shot, so the buffer is sizing forecast error, not lead-time risk.
Worked example: one hero SKU, end to end
A beauty brand's hero SKU is a 3-piece skincare set, produced in 30 days by a contract manufacturer in Maharashtra.
Step 1 — Baseline. Over the trailing 60 days, excluding the 9 days it was out of stock, the set sold at 12 units/day.
Step 2 — Uplift. Last year the SKU sold 252 units across the 5-day window (Thanksgiving Thursday to Cyber Monday) against a baseline-equivalent of 60. Uplift = 252 ÷ 60 = 4.2×.
Step 3 — Promo delta. Last year's 4.2× came from 25% off and ₹3.5 lakh of ad spend. This year: 35% off and ₹5.6 lakh. The brand's planning rule is roughly +12% units per 5 extra points of depth, and it discounts the ad increase heavily because auction CPMs rise for everyone that week:
Step 4 — Window demand.
Step 5 — Window buffer. This is a hero SKU carrying the paid campaign, so it gets 20%: 68 units. Mid-tail SKUs on the same buy get 5-10%; the long tail gets none.
Step 6 — The December trough. Last year, 2-31 December ran at 8.4 units/day against a 12/day baseline — a trough factor of 0.7×. Plan December at the troughed rate, not the baseline rate:
Step 7 — Net it off. On hand at the 18 November cut-off is 180 units, and a PO for 150 units is already landing on 12 November.
That is the point of the method. When the founder asks why 350 and not 600, you can point at the exact assumption that moves the answer.
Typical BFCM uplift ranges by category
If you have no per-SKU history, start here — then throw this table away after your first cycle and use your own numbers. These are planning starting points, not benchmarks. The ranges are wide on purpose, because uplift depends far more on your discount depth, list size and ad budget than on which category you sell in.
| Category | Typical window uplift vs baseline | What breaks first |
|---|---|---|
| Apparel — core basics | 3-5× | The middle of the size curve (M/L) sells out days before XS and XXL |
| Apparel — seasonal / fashion | 2-4× | Hero colourway sells out; the rest becomes January markdown |
| Beauty & personal care | 4-7× | Sets and bundles outsell singles; components run out, not SKUs |
| Jewellery — fashion / gifting | 3-6× | Demand skews late and skews to gift packaging |
| Footwear | 3-5× | Size curve, plus a returns wave that lands in January |
| Home & lifestyle | 2.5-4× | Higher AOV means slower decisions and a longer tail |
| Food & beverage | 2-3× | Shelf life caps the buy before the forecast does |
| Small accessories / impulse gifting | 4-8× | Highest variance of any category — buy the low end |
One structural note for bundles and kits: uplift applies to the finished good, but the shortage happens at component level. A 6× uplift on a gift set is a 6× uplift on every line of its bill of materials, including the box. Run the multiplier through the production and material-purchase stages before you sign the PO.
Order in two waves, not one
A single purchase order commits your whole season on the worst information you will have all year. Split it where lead time allows:
- Wave 1 — roughly 70% of plan. Must land, be received, QC'd and live on the store before the window opens. This is the part you cannot get wrong, so it covers window demand plus the full buffer.
- Wave 2 — a booked slot, not a quantity. Reserve capacity at the factory with the quantity confirmable up to a stated date. Be honest about the physics: with a 30-day lead time, wave 2 cannot land inside a 5-day window. What it can do is cover the December gifting tail and the January restock, sized against real BFCM data instead of a forecast.
Booking capacity rather than units is the negotiation that matters. Factories in India are full from September onward, and most will hold a slot if you commit to a floor quantity and a confirmation deadline.
Decide now which SKUs are allowed to sell out
You cannot protect the whole catalogue, and trying to is how brands end up with 20% more inventory and the same stockouts. Sort it into three written lists before you order:
- Never out of stock (5-8 SKUs maximum). These carry the paid campaign, the email hero slot and the retargeting pool. A stockout here does not just lose the sale — it burns ad spend at its most expensive all year. Full buffer, wave 1 only.
- Allowed to sell out. Mid-tail products bought to forecast with little or no buffer. Selling out on day three costs a handful of orders; overbuying costs a January markdown and the dead stock that follows. Plan the “notify me” flow instead of the extra units.
- Not promoted at all. Thin-margin items, anything supply-constrained, and anything you want at full price in December. Excluding a SKU from the promo is an inventory decision too — it drops out of the uplift calculation entirely.
The list is not a ranking of favourites. It is a ranking of what your ad spend is pointed at. Our stockout playbook goes deeper on which SKUs justify protection year-round.
Work backwards to your cut-off dates
BFCM inventory planning fails in November because the real decision date was in August. Work backwards from the window. Assume a 5-day window opening 26 November, 30-day production, 15-day material lead time:
| Milestone | Domestic (India → India 3PL) | Sea freight (India → US/EU 3PL) |
|---|---|---|
| Live on store, pick-ready | 18 Nov | 18 Nov |
| Received, QC'd, put away | 14 Nov (4 days) | 4 Nov (14 days — 3PLs are backed up in November) |
| Dispatch from factory | 9 Nov | 25 Sep (~40 days transit + customs) |
| Production complete | 8 Nov | 24 Sep |
| PO issued, production starts | 9 Oct | 25 Aug |
| Materials ordered | 24 Sep | 10 Aug |
| Promo plan locked | 20 Sep | 5 Aug |
Read the last row again. If you ship by sea, your Black Friday discount is an August decision. Build this table once for your own lead times and put the dates in a shared calendar — the same backwards planning drives your reorder points all year, with less at stake. Then add slack for what always slips: dye lots that fail QC, the labour gap around Diwali, and customs. Read it alongside our festive season planning guide — the Indian festive calendar eats the same factory capacity two months earlier.
What to do with unsold BFCM stock in December
Some leftover is correct. Finishing 1 December at zero across the board means you underbought. The mistake is panicking on 2 December and marking down into the strongest gifting month of the year.
- 1-8 December: hold price. Traffic is still elevated and the discount hunters have already bought. Compete on delivery speed.
- 9-18 December: bundle, do not discount. Gifting demand peaks against shipping cut-offs. A set at full price with free gift wrap converts better than the same units at 30% off, and protects your price perception.
- 19-24 December: last-minute and digital. Whatever cannot arrive in time is a gift card sale, not a discount opportunity.
- 26 December onward: this is when you clear. Now discount hard, in one decisive move rather than three timid ones.
Then split what remains by shelf life, not sales rank. Carrying cost runs roughly 20-25% of inventory value a year — about 2% a month — so holding an evergreen unit for three months costs about 6%, usually cheaper than a 30% markdown. Seasonal prints and holiday packaging cannot be carried at all. Clear those while they still say “gift.”
The biggest BFCM inventory mistake
It is not under-ordering and it is not over-ordering. It is ordering to the forecast of a discount you have not yet decided on.
The pattern is always the same. Ops builds the buy in September on last year's uplift. Marketing decides in early November to go 40% off instead of 25% and doubles the ad budget. The forecast was correct — for a different promotion. The hero SKU is gone by Saturday morning while ₹5 lakh of paid traffic lands on a sold-out page, and the mid-tail nobody promoted sits untouched until February.
The fix is a sequencing rule, not a better model: the promotion is locked before the purchase order is issued. Discount depth, which SKUs are in and out, and the ad budget ceiling get signed off at the cut-off date in your calendar. After that, the promo plan is an input to a manufacturing commitment, not a marketing opinion.
If marketing genuinely cannot commit that far ahead, invert the constraint: buy the conservative version and cap the discount at what the stock supports. Discount depth is a lever you can still pull on 20 November. Inventory is not.