Supply Chain · Retail

Fashion without a crystal ball

Mohit Saini · 18 May 2022 · 4 min read

Fashion without a crystal ball

Twice a year, many fashion and textile companies run the same ritual. Dealers and wholesalers are invited to see the new collection. They study hundreds of designs and place their bookings for the next six months. The company then produces against these bookings.

It sounds sensible: the people closest to the market tell the company what they will sell. So let us test how well it works.

How accurate are the people who know the market best?

Take a dealer’s regular items, the ones sold year-round, not new fashion. Compare what he booked for the season with what he actually sold.

In most cases, the gap is huge. Some shades and qualities sell twice what was booked; others sell half. This happens with experienced dealers and new ones, with big operations and small ones, with and without software.

If forecasting is this hard for regular items, imagine the error on brand-new designs.

Bad tools, or impossible task?

There are two possible explanations:

  1. Dealers forecast badly because they rely on gut feel instead of scientific tools.
  2. Forecasting demand for a fashion item, in one dealer’s market, six months ahead, is inherently unreliable, whatever the tool.

Software vendors and many consultants favour the first explanation, because it sells better forecasting tools. Theory of Constraints starts from the second.

Why? Because demand for fashion behaves like a complex system. A competitor’s delayed shipment, a celebrity seen in a particular style, word spreading that something is selling well: small events can create sudden surges, and those surges feed on themselves. Mathematical models, however sophisticated, can’t reliably predict such systems. Using equations doesn’t make a forecast scientific.

Aggregating forecasts (a national forecast instead of a dealer-level one) does reduce error. But the benefit is lost as soon as stock is pushed down to the shop, where the error is largest.

Don’t predict faster. React faster.

If we can’t predict, we must react. The faster the supply chain responds to actual sales, the less it suffers from forecast errors.

For regular items, this means make-to-availability:

  • Hold stock in a central warehouse.
  • Cut production lead time so the warehouse can be refilled quickly.
  • Replenish dealers several times a month according to what they actually sold, instead of once per season.
  • Let dynamic buffer management adjust each dealer’s target: items that sell fast get a higher target within a cycle or two; items that don’t sell are run down within a few cycles.

With replenishment time down from a whole season to a week or two, dealers hold far less stock, and the stock they hold is what sells.

Picture one dealer and one regular shade. He books 600 metres for the six-month season, about 100 a month. If it sells 200 a month, he runs out in the third month and loses sales for the rest of the season. If it sells 50 a month, he ends the season with 300 metres of unsold cloth. Either way, on average he holds around three months of stock.

With refills every two weeks against what he actually sold, he holds perhaps three to four weeks of that shade. When it sells 200 a month, his refills rise within a cycle or two. When it sells 50, they shrink. He carries about a third of the stock, and both the stock-out and the leftover largely disappear.

New products without new development

There is a bonus. Under the booking system, dealers and retailers turn their stock only once or twice a year. Shelves fill up with slow items left over from past seasons, and there is little room or money for anything new. Ironically, even though the company launches a large range every season, shops rarely look very different.

Under make-to-availability, dealers earn a much better return on their money and need only a few weeks of stock for any item. They can afford to try more of the company’s existing range, knowing that winners will be refilled quickly and losers will be run down without much risk. For the customer walking into the shop, items new to that shop are effectively new products, without the company having to develop anything.

What about genuinely new designs?

New designs still need forecasts in a booking system, and that is where the worst errors happen. Two changes help:

  • More, smaller collections. Instead of two big seasons with hundreds of designs presented at once, offer several smaller collections through the year. Dealers forecast over a shorter horizon, so errors are smaller. A smaller collection gets more attention, so each design gets a fairer chance and larger quantities.
  • Reliable supply. Dealers usually wait until the old season’s supplies are complete before starting the next, which creates a blind spot at the start of each season. When supplies are reliable, dealers can order the next collection without waiting.

New designs are then made to confirmed orders. Those that keep selling move into make-to-availability. Those that don’t are dropped by the buffer rules. No crystal ball needed at any stage.

The secret of a fashion supply chain is not an accurate forecast. It is a fast reaction.

Mohit Saini

Strategy, supply chain & sales transformation consultant. About · Book a discussion