INVInventoryStop 11 of 11
Postponement
Holding generic stock and making the final version only when real demand is known.
Read atChange any time
A paint shop doesn’t stock 2,000 ready-mixed colours. It stocks a few white bases and a tinting machine, and mixes the exact shade when you ask for it. That is postponement: keep products generic for as long as possible and make the final, specific version only once real demand is known.
Why it works: forecasting total paint demand is far easier than forecasting demand for each of 2,000 shades. The generic base pools the uncertainty of all the colours, so you need much less stock overall, and you rarely end up with cans of a colour nobody wants.
The classic business case is Hewlett-Packard’s printers. HP redesigned them so the country-specific parts, such as the power supply and the manuals, were added at regional warehouses instead of the factory. Regional stock stayed generic until orders arrived, which cut inventory and improved availability.
Postponement comes in three forms:
- Form: delay labelling, packaging, assembly or mixing (paint, printers, multi-language packs).
- Time: make or ship only after the order arrives.
- Place: hold stock centrally and send it out only when a region needs it.
It isn’t free. The final step close to the market can cost more per unit, products may need redesigning, and customers may wait a little for the last step.
Words to use
- Delayed differentiation
- Another name for postponement: products split into variants as late as possible.
- Decoupling point
- Where stock is held and work switches from forecast-driven to order-driven.
- Form postponement
- Delaying the physical customization: labelling, packaging, assembly or mixing.
The arithmetic of staying generic
Suppose a printer comes in four country versions, each with weekly demand that varies independently with = 300 units, and the target is = 1.65.
If each version is finished at the factory, each needs its own buffer. If they share a generic printer, the generic stock faces the combined demand, whose variability is only , not 4 × 300 = 1,200. It is the same square-root effect as pooling warehouses, applied to variants instead of locations.
Shipping from the factory takes 4 weeks and final customization takes 1 week.
| Set-up | Buffers needed | Safety stock |
|---|---|---|
| Customize at the factory | Each version covers 5 weeks: | 4,427 |
| Postpone; customers wait 1 week | Generic stock covers 4 weeks: | 1,980 |
| Postpone; customers need stock now | Generic 1,980, plus each version covers 1 week: | 3,960 |
Postponement cuts safety stock by 55% if customers accept a one-week wait, but only by about 11% if finished stock must still sit on the shelf. The benefit depends on whether the last step can happen after the order arrives.
The decoupling point
The customer order decoupling point is where a product stops being made to forecast and starts being made to a specific order. Upstream of it, stock is generic and planned from forecasts; downstream, work is driven by real orders.
| Strategy | Decoupling point | Example |
|---|---|---|
| Make to stock | Finished goods | Shampoo, packaged food |
| Postponed or assemble to order | Generic base or modules | Paint tinting, configured laptops |
| Make to order | Raw materials | Custom furniture |
| Engineer to order | Design | Industrial machinery |
Postponement moves the decoupling point as late as the process and the customer’s patience allow. The customer’s acceptable waiting time is the limit: every step after the decoupling point must fit inside it.
Postponement starts on the drawing board
Postponement usually needs the product or the process to change:
- Modular design: country- or customer-specific parts become separate modules added at the end (power supplies, plugs, manuals).
- Standardization: one common part replaces several, so variants share more.
- Process resequencing: reorder the steps so the differentiating one comes last. Benetton famously knitted sweaters in undyed wool and dyed them later, instead of knitting dyed yarn, so colours could follow what was actually selling.
What it costs
- Higher unit cost at the end: customizing in a regional warehouse rarely has the scale or automation of the factory.
- Investment: tinting machines, local assembly, trained staff and quality checks in many places.
- Redesign cost, and sometimes a compromise in the product: a universal power supply may cost more than a country-specific one.
- A longer wait for customers if the final step happens after the order.
- Tax and duty effects: moving value-adding steps into another country can change customs duties, which may help or hurt.
When it backfires
- Variants are few, or their demands move together, so pooling saves little.
- Customers want instant availability and the final step is slow, so finished stock is still needed.
- The last step is expensive or hard to do well outside the factory.
- Volumes per variant are high and steady, so forecasting each one is already easy.
Related ideas
- Logistics postponement: keep finished stock centrally and ship only when a region orders, trading inventory for transport speed.
- Speculation: the opposite strategy, committing early to cut cost (long factory runs, cheap sea freight). Bucklin (1965) framed channel design as a balance between postponement and speculation.
- Services: generic training followed by specialization, or standard software configured per customer, follow the same logic.
Interview kit
Practise it out loud
Answer each question yourself before opening the model answer.
Postponement means keeping products generic as long as possible and customizing only when demand is known, like paint tinted at the store or HP adding country-specific parts at regional warehouses. It pools forecast uncertainty across variants, so you hold less stock and avoid obsolete variants. The trade-offs are a costlier final step, product redesign and sometimes a little extra lead time.
Why does postponement reduce inventory?
Because aggregate demand for the generic product is more predictable than demand for each variant. Holding stock before differentiation pools the uncertainty, which is the same risk-pooling effect as centralizing warehouses, applied across product variants.
When would postponement be a bad idea?
When customization is expensive or slow near the market, when customers won’t wait for the final step, or when there are few variants with stable demand and little uncertainty to pool. It shines with many variants and unpredictable mix.
Connected stops
Where this idea feeds in, or gets its inputs from.
Sources and further reading
- Feitzinger, E. & Lee, H. L. (1997). Mass customization at Hewlett-Packard: The power of postponement. Harvard Business Review, 75(1), 116–121.
- Chopra, S. & Meindl, P. Supply Chain Management: Strategy, Planning, and Operation. Pearson. Section on postponement and aggregation.
- Bucklin, L. P. (1965). Postponement, speculation and the structure of distribution channels. Journal of Marketing Research, 2(1), 26–31.