Real-time sales data and social listening were expected to make trend forecasting obsolete. The firms remain in business because the decisions they inform are made before any data exists.
Manufacturing commits early
Yarn is spun, dyed and woven months before a garment is cut, and fibre is contracted before that. Each stage requires a decision on colour and construction.
No sales signal exists at that point, because the product does not exist and neither does the season it will sell into.
Somebody has to make a call anyway, and a forecast is a structured way of making it rather than a guess.
The industry coordinates through shared direction
Mills, trim suppliers, brands and retailers all need to be working toward compatible outputs, or the components will not match when they meet.
Colour direction distributed across the supply chain lets independent companies converge without negotiating with each other.
The forecast is partly a prediction and partly a coordination mechanism, and the second function works regardless of whether the first is accurate.
Real-time data only describes the present
Engagement and sales data are excellent at showing what is selling now, which is exactly what fast fulfilment operations need.
They say nothing about a season eighteen months out, and extrapolating current momentum forward has an obvious failure mode.
Firms that operate entirely on live data still buy long-lead materials, and for those decisions they use the same forecasts as everyone else.
Fast fashion is often described as having escaped the calendar, but it has only escaped the final stage of it, since the fabric it prints on was committed to months earlier.
Forecasts are self-fulfilling to a degree
When most of an industry receives similar direction, shops fill with similar products, and shoppers choose from what is available.
Adoption then appears to confirm the forecast, which strengthens the case for buying the next one.
This circularity is well understood inside the industry and is one reason the output is framed as direction rather than as prediction.
The value has shifted toward the unglamorous end
Silhouette forecasting has weakened as trends fragment and emerge from too many places to track.
Colour, fibre, yarn and finish forecasting has held up better, because those decisions have the longest lead times and the fewest alternatives.
The business has consequently moved toward the technical end of the supply chain, which is less visible than the runway commentary the firms were known for.