Buying discipline: the money is lost before purchase

By the time a pile of stock is old enough to worry about, the decision that created it is six to nine months in the past. Somebody approved a quantity. Somebody accepted a minimum order quantity without pushing back. Somebody launched a line with no agreed definition of failure.

Everything after that is damage control. The money is lost at the purchase order.

Minimum order quantity is a negotiation

Suppliers quote a minimum order quantity because it suits their production planning. It is a starting position, not a law of physics, and Indian manufacturers negotiate on it more often than founders assume.

The levers are usually these. Pay a higher per unit price for a smaller first run and treat the difference as the cost of a test. Commit to a total volume for price protection but take delivery in tranches against a written schedule. Separate one time costs such as tooling, plates or dies and pay them upfront so the unit quote stops carrying them. Accept a longer lead time for a smaller batch, often the cheapest concession available if you plan ahead.

Packaging is where minimum quantities usually bite hardest. Cartons, labels and pouches carry their own minimums, and they frequently force the product quantity higher than the product itself required. Where the category permits, design a shared base pack with variant specific labels, so the large minimum sits on the component that fits every line rather than on the one that might fail.

The number to compare is landed cost per unit sold, not landed cost per unit bought. A price at ten thousand units looks excellent right up to the point where three thousand of them are still sitting eleven months later. Run both versions before signing.

Forecast error concentrates in the long tail

Your top sellers have history, volume and seasonality you have already observed. Forecast error on them exists but it is proportionally small and it is recoverable, because a shortfall on a fast mover is a stockout you can fix with a replenishment order.

A new tail SKU has none of that. No history, low volume, and a demand signal so thin that two weeks of ordinary noise reads like a trend in both directions. Percentage error on those lines is large by construction, and no forecasting method fixes it, because the information does not exist yet.

That asymmetry gives a clean buying rule. Buy the head deep and the tail shallow. Being wrong on the head is inconvenient and correctable. Being wrong on the tail creates the pile that eventually needs marking down.

Remember also that a tail SKU costs more than its purchase order. It needs listing work, photography, catalogue maintenance across every platform, and a share of somebody’s weekly attention. Those costs land whether the line sells or not.

How many new lines can you actually afford to buy

The useful question is not how many SKUs the brand should carry. It is how many first buys the team can fund, forecast, list properly and review honestly in a single quarter.

The binding constraints are working capital, warehouse and pick face capacity, the dark store or shelf facings you can realistically win at your current velocity, and the attention of the one person who will actually own the review. Attention gets ignored, and it is usually the tightest of the four.

A practical filter before any purchase order is raised. Every new line needs a named cash budget and a named review owner. If either is missing, the line is not ready to be bought, regardless of how good the sample looks.

Pre-orders and staggered buys for untested lines

An untested line does not need a full buy. It needs a demand read.

Pre-orders on your own site are the cheapest test available to an Indian D2C brand, because the audience is already yours, the cash arrives before the stock does, and the signal comes from people committing money rather than clicking a survey. Be honest about dispatch dates and keep the window short.

A staggered buy gives you price and optionality together. Commit the total volume so the supplier holds the rate, take delivery in tranches, and negotiate a deferral or cancellation clause on the later tranches. Get the clause written into the purchase order itself, because a verbal understanding with a factory is worth nothing when the goods are ready.

A geographic pilot is the third route. Launch in one city or one platform before national listing. Quick commerce makes this straightforward, since dark store lists are managed city by city, and a small footprint gives a cleaner read because availability is easier to control.

Write the exit criteria before the purchase order

This is the highest value ten minutes in the entire process, and almost nobody spends them.

Before the purchase order is approved, write down the expected rate of sale in units per week by week six, the minimum acceptable rate below which the line is considered a miss, and exactly what happens if the miss occurs. The options are limited and should be named in advance. Fire the markdown trigger. Move it to a different channel. Hold and retest once with corrected content or a defined spend. Or stop the repeat buy.

The most valuable sentence to put in writing is that the repeat buy is not automatic. The default is no repeat unless the criteria are met. Reversing that default removes the single most common way a modest mistake becomes a large one, which is reordering a slow line because the reorder point said so.

Record all of it in the purchase order approval note. That is what stops the criteria being re-argued in month four by the person who chose the product.

Week six, not month six

Most brands discover a slow mover when someone notices the stock. That is usually month five or six, by which point the repeat buy has often already been placed.

Put new lines on a fixed review calendar instead. Three checkpoints.

• Week two is an execution check only. Is the listing live, is stock actually available, does it appear in search, is the price correct across channels, are images and content complete. Do not judge demand here, because most week two failures are availability failures wearing a demand costume.
• Week six is the first honest demand read, against the rate of sale written at launch. Execution issues should be closed by now, so a miss at week six is a real miss.
• Week twelve is the repeat buy decision, taken against the criteria set before the first order.

Keep the attendance small and fixed. Buying, planning, the channel owner and finance. The output is a recorded decision with an owner and a date, not a discussion. A line that gets three honest reviews in its first quarter almost never becomes a write-down problem, because the miss is caught while the only cost is the first buy.