Say you nail the forecast. The model is purpose-built, probabilistic, tuned on your sell-through at SKU by size by channel. It is genuinely excellent. You are still not done, and here is the uncomfortable part: an excellent forecast is not a plan, and treating the two as the same thing is one of the most expensive mistakes in retail.
A forecast is a prediction of what will sell. A plan is a commitment about what the business will do, and those are not the same document. The forecast answers 'what does demand look like?' The plan has to answer 'what are we going to buy, spend, and promise finance, given that demand and everything else we know?' The gap between the two is where two different truths live, and if you do not reconcile them on purpose, they reconcile themselves against you.
The two truths are bottom-up and top-down. One is the SKU-level reality of what customers will buy. The other is the financial reality of what the business has to hit. A real plan holds both at once. Most brands run them in separate spreadsheets, owned by separate people, and only discover they disagree when it is too late to do anything about it cheaply.
Bottom-up is the truth of what will sell
Every SKU, every size, every channel, added up from demand. Precise, honest, and blind to what the business needs.
Bottom-up starts at the smallest unit that matters and builds. It is the forecast for this style, in this size, for this channel, on this lead time, summed into options, categories and totals. This is the honest number. It reflects what customers actually do, size curve and all, and when it is built on a real demand model it is the most accurate picture of the season you can get.
But bottom-up has a blind spot the size of the business. It does not know your revenue target. It does not know the margin the board committed to, the working-capital ceiling the CFO set, or that marketing is holding spend for a launch in month two. It answers 'what will sell', beautifully, and stays silent on 'what do we need to hit'. Run on bottom-up alone and you can build a perfectly accurate plan that quietly misses every financial goal the company has, and you will not find out until the numbers roll up.
Picture the roll-up. Every category forecast is defensible on its own, built from real demand, and the sum lands eleven points under the revenue line finance already gave the street. Now what? You cannot invent demand that is not there, so someone starts padding buys to close a financial gap with units the market never asked for. That is bottom-up failing not because the forecast was wrong, but because it was never asked the second question. Accuracy at the SKU level and a miss at the P&L level are not a contradiction. They are what you get when the only truth in the room is bottom-up.
Top-down is the truth of what you must hit
The revenue line, the margin target, the open-to-buy the CFO signed. Real, non-negotiable, and totally blind to the size curve.
Top-down comes from the other direction. It starts with the number the business has to make: the revenue plan, the margin target, the merchandise financial plan, the open-to-buy by category. This is also the truth, a different one. It is what finance is measured on and what the board approved. It is not a suggestion.
And it has the mirror-image blind spot. Top-down does not know your size curve. It does not know that half the growth it just allocated to a category will land in a size that already stocks out, or a door that skews the other way. It hands down a spend number that is financially correct and operationally naive. Run on top-down alone and you buy to a budget instead of to demand, which is exactly how you end up with the right total dollars in the wrong units. The category number is green. The markdown report is red.
Watch how that plays out. Finance grows dresses eight percent, so eight percent more dollars flow into dresses. But the growth in real demand was in two styles and one size band, and the budget cannot see that, so the extra dollars spread evenly across the range like butter. The styles that were already pulling stock out anyway. The styles that were dead get bought deeper. You spent the exact number finance approved and manufactured overstock and stockouts in the same category, at the same time. The budget was hit to the dollar. The units were wrong in every direction that costs money.
of working capital freed on average when spend is planned against a live, reconciled number instead of a top-down budget and a bottom-up forecast that quietly disagree.
That capital is not freed by spending less. It is freed by spending against one number that both truths agree on, so money stops getting stranded in the gap between the budget finance defended and the demand the merchants actually saw. The 22 percent is the cost of the disagreement, handed back.
The plan is the reconciliation, and it has to be live
One number finance and merchants both plan from. Not two spreadsheets that drift apart by week three.
So the plan is not the forecast, and it is not the budget. The plan is the reconciliation of the two, the single number that respects both what will sell and what the business must hit. And the word that matters most is live.
Reconcile once, in a room, in January, and you have not solved the problem. You have scheduled it. The bottom-up number keeps updating as real sell-through arrives. The top-down number keeps updating as the business re-forecasts revenue and adjusts the open-to-buy. If the reconciliation is a static spreadsheet, the two truths start drifting apart the moment the meeting ends, and by week three they describe two different businesses. You pay for that gap twice: in the inventory you over-bought to a stale budget, and in the markdown you take to clear it.
And the drift is not slow. One promo, one viral style, one soft week in a key door, and the bottom-up number has moved while the budget sits frozen where you left it. A plan that only reconciles at the start of the season is at its most accurate on day one and decays from there. The whole point of holding both truths is lost if you only hold them together once.
Reconciled plans clear more before markdown
The 14-point gap is not merchandising instinct. It is what happens when finance and merchants plan from one live number instead of two that drift.
Live reconciliation is what closes it. Bottom-up and top-down sit on the same model, so the moment sell-through moves the bottom-up number, the plan re-plots against the financial guardrails automatically, inside the ceilings finance set. Working-capital cap, margin floor, category open-to-buy: these become limits the system plans within, not rules a buyer has to remember and finance has to police after the fact. When the two truths disagree, the disagreement surfaces the same week, with the reasoning attached, instead of at quarter close as a surprise.
A forecast tells you what will sell. A budget tells you what you must hit. The plan is the one number that respects both, and it is only worth anything if it moves the day either one changes.
This is the job an AI-native planning stack should actually be doing, and it is a harder, more valuable job than producing a forecast. The purpose-built model gives you the bottom-up truth. The financial plan gives you the top-down truth. The platform reconciles them into a single number that finance and merchants both plan from, and keeps them reconciled as the season moves. Tightly runs this live, with the agent staging the moves when the two drift and putting them in front of a human to approve, so the reconciliation holds all season instead of dying in week three.
So build the best forecast you can. Then remember it is an input, not the answer. The answer is the plan, and the plan is the reconciliation, held live, between what will sell and what you have to hit.