The In-Season Reforecast

An in-season reforecast is the mid-season revision of the sales, receipt, and inventory plan against actual sell-through — the point where a merchandising team replaces the pre-season plan with one built on real demand instead of last spring's assumptions. It is the least glamorous artifact in the planning calendar and the most consequential. The pre-season plan gets the budget meeting, the line review, and the executive sign-off. The reforecast gets a spreadsheet, a Tuesday, and whatever time is left after the current fire.

That allocation of attention is exactly backwards.

The pre-season plan expires when the season opens

A pre-season plan is a hypothesis. It is built months before the first unit sells, on carry-over hindsight, category intuition, and whatever the assortment strategy needed it to be. It is the best guess available at the time, and it is wrong in specific, unknowable ways the moment real customers start voting. Not wrong everywhere — wrong in the places that matter. A hero style pulls forward faster than planned and blows through its buy in the first four weeks. A core category runs cold and the receipts already on the water are suddenly overbought. The size curve you templated off a regional average breaks in week one, and you are stocked deep in a size the customer isn't reaching for.

None of this is a planning failure. It is the nature of fashion demand: the signal does not exist until the season opens, and once it exists it decays. The pre-season plan's job was to get you to the start line with committed inventory and a defensible open-to-buy position. The moment the season opens, its shelf life starts counting down. Every week you keep planning against it, you are steering by a map drawn before the road was built.

The teams that treat the pre-season plan as the plan — the thing they defend at buy review, reconcile against, and revisit at month-end — are managing a document. The teams that treat it as a hypothesis to be disproven, fast, are managing a business.

What a reforecast actually moves

A reforecast is not a reporting exercise. It is a set of decisions, and every one of them has a clock on it. Four things move:

• →The open-to-buy that's still open. The number that matters is not what you planned to spend — it is what you have left to spend and how fast the window to commit it is closing. A reforecast reprices the remaining OTB against actual sell-through: pull dollars off the category that's soft, push them behind the one that's proving out, while there is still receipt flow you can redirect.
• →Receipt flow and cancellations. A style tracking behind plan with units still in production is a cancellation, a push, or a chase-order conversation with the vendor — but only if you see it while the PO is still movable. A reforecast that lands after the goods ship is a markdown plan wearing a reforecast's clothes.
• →Markdown timing. Weeks-of-supply by style and channel tells you which styles are going to end the season long. The earlier you know, the cheaper the exit — a first markdown taken while the style still has full-price velocity protects more AUR than the clearance event you get forced into near the end of the selling window. A reforecast that surfaces the long styles in week five gives you a graduated markdown cadence; one that surfaces them in week twelve gives you a fire sale.
• →Size and channel re-allocation. When the curve breaks or one channel outruns another, the fix is redistribution, not another buy — pull the size that isn't selling out of the doors stocked deep, and move units to the channel that's actually proving demand.

Every one of these decisions gets cheaper the earlier it's made and more expensive — or simply impossible — the longer the reforecast waits.

The cadence problem

Here is the structural failure. Most teams reforecast monthly. The demand signal arrives weekly.

A monthly reforecast in a fashion season is a cadence mismatch you can measure in margin. By the time the month-end reforecast confirms that a category is running cold, three or four weeks of receipts have landed against a plan everyone already suspected was wrong. The buyer knew in week two. The system caught up in week five. The gap between those two dates is where the overbuy accumulates and the markdown liability quietly builds.

The reason the cadence is monthly is almost never that monthly is the right rhythm. It is that the reforecast lives in spreadsheets. Pulling actuals, reconciling them against plan across the OTB file, the assortment file, and the allocation file, re-deriving the open-to-buy, and re-checking every size and channel position is a multi-day job for a small team. You cannot run a multi-day job every week, so the cadence bends to the tooling — and the plan updates a beat slower than the business moves, every single week of the season.

That is the whole problem in one sentence: the reforecast runs at the speed of the spreadsheet, not the speed of the signal.

What connected reforecasting looks like

The alternative is not "reforecast faster by working harder." It is to remove the reconciliation step that makes the reforecast slow in the first place.

On a single, shared data model, plan and actuals are not two datasets that have to be joined — they are one object seen from two sides. Sell-through lands and the open-to-buy updates because the OTB is a view of the same data, not a downstream file that has to be refreshed. When a receipt posts, the on-order position, the projected weeks-of-supply, and the remaining OTB all move together, because they were never separate numbers. The reforecast stops being an event the team schedules and becomes a state the plan is always in.

That changes what the merchant does with the week. Instead of spending the first two days rebuilding the reforecast, the team spends the week reading a plan-vs-actual that is already current and making the decisions the reforecast exists to inform: where the open-to-buy should move, which POs to pull or cancel, which styles to mark first, where to re-allocate size and channel. The work shifts from reconciling the numbers to acting on them — which is the only part of the job that protects margin.

It also dissolves the cadence question. When the reforecast is live rather than assembled, "how often do you reforecast" stops being a resourcing decision. You reforecast continuously, because there is nothing to run — the plan is current the moment the data is. AI-assisted planning helps here too, but only on this footing: a model that flags the style pulling forward or the category going cold is useful exactly to the degree that the OTB underneath it can move in response.

Where this leaves the planning team

The pre-season plan will always be a hypothesis. No amount of tooling changes that — fashion demand is unknowable until it isn't. What a team can change is how fast it responds when the hypothesis meets the market. A reforecast run monthly in spreadsheets responds a month late. A reforecast that lives on the same data model as the plan responds as fast as the signal arrives.

This is the reason RetailNorthstar keeps plan and actuals in one model instead of syncing them across files. The point is not a tidier spreadsheet. It is that the in-season reforecast — the decision layer where merchandise planning actually earns its keep — runs at the speed of sell-through instead of the speed of reconciliation. The pre-season plan was the team's best hypothesis about the season; the reforecast is the team's judgment applied to what the season actually did. Everything that makes a planner valuable — knowing which soft category to give another week, which hero to chase, which size break is a data artifact versus a real signal — lives in the reforecast, not the pre-season plan. For a mid-market or emerging apparel brand, live in weeks, that is the difference between catching the overbuy while the PO is still open and reading about it in the markdown report.

See how connected open-to-buy keeps plan and actuals in one model.