A forecast is not a number. It's an argument.

Somewhere this week, a planning meeting will open with a number on a slide. March: ₹47 crore. Heads will nod. And if anyone in the room asks the only question that matters, where did that come from, the honest answer at most retailers is some version of: last year, times a growth number we felt good about. The number isn't wrong, exactly. It's unexamined. It can't be argued with, because there is nothing inside it to argue about. That is the real difference between a forecast and a plan, and it is the whole subject of merchandise financial planning.

This piece walks one division of a sample dataset, menswear, from raw actuals to a March plan that reconciles line by line. The numbers are illustrative; the mechanism is the point. It is the same walk we run on live data, and every step of it is visible.

Where the baseline comes from (and why it isn't last year × 10%)

Start with what actually happened. January sold ₹54.85 crore; February sold ₹35.08 crore. A 36% month-on-month drop, which looks alarming until you know the season, and the season is exactly the kind of thing a naive trend line doesn't know. Project that slope forward and March collapses. Project last year plus a hopeful ten percent and you've ignored this year entirely. Both are guesses wearing spreadsheets.

A better baseline asks a sharper question: not how much did we sell, but how far did sales run ahead of what the buy was planned against. In January, sales beat the division's budgeted cost of goods by 41%. In February, by 31%. Weight the recent month heavier, because it carries the fresher signal, and the blend lands at 34.4%. Grow February by that, and the suggested March is ₹47.13 crore.

That distinction matters more than it looks. A revenue trend rewards whatever just happened. A margin-versus-plan signal rewards a division that is outperforming its own buy, which is the thing you actually want to fund with more inventory. It is the same logic that makes sell-through a better grade than units sold: measure against the plan, not against the applause.

What one number still can't see

So far, so good, and still not a plan. ₹47.13 crore is one number for an entire division, and a division is not one thing. It is flagships and small-format stores, zones with a March festival and zones without one, a price revision landing mid-month, clearance racks doing their end-of-season work, an early summer pulling short sleeves forward, an online channel the store history never counted, and a new store opening on March 10 that today's data literally cannot see, because it doesn't exist yet.

Treat all of that as one average and the average will be wrong everywhere at once: over-bought where demand is flat, under-bought where the calendar spikes. The fix is not a smarter single number. It is taking the baseline apart, one named factor at a time, in an order that ties out.

Nine factors, one running total

Read the stack top to bottom and notice there are only three kinds of line.

• Correctors reshape the baseline. Last year's February-to-March index says the season usually lifts about 28%; blended with today's trend (one year's index is itself a small sample, so blend, never replace), the estimate settles at ₹46.08 crore.
• Splitters move money without changing the total. Store grade and format divide the division number into fair slices: 45% to Grade A flagships, 35% to Grade B, 20% to small-format Grade C, with a ramped share for stores still warming up. Vendor lead times and minimum order quantities do the same at the end: same total, now dated and orderable.
• Movers each carry a sign and a size. A South-zone festival adds ₹0.92 crore to that zone alone. A planned 2% price revision adds ₹0.45. Clearance takes ₹0.60 away, because markdown units sell at a lower realized value. An early summer adds ₹0.30, the online channel ₹0.50, and the Salem store opening March 10 contributes ₹0.36 at a 40% first-month ramp.

The total lands at roughly ₹48.0 crore, about two percent above the baseline. If that sounds anticlimactic, good. The point was never that the number moves dramatically. The point is that every rupee of movement now has a name, a formula and an owner. The festival line can be challenged by whoever owns the calendar. The markdown line can be switched off if the clearance plan changes. Nothing hides inside a multiplier.

There is also a quieter benefit. When the plan is a stack instead of a slide, the meeting changes shape. Nobody argues about whether ₹48 crore "feels right," which is a conversation with no exit. They argue about whether the festival uplift should be 8% or 6%, whether the new store really ramps at 40%, whether the price revision holds. Those are arguments the room can actually settle, and each one settled makes the plan better.

Where the plan goes next

A reconciled plan is the middle of the story, not the end. Two things happen to it from here. First, once enough history is connected, the same factors become features, and a model layer learns the interactions the rules only approximate, festival × region × store grade, and prices the uncertainty around the number. Second, and more urgently, the sales plan has to become a buying decision: what you may actually spend once the stock you already own is counted. That number, Open-to-Buy, is the true output of merchandise financial planning, and it is almost always smaller than anyone in the room expects.

But it starts here, with a baseline you can defend and a stack you can argue with. The retailers who plan well are not the ones with the cleverest multiplier. They are the ones whose March number can survive a room full of people paid to disagree with it.