Demand planning vs supply planning for D2C buying

Demand planning projects, supply planning commits

ASCM, the Association for Supply Chain Management, treats demand planning and supply planning as separate processes. On that topic page, demand planning equips supply chain organisations to project future demand within a reasonable margin of error. If the demand plan indicates high product volumes, ASCM's supply planning page says the supply plan must detail how that product will be manufactured and supplied. A demand plan built from data comes before that work starts.

For a brand that buys finished goods, the released purchase order is the commitment in this comparison: a quantity, a supplier date and cash. Treating that order as the commitment is an interpretation of the manufacturing role boundary.

The boundary is on ASCM's demand planning careers page. There, the demand planner does not release purchase orders, does not sequence the plant, and in most organisations does not own the master production schedule. The master scheduler converts agreed demand into a buildable schedule, inside capacity and inside the time fence, the horizon in which the schedule is frozen against change.

The planner also does not usually own inventory policy. The planner supplies the demand variability that policy is set against. Those limits sit with supply planning: manufacturing lead times, minimum order quantities, production levelling and safety stocks on SoftwareConnect, and capacity, lead time, minimum order quantity and safety stock on Intuendi.

Applied to finished-goods buying, there is no plant sequence to freeze. The closest equivalent to the time fence is the purchase-order release, or a supplier freeze window if the contract has one.

What each side takes in, and what it is scored on

On a Prediko page aimed at Shopify brands, demand inputs include historical sales, market trends, seasonality, and promotions or competitor activity. Prediko also says demand planners work with sales, marketing and finance so forecasts match business goals. Logistics Bureau, a consultancy, contrasts less defined external data on the demand side with more concrete internal data on the supply side.

Internal data can still move. A 19 May 2025 article on demand-planning.com says supply planning's practical calculations, using more reliably sourced data, are typically less volatile. Netstock, an inventory software vendor, says supply planning feeds the demand plan with historic sales, inventory planning information, actual lead times and supplier delivery performance. Actual lead times are in that feed, so a change in supplier performance can come back into the next demand plan.

The outputs differ in the same way. The 19 May 2025 article says demand planning produces predictions that then affect supply planning and other decisions, and describes supply planning's payoff as inventory optimisation. For a finished-goods brand, Prediko's supply scope is the relevant list: purchasing, production, inventory allocation and logistics. Purchasing and logistics will do more of the work than plant sequencing.

The careers page says demand planners defend one monthly forecast and are measured on accuracy and bias. Fill rate, on-time-in-full delivery to customers, inventory turns, and excess and obsolete write-offs are not a forecast score on their own. Supply, capacity, procurement and logistics sit in those results, and so does the forecast. Keep accuracy and bias on the forecast, and read availability and excess as shared outcomes, even when one person made both decisions.

Forecast first, then a check, then a return path

ASCM's supply planning page says the demand plan comes first, and describes the relationship as a two-way street and a balancing act.

Demand planning is the key input to supply planning, and the supply plan is the output of that work: a feasible plan within capacity, lead time and minimum order quantity (Intuendi).

The sequence that fits these sources is a forecast, then a feasibility check, then feedback. Demand still comes first. Supply can still send constraints and actual supplier performance back.

Inbound Logistics describes another place to put the limits: constrained demand forecasting, which accounts for business limitations so the forecast is realistic and achievable.

If the supplier cap is written into the forecast, demand you might have served drops out of the sheet. Keep the unconstrained forecast, the units you think customers would buy, beside the committed plan, the units you will place. The difference is the demand you are choosing not to cover.

Three lines to hand across

1. Hand across unconstrained units by SKU and week, with the promotion and paid-spend assumptions written next to them.
2. For the feasible order, name lead time, minimum order quantity, capacity or supplier limit, safety stock, stock on hand, stock already on order, and the cash the order will tie up, then name the quantity and the date you will release.
3. On the return, send actual lead time and supplier delivery performance back to the person who owns the forecast. Send constraints back as demand decisions, not as a silent edit to last month's units.

Intuendi frames demand planning as forecasting and as influencing customer demand. The careers description does not give the planner the purchase order, the plant sequence or, in most organisations, the master schedule. Marketing in a D2C brand can change demand through promotions and ad spend. Record those actions as assumptions on line 1. Change them on line 3, once the gap is known. Accuracy and bias still refer to the unconstrained units, not to the units you later decided not to chase.

Logistics Bureau's examples of supply feedback are specific: phase a promotion, limit a low-margin SKU, or prioritise a key customer. Extending the promotion example to a Meta or Google Ads change is this article's inference. If the order cannot cover the forecast, hold or phase the spend that sat in the assumptions. Whether the product can carry paid acquisition is a separate margin test: when a product can afford paid acquisition. For the SKU you might limit, see Do you retire product rules, or judge SKUs on CM2?.

Line 2 also needs a location check. Units on hand in another node are not automatically available inside the supplier lead time. See 3PL, owned fulfilment or hybrid: how to choose without guessing.

A small business can combine both functions

SoftwareConnect says a small business can combine both functions into one department. Housing the work in one department is not the same as one person holding both roles.

The safeguard is still two passes. The person closest to promotions and ad spend writes the unconstrained forecast and the assumptions. The person who can commit cash signs the purchase order after the feasibility check. When that is the same person, write the unconstrained units before opening the cash balance, then write what you cut and why. If both happen in one pass, the risk is that the forecast becomes the quantity you can afford.

Once a month, a founder can set expected units next to what the supplier can ship, and next to whether the cash has to leave before the goods can sell.

An invented order, so you can see the gap

Every figure below is invented, including the safety stock.

Take one finished-goods SKU. The unconstrained forecast is 30 units a week. Supplier lead time is 6 weeks, and you review orders every 6 weeks, so the next buy is not next week. The protection window, lead time plus review interval, is 12 weeks. Expected demand in that window is 30 times 12, which is 360 units. Add assumed safety stock of 60. You have 40 on hand and nothing inbound. A full cover means buying 360 + 60 - 40, which is 380 units.

The minimum order quantity is 200. At an assumed cost of 8 per unit, in whatever currency the supplier invoices, one minimum order commits 1,600. If orders have to be whole multiples, the feasible buys are 200 or 400. Four hundred units would cost 3,200 and would cover the 380. The cash assumption in this example is that this month funds 1,600 and does not fund 3,200. You release 200.

The 380 only holds if demand from before the receipt can still be filled from the receipt. This example assumes lost sales, so those missed units are not filled when the goods arrive. The 40 on hand last about 1.3 weeks. With receipt at week 6, demand before receipt is 180 and 140 of those units go unserved. Releasing 400 instead of 200 would not recover that 140, because the goods still arrive at week 6. From receipt to week 12 the forecast is another 180. Holding 60 of safety stock after that 180 would take 240 at receipt. The released 200 covers that 180 and leaves 20, which is 40 below the 60 safety stock. Four hundred would cover the 380 on the netted formula, and under lost sales it would leave 220 at week 12, 160 above the 60 safety stock. Leave the forecast at 30 a week.

At 30 a week, 200 units cover about 6.7 weeks of unconstrained demand, which is longer than the 6-week review interval. This example assumes the supplier invoices on shipment and that payment is due on the invoice. The 6-week lead time is the time to receipt, so the 1,600 leaves before the goods are in the warehouse, while those units are still unsold. The check is whether that cash can sit out until they sell.

Then send performance back. Suppose the supplier delivers in 8 weeks, not 6, and misses the date. Netstock says actual lead times and supplier delivery performance feed the demand plan. This example also applies the 8-week lead time to the next cover. With the review still every 6 weeks, that cover uses a 14-week protection window. Applying the longer lead time to the next cover does not edit the earlier forecast of 30 a week down to whatever arrived.