Retail Planby RetailNorthstar

How to reforecast in-season

An in-season reforecast re-projects the remaining season — sales, receipts, markdowns and closing stock — from actuals to date, and re-derives the open-to-buy from the result. It is not a re-plan. The pre-season sales plan stays frozen as the yardstick; the reforecast is a versioned projection beside it, and the season total moves only when someone with the authority to move it decides to.

This guide is the operational how-to. For the cadence question — how often, derived from the last point a decision can still change — read how often to reforecast a merchandise plan on RetailNorthstar. For the plan being reforecast, read how to set a sales plan and how to plan receipt flow; for the weekly grid it reads from, how to read a WSSI.

What this guide covers: triggers, scope, the seven-step method, a worked month-three reforecast, reading the outputs, the six-lever decision menu, governance, and vertical differences. All figures are illustrative — not benchmarks, not drawn from any brand.

The short version
Freeze the original plan. Replace plan with actuals for every closed period and let closing stock fall out of the identity. Hold committed receipts and the season exit; re-forecast sales and markdowns by category from run-rate, trend and last year’s shape. Rebuild closing stock targets from the revised sales, re-solve required receipts, net the commitments, and read the new open-to-buy front to back. Then choose — chase, cancel, push, re-allocate, markdown or hold — cheapest lever first. Sign it off, version it, write down why.
Definition — In-season reforecast
An in-season reforecast is the re-projection of a season’s remaining periods from actuals to date: closed periods restated with actual sales, markdowns and receipts, the balance re-forecast by category, closing stock targets rebuilt from the revised sales, and required receipts and open-to-buy re-derived from the inventory identity — with the original plan frozen beside it.
Required receipts = revised closing stock + revised sales + revised markdowns − opening stock; open-to-buy = required receipts − committed on order
Used by: Merchandise planners, buyers, allocators and trading teams, monthly with a weekly WSSI read
Related: Sales plan, open-to-buy, WSSI, forward cover, stock-to-sales ratio, markdown plan

Why the distinction matters

A plan is a commitment; a forecast is an expectation. The pre-season plan committed money — to receipts, a markdown budget, a margin the season promised — and those commitments are no less real because the first eight weeks traded below them. A re-plan rewrites the commitment; a reforecast measures the season against it so the decisions that can still change the outcome are taken while they exist. Conflate the two and the plan drifts toward the actuals every cycle, until the hindsight compares the outcome to a forecast adjusted to match it. The forward view is re-solved every cycle; the total moves only when its owner decides, with a reason, because it reprices everything downstream at once.

Calendar cadence and threshold breach

Two triggers, agreed before the season starts. The first is the calendar: a fixed cadence on which the full reforecast runs whether or not anything looks wrong — monthly for a seasonal business, because that is the open-to-buy cycle a changed sales view can still move, with the WSSI read weekly in between. It exists because the absence of an alarm is not evidence the plan is right; it means nobody has looked.

The second is a threshold breach: a pre-agreed tolerance on cumulative variance that, once exceeded for several consecutive weeks at the same category or channel, forces a reforecast between calendar slots. It is tested at the lines, because a total can sit inside tolerance while two categories cancel out. A material event is a breach in its own right — a slipped delivery, a cut style, a door closure. Neither trigger fires on one bad week: a phasing miss self-corrects across adjacent weeks, a level miss repeats, and it takes several weeks to tell them apart.

What is re-forecast and what is held

Three lines behave differently. The sales forecast for the remaining periods is re-projected — that is the point. The markdown forecast moves with it, because a smaller season with the same receipts has more to clear, and the markdown line has to say so in the cycle the decision is taken rather than at the exit. Receipt commitments are held: goods shipped, orders past the cancellation point and confirmed prebooks are facts, carried as on-order against the periods they land in.

The season exit stock is held as policy, not re-forecast as an outcome. The carry-out is set by when the space and the open-to-buy are needed for next season, and that does not change because this season traded soft. Holding the exit is what makes the reforecast bite: with the exit fixed and sales lower, the surplus has nowhere to go but fewer receipts or more markdown. Between the two sit the intermediate closing stock targets, rebuilt from the revised sales by applying the planned stock-to-sales ratio or forward-cover target, so stock falls with the sales.

Seven steps, three of them arithmetic

Steps two, four and five are the arithmetic — restate actuals, project the balance, rebuild stock and open-to-buy — and a spreadsheet can do them. The other four are governance, and they decide whether the arithmetic changes anything.

  1. 1

    Confirm the trigger, then freeze the original plan

    Run it because the calendar says so or a threshold was breached — never because one week disappointed. Before touching a number, lock the pre-season plan as a saved version: it is the yardstick every hindsight is measured against, and once overwritten the season can only say how wrong it became, not how wrong it was.

  2. 2

    Restate actuals to date

    Replace planned sales, markdowns and receipts with actuals for every closed period and let closing stock fall out of the identity: closing stock = opening stock + receipts − sales − markdowns. Reconcile it to the stock ledger before going further; a reforecast built on a closing stock the ledger does not recognise is wrong from its first cell.

  3. 3

    Decide what is re-forecast and what is held

    Sales and markdowns for the remaining periods are re-projected. Receipts already committed — shipped, past the cancellation point, prebooked with dealers — are held as facts and carried as on-order. The season exit stock is held as policy, because the space and the open-to-buy are needed next season regardless. Write the split down; it is the scope.

  4. 4

    Project the balance of the season by category

    Re-forecast the remaining periods line by line, never by scaling the total. Blend three views per category — the run-rate from the weeks traded, its trend, and last year’s shape for the periods ahead — weighted toward last year’s shape early, when the run-rate is a few noisy weeks, and toward the run-rate late. Then answer what the blend cannot: less demand, or later demand?

  5. 5

    Rebuild closing stock and re-solve open-to-buy

    Re-derive each remaining closing stock target by applying the planned stock-to-sales ratio or forward-cover target to the revised sales, holding the exit. Solve required receipts from the identity in period order: required receipts = revised closing stock + revised sales + revised markdowns − opening stock. Open-to-buy is required receipts less committed on-order; negative is the overbought.

  6. 6

    Read the outputs and choose from the decision menu

    Three readings come out: which periods are over- or under-bought, what forward cover the revised stock represents against the revised sales, and where trailing weeks-of-supply disagrees with forward cover. Each maps to a lever — chase, cancel, push, re-allocate, markdown or hold — and the levers expire cheapest first. Choose deliberately and attach the reason.

  7. 7

    Sign off, version, and record the narrative

    The revised forecast becomes the working plan only when the owner of the season total signs it. Save it as a numbered version beside the frozen original and record, against every material change, what moved and why. The numbers are superseded within weeks; the narrative is what stops next season inheriting this season’s accidents as demand.

A month-three reforecast on the shared six-month season

The same illustrative six-month season used in the sales-plan, receipt-flow and markdown-plan guides, in thousands of retail dollars: sales planned at 1,500, markdowns 145, opening stock 520, carry-out 150. Two months have traded; the reforecast runs at the start of month three. Plan receipts are shown without the small shrink provision the receipt-flow guide carries, so every row can be checked by eye. First, the restated actuals.

Hypothetical actuals for months one and two against plan, in thousands of retail dollars: sales, markdowns, receipts and closing stock, with variances.
LineM1M2Season to date
Sales plan180240420
Sales actual168222390
Sales variance−12−18−30
Markdowns plan / actual0 / 00 / 00 / 0
Receipts plan220280500
Receipts actual220265485
Receipts variance0−15−15
Closing stock plan560600
Closing stock actual572615
Closing stock variance+12+15

Sales are 390 against 420 to date — a little over seven percent behind — at the same categories in both months: a level miss, not a phasing miss. Receipts landed at 485 against 500 because 15 slipped from month two into month three. Closing stock is 615 against 600: the sales shortfall added 30, the slipped receipt took 15 away. Opening month three with 615 against revised sales of 280 is a stock-to-sales ratio near 2.2 where the plan had 2.0 — the surplus is visible before any projection is made. Full-price sell-through on the goods that were on the floor ran below plan rate, so this is read as a smaller season, not a later one.

Hypothetical balance-of-season reforecast for months three to six, in thousands of retail dollars: original and revised sales and markdowns, revised opening and closing stock, required receipts solved from the identity, committed on-order, and the resulting open-to-buy by period.
LineM3M4M5M6Balance
Original sales plan3003402601801080
Revised sales forecast2803202451751020
Original markdown plan10204570145
Revised markdown forecast10255080165
Opening stock (revised)615544441315
Closing stock target (revised)544441315150
Required receipts (solved)21924216990720
Committed on order30525012030705
Open-to-buy (required − committed)−86−8+49+60+15

Revised sales for months three to six are 1,020 against 1,080; the season is now 1,410 against the 1,500 commitment — a difference the owner of the total acknowledges, not one the spreadsheet makes quietly. Markdowns rise from 145 to 165 because the surplus must clear by the same exit. Closing stock targets are rebuilt by applying the original stock-to-sales ratios to the revised sales, rounded down toward the exit — month four opens at 544, about 1.7 times its revised sales, as planned — and the exit is held at 150. Required receipts fall out of the identity: month three 544 + 280 + 10 − 615 = 219; month four 441 + 320 + 25 − 544 = 242; month five 315 + 245 + 50 − 441 = 169; month six 150 + 175 + 80 − 315 = 90.

The balance requires 720 where the original plan, including the 15 that slipped, still had 790 to land — the reforecast has taken 70 off the requirement. Against that, 705 is committed. Before the reforecast the open-to-buy was 85 of room; after it, the season nets to 15, and the net hides the shape: month three is overbought by 86 and month four by 8, while months five and six show 49 and 60 of room. If the month-three commitments land in full, month three closes at 615 + 305 − 280 − 10 = 630, 86 above target — the same number from the stock side.

Every chain closes: 615 + 219 − 280 − 10 = 544; 544 + 242 − 320 − 25 = 441; 441 + 169 − 245 − 50 = 315; 315 + 90 − 175 − 80 = 150. And the season: 520 opening + 1,205 of receipts (485 landed, 720 required) − 1,410 sales − 165 markdowns = 150 carry-out. A reforecast that does not reconcile this way has a typed number in it, and the typed number is the one that is wrong.

Over-bought, under-bought, and the two kinds of cover

Read the open-to-buy row by period, front to back. A negative figure is that period’s overbought; a positive figure is room — but room at the back of a season whose front is overbought is what remains after the surplus rolls through, and spending it first plans the same stock twice. A surplus in month three becomes month four’s opening stock whether or not anyone decides it should.

The second reading is forward cover: revised opening stock over the revised sales ahead. Because the denominator is the new forecast, a reforecast changes every period’s cover without moving a unit — a line whose forward cover now exceeds the trading weeks left has surplus already committed to clearance. The third is where forward cover and trailing weeks-of-supply disagree: going into a peak, trailing WOS flatters cover; coming out of one, it hides overstock. Where they disagree the season is turning, and the trigger belongs on forward cover.

Six levers, in the order they expire

The arithmetic produces readings; the readings need decisions. Six levers cover the ground, and they expire roughly in the order listed, cheapest first — which is why the timing of a reforecast matters as much as its accuracy.

Chase

Buy more of what is ahead of forecast — only where a reorder can land with weeks left to sell it. A chase that arrives with fewer weeks remaining than its own cover is a markdown ordered in advance. Open-to-buy says whether the room exists; the T&A calendar says whether the lead time fits.

Cancel

Remove committed receipts before they ship, where contract and production stage still allow. It is the cheapest cure for an overbought and the first lever to expire — once goods are cut or containerised the option is gone. That is why a reforecast belongs at every buy milestone, not only on the reporting calendar.

Push

Move a delivery later without cancelling it, so it lands against periods that now have room rather than periods already heavy. Pushing holds the total and changes the shape — right when demand is later, not smaller. The pushed goods still need weeks to sell, and the exit date does not move with them.

Re-allocate

Move stock between doors, channels or regions before it is repriced. When the miss is uneven — one channel behind, another ahead — surplus and shortage partly cancel with no markdown at all. Usually the lever with the smallest margin cost, and the one that needs the best data on where stock actually is.

Markdown

Reprice the surplus so it clears by its exit date. Once revised forward cover on a line exceeds the trading weeks left, the surplus is committed to clearance and only the price is open. Move the markdown budget explicitly, in the markdown line, so the maintained-margin cost is visible when the decision is taken.

Hold

Do nothing to this line this cycle — as a recorded decision, not an omission. Right when the miss is a phasing miss that adjacent weeks will absorb, when the lever costs more than the problem, or when one more period of data changes the answer. A hold with a reason and a review date is a decision; without either it is a delay.

Applied to the example: the 86 overbought in month three is worked first and in cost order — cancel what remains cancellable of the 305, push what can be pushed into months five and six where the room now sits, re-allocate what has landed toward the doors and channels that are ahead, and only then let the markdown line carry the rest. Nothing is bought against the room at the back until the front is resolved.

Sign-off, versioning, and the frozen original

Three rules make a reforecast an instrument. Ownership: category owners sign their lines and the owner of the season total signs the total, so the revised forecast becomes the working plan by decision rather than by circulation. Versioning: the pre-season plan is frozen before the first reforecast, and every reforecast is saved as a numbered version beside it — never over it — so the season-end hindsight can measure forecast error against the original and reforecast error against each version at the point it was made. The forecast accuracy guide covers why grading a plan that has since been reforecast voids the number.

And the narrative: against every material change, record what moved and why — a late delivery, a stock-out that capped a week, an event that did not repeat — because next season’s planner will meet this season’s actuals as bare figures.

How the cadence and the levers differ

The identity is the same everywhere; what changes is how much of the receipt line is held as fact, how early the decision points fall, and which levers are available. Apparel is the reference case; the differences are mostly lead time, prebook share, and what the exit date is anchored to.

Apparel

Weekly in the WSSI, with a monthly re-solve of the open-to-buy. The levers are broadest here: DTC receipts can be split and pushed, chase is real on core styles with short lead times, and re-allocation works because the style-color matrix trades in many doors at once. Read sell-through by style-color against weeks elapsed; size-curve breakage shows there before it shows in the total.

Footwear

Prebook-heavy, so most receipts are held as fact and the reforecast works the at-once and replenishment share. The distinctive reading is size-run breakage: a style can be on forecast in pairs while the core sizes are gone and only the tail is left — cover in the grid, a markdown in the store. Reforecast pairs by size run wherever the data allows.

Accessories and bags

Split the evergreen core from the seasonal colourways: the core is a replenishment rate and is reforecast as one; the hero colours are a sell-through read against their window. Check attach rate to the apparel or footwear they sell alongside first — an accessories miss is often an apparel miss arriving late.

Home and furniture

Ocean lead times pull every decision point forward: the last date a container can be cancelled or pushed falls months before the goods sell, so the reforecast that can still change the season runs early, on less data. Receipts move in container multiples, so the required figure is rounded to the container and the rounding is itself a decision. Cover is read in weeks of supply on the options and finishes that repeat.

Health and beauty

Mostly replenishment-heavy core, reforecast as a rate against a launch calendar rather than a season total. The distinctive lever is shade rationalisation: a shade range broken at the tail behaves like a broken size run, and the decision is whether to re-buy the tail or narrow the range. Shelf life and period-after-opening put a hard date on stock that no markdown can move, so holding costs more than it looks.

Outdoor and sporting goods

Dealer prebooks fix much of the buy pre-season, and the model-year transition sets the exit: whatever is unsold at changeover meets the new model at full price. The reforecast works the at-once share, the counter-seasonal categories, and the soft-goods versus hard-goods mix with its different cover targets. Where MAP pricing applies the markdown lever is narrow, so cancel, push and re-allocate carry the weight.

Toys and games

Volume concentrates into a few weeks at year end, and the receipts for those weeks are committed long before any sell-through signal exists. The reforecast that matters is the one before the last commitment; after it, the levers are re-allocation between retailers and the post-peak exit. Licensed windows and retailer reset dates fix an end that no revision can extend.

Baby and juvenile hard goods; jewellery and watches

Two cases where the reforecast is a rate more than a season. Juvenile hard goods carry long product lives and registry-driven demand, so it is a replenishment read with a watch on safety and regulatory cycles that can end a product regardless of sell-through. Jewellery and watches carry metal-price exposure on the cost side and a low-markdown culture, so a miss is worked through hold and re-allocation across gifting peaks before repricing is considered.

See the connected workflow in RetailNorthstar

Frequently asked questions

What is an in-season reforecast?
An in-season reforecast is the re-projection of a season’s remaining periods from actuals to date: closed periods are restated with actual sales, markdowns and receipts, the balance is re-forecast by category, closing stock targets are rebuilt from the revised sales, and required receipts and open-to-buy are re-derived from the inventory identity. Unlike a re-plan, the original plan stays frozen as the yardstick, and the season total moves only as an owned decision.
How often should you reforecast in-season?
On a pre-agreed calendar cadence — monthly for a seasonal business, because that is the open-to-buy cycle a changed sales view can still move, with the WSSI read weekly in between — plus a threshold trigger when cumulative variance breaches an agreed tolerance at the same lines for several consecutive weeks, or when a material event lands: a slipped delivery, a cut style, a channel change. The deeper rule is that cadence follows the last point at which a decision can still change.
What is held constant in a reforecast, and what is re-forecast?
Sales and markdowns for the remaining periods are re-forecast. Receipts already committed — shipped, past the cancellation point, prebooked — are held as facts and carried as on-order. The season exit stock is held as policy, because the space and the buying room are needed next season whatever this one did. The stock-to-sales ratio or forward-cover target that rebuilds intermediate closing stock is carried from the original plan but checked each cycle, since the remaining window is shorter than when it was set.
How do you project the balance of the season?
Category by category, from a blend of three views: the run-rate from the weeks traded, the trend in that run-rate, and last year’s shape for the periods ahead. Early in the season last year’s shape deserves most of the weight, because the run-rate is a few noisy weeks; late in the season the run-rate is the demand you actually have. Then the projection must answer what the blend cannot: is the gap less demand, which says cut receipts, or later demand, which says push and hold the total?
How does a reforecast change open-to-buy?
By re-solving required receipts for every remaining period from the revised sales, revised markdowns and rebuilt closing stock targets, then netting the committed on-order against them. A negative figure means the period is overbought by that amount; a positive figure is room, which is not a licence to spend until the overbought periods in front of it are resolved. In the worked example on this page, a month-three reforecast takes seventy off the balance-of-season requirement and leaves the front of the season overbought while the back shows room.
Who signs off a reforecast, and why version it?
The owner of the season total signs it; category owners sign their lines. It is saved as a numbered version beside the frozen pre-season plan, never over it, because the frozen original is the only honest measure of forecast error at season end — overwrite it and the hindsight compares actuals to a forecast adjusted toward them. The narrative attached to each version — what moved and why — is what stops next season restating this season’s accidents as demand.

See how RetailNorthstar keeps the sales plan, the open-to-buy, and the WSSI on one data model — so a reforecast re-solves receipts, cover, and buying room the same day, measured against the original plan.