Demand forecasting for consumer durable goods

Demand forecasting for consumer durable goods separates replacement purchases, new buyers and channel inventory changes before comparing a forecast with actual orders. That separation matters for appliances, furniture and other products purchased infrequently. A strong month of manufacturer orders can reflect retailer restocking even when consumer purchases have barely changed.

Define the demand you are measuring

Start with one measure: net customer orders in units, assigned to the month the order was accepted. Keep shipments, consumer sales and channel inventory in separate columns. The Census Bureau's manufacturing definitions distinguish new orders, shipments, unfilled orders and inventory; these measures answer different questions.

For a manufacturer, the practical question is whether each channel will order the supplied quantity. For a retailer, it may be whether consumers will buy it. Use the same definition in the forecast and actual columns. An order forecast evaluated against shipments mixes demand with fulfillment timing.

This is a durable-product application. Consumer-goods articles cover nearby commercial measures, while industrial-goods demand forecasting separates repeat purchasing from customer projects.

Gather the export and the assumptions

Bring a forecast snapshot and an order export containing product, channel, forecast month, forecast date, units ordered and cancellations. Add opening channel inventory, target closing inventory, and the demand assumptions supplied by the commercial team.

For replacement demand, record the installed population the estimate covers, its date, the assumed replacement rate and the evidence behind it. National ownership data can provide context: the EIA's Residential Energy Consumption Survey publishes appliance characteristics by housing type and other household attributes. Those tables do not establish your brand's installed base or replacement rate.

The sales or category owner should therefore label replacement assumptions explicitly. Warranty registrations, service records and distributor reports may cover different populations. An undocumented total cannot become a precise demand estimate simply because it appears in a workbook.

Build a replacement and inventory bridge

For a simple scenario, use:

Replacement purchases = eligible installed units × assumed period replacement rate

Implied channel receipts = replacement purchases + new-buyer purchases + target closing stock − opening stock

The second identity assumes no returns, losses or interchannel transfers. Add those movements separately if they exist. The replacement rate must cover the forecast period; an annual assumption cannot be applied unchanged to a monthly forecast.

A fictional US refrigerator manufacturer prepares a quarterly scenario. Its channel pools below are mutually exclusive planning populations, not estimates of the entire US market. All quantities are units; replacement rates are supplied assumptions.

Put the table in columns A:H, with the first data row at row 2. Compute replacements in D2 and receipts in H2, then fill down:

=B2*C2
=D2+E2+G2-F2

Retail dealer receipts are 720 + 180 + 350 − 300 = 950. Builder receipts are 400 + 250 + 180 − 200 = 630. Replacement rates stay at the channel level; adding percentages would have no useful interpretation.

The stock identity proves the total: 500 opening units + 1,580 receipts − 1,550 consumer purchases = 530 closing units. Consumer purchases are 1,120 replacements + 430 new-buyer purchases. Receipts exceed purchases because the channel increases stock.

Compare with the supplied forecast and actual orders

Keep the original forecast unchanged. This example assumes all planned receipts are ordered and received within the quarter, with no change in outstanding channel orders, so the receipt scenario also represents net orders. Otherwise, bridge receipts to orders using the change in outstanding orders before comparing them. Once the quarter closes, join actual net orders to the same product-channel keys.

With this table in A:G, calculate the signed and absolute differences:

=B2-D2
=C2-D2
=ABS(F2)
=SUM(G2:G3)/SUM(D2:D3)

The scenario's absolute error is 60 / 1,580 = 3.8%, rounded to one decimal place. The supplied forecast's absolute misses are 60 and 10, giving 70 / 1,580 = 4.4%. Although the scenario matches the total exactly, its channel misses remain. The SKU and site bias guide explains how to investigate repeated directional errors.

Where the comparison goes wrong

• Replacement assumptions become facts. A documented planning rate remains uncertain. Review it against observed purchases and available service evidence.
• Channel stock is counted twice. Consumer purchases and retailer receipts are separate flows. Use the stock bridge rather than adding both to demand.
• Installed populations overlap. The same appliance cannot support replacement demand in both channel pools unless an explicit allocation removes duplication.
• Forecast dates move. Compare the snapshot available when purchasing decisions were made, rather than a revised forecast created after orders arrived.

Use forecast accuracy guidance to keep the comparison horizon consistent. A national indicator or a replacement assumption can challenge the forecast, but neither replaces the company's actual order reconciliation.

Compare your own durable-goods file in Covirage

Upload a file or Google Sheet with supplied forecasts, actual orders and documented replacement assumptions. Covirage maps the columns, and its tools compute and check the comparisons; the external AI model explains the results with every figure traced to its rows. Create a free Covirage account, which takes about a minute, then set up your business and bring a sample of your own data.

See analytics for industrial manufacturers for the supporting industry context.