How is AI used for capacity booking and production planning in fashion?
AI can improve capacity forecasts and scenario planning, but booking quality still depends on order stability, honest lead times and supplier relationships.
KEY TAKEAWAYS Summary by the editors
- Capacity booking reserves factory or line capacity ahead of orders, and AI helps by forecasting demand by category and testing scenarios, not by guaranteeing supplier availability.
- The Better Buying Purchasing Practices Index 2025 reported an overall softgoods score of 66, with Planning and Forecasting showing the largest decline, down 3 points.
- In that survey, 37% of suppliers named planning and forecasting as their top area for improvement, which suggests that forecast quality and buyer behaviour are bigger constraints than software.
- Better Buying's framework measures how closely planned production matches the orders actually placed, so a booking model should be judged against realised orders, not only against the forecast.
- McKinsey's State of Fashion 2026 describes agility as the defining factor for brands and suppliers, which makes scenario-based capacity planning more relevant than a single annual plan.
What is capacity booking in fashion production?
Capacity booking is the practice of reserving production capacity, such as sewing lines, cutting time or fabric mill slots, before a style is finally ordered. AI is used to forecast how much capacity a brand will need by category, factory and time window, and to simulate what happens if volumes shift. It supports the decision but does not create capacity that the supplier does not have.
This explainer covers how the process works, where machine learning adds value, what data it needs and where it fails.
Most brands set a seasonal buy plan, translate it into volumes by product type and fabric, and share a forecast with suppliers. Suppliers confirm capacity, often as lines or units per week, and the brand then places detailed orders as designs and fabrics are approved. The gap between the early forecast and the final order is where problems arise.
Better Buying measures this gap directly. Its Planning and Forecasting category looks at how much visibility buyers give suppliers into ordering plans and how closely planned production matches the orders actually placed. A separate category, Management of the Purchasing Process, assesses whether suppliers receive production lead time and whether it is maintained through development and launch.
Where can AI help with capacity planning?
| Use | How AI contributes | Main data needed | Typical limit |
|---|---|---|---|
| Volume forecasting by category | Machine learning on sales and pre-order data | Sales history, calendar, promotions | New styles have little history |
| Capacity needs by factory | Convert forecast to minutes or lines using style complexity | Standard minute values, line efficiency | Operations vary by site |
| Scenario testing | Simulate demand swings and delays | Forecast ranges, supplier constraints | Quality depends on assumptions |
| Allocation suggestions | Propose splits across suppliers | Cost, quality, risk and lead time data | Cannot see supplier priorities |
| Early warning | Flag when orders drift from the booked plan | Booked versus placed volumes | Needs timely order data |
The value is mostly in speed and range. Planners can compare several scenarios quickly and see the capacity implications of each, rather than rebuilding spreadsheets for every change.
What data does AI need for production planning?
- Style-level attributes such as product type, construction complexity and fabric, so that demand can be converted into workload.
- Supplier data covering sites, process capabilities, historical output and on-time delivery.
- Booked and placed volumes over several seasons, to measure forecast accuracy and order drift.
- Calendar effects including holidays and factory closures in producing regions.
- Lead time history by supplier and product type, preferably including the reasons for delays.
Complexity data is often the weakest link. If every style is treated as equal workload, the converted capacity figure will mislead planners and suppliers alike.
Why do capacity bookings fail even with good forecasts?
Forecast accuracy is only one cause. The Better Buying Purchasing Practices Index 2025, based on 1,340 completed supplier surveys, gave the softgoods industry an overall score of 66, down 1 point from 2024. Planning and Forecasting fell most, by 3 points, and 37% of suppliers named it their top area for improvement. The release also noted that the 2025 cycle ran as new United States tariffs were imposed, a reminder that external shocks change orders after capacity has been booked.
Order changes, late approvals and shifting volumes all reduce the value of a good forecast. An AI model that predicts demand well still cannot fix a buying process that releases orders late.
How should a brand introduce AI into capacity planning?
- Start by measuring booked versus placed volumes by supplier for the last few seasons.
- Add style complexity and standard time data so that demand converts into workload.
- Run the model beside the current process for a season and compare the results.
- Share forecast ranges and confidence with suppliers, instead of single figures.
- Review with suppliers where the plan changed and why, then adjust process rules.
McKinsey's State of Fashion 2026 states that agility will be the defining factor for brands and suppliers, and that larger suppliers are investing in digitisation while smaller players face growing pressure. Capacity tools that depend on detailed data from suppliers should therefore account for the fact that not every factory can supply it.
How do transport delays affect booked capacity?
Capacity booked at the factory is only useful if goods arrive when the sales plan needs them. UNCTAD's Review of Maritime Transport 2025 reports that the average voyage haul rose from 4,831 miles in 2018 to 5,245 miles in 2024, and that ships which once crossed the Red Sea in days now sail for weeks around the Cape of Good Hope. It also describes freight rate volatility as becoming the norm.
For planners this means that a production slot and a ship date are linked decisions. Models that treat transit time as a fixed number will understate the risk of late arrival, so lead time assumptions should carry ranges and be updated as conditions change.
- Which forecast version is shown, and how often will it change?
- What volume range can the line absorb without overtime or subcontracting?
- Which styles are the most complex, and have standard times been confirmed?
- What notice does the supplier need for a change in size split or colour?
- How are late fabric or trim approvals treated in the booked slot?
Asking these questions turns a model output into a shared plan. It also gives planners an early indication of where the model's assumptions differ from factory reality.
What are the risks of AI in capacity booking?
The main risks are false precision, since a model output can look more certain than the data supports, and the temptation to shift volatility onto suppliers because the tool makes it easy to re-plan. A model that improves the brand's internal view while worsening supplier stability may reduce delivery reliability in the longer run. Human review, supplier dialogue and clear change rules remain part of the process.
Frequently asked questions
What is capacity booking in garment manufacturing?
It is the reservation of factory capacity, such as sewing lines or weekly units, ahead of final orders. Brands share forecasts, suppliers confirm capacity, and orders are then placed against that reservation as styles are approved.
Can AI guarantee that a factory has enough capacity?
No. AI can estimate the capacity a plan requires and test scenarios, but actual availability depends on the supplier, its other customers and conditions at the site. Confirmation still comes from the supplier.
What data is most important for AI production planning?
Booked versus placed volumes, style complexity or standard times, supplier output history and lead time history are most important. Without complexity data, demand cannot be converted reliably into workload.
Why do suppliers rate planning and forecasting poorly?
Better Buying's 2025 index recorded the largest score decline in Planning and Forecasting, and 37% of suppliers named it their top improvement area. The category reflects visibility into ordering plans and how well planned production matches orders.
One edition every weekday morning. Read in five minutes. Free for industry professionals.
