The Demand Planning Process, Step by Step

The demand planning process is a monthly rhythm that turns recent sales into an agreed plan for the next several months, and the companies that run it with discipline tend to have fewer surprises than those that treat planning as an occasional exercise. The process does not have to be complicated. It has seven steps, a fixed calendar, clear owners and a habit of looking back at what happened. This guide walks through each step, shows a sample monthly calendar, works through a top-down split of a family forecast with checked arithmetic, describes who does what, and lists the failure points that break the cycle. It is written for operations, supply chain and finance leaders at consumer product companies who want a practical process they can start running next month. It is part of demand planning and inventory optimization, follows demand planning vs. demand forecasting, and leads to the role of the demand planner and demand planning KPIs. The agreed plan also feeds cash flow management.

The seven steps

1. Collect and clean the data. Pull the latest actuals by item, customer and channel: shipments, orders and, where available, point-of-sale data. Close the period, remove or flag anomalies, and note events that distorted sales, such as stockouts, large one-time orders, returns and system outages. Update the item master for launches and discontinuations. This step sets the quality ceiling for everything that follows.

2. Produce the statistical baseline. Run the forecasting method on the cleaned history. Segment items first: stable high-volume items suit trend and seasonal models, intermittent items need methods designed for sporadic demand, and new items need analogs. Review exceptions where the forecast changed sharply from the last cycle or where recent actuals fell far outside the previous forecast. See statistical forecasting for demand planning.

3. Gather market intelligence. Collect what the history cannot know: planned promotions and their timing, price changes, new listings and delistings, product launches, competitor activity, macro shifts and customer-specific forecasts. Ask for the information in a standard template, with the quantity, the dates, the reasoning and the owner. Vague inputs such as "expect strong growth" should be turned into numbers or set aside.

4. Build the proposed plan. The planner combines the baseline and the inputs into a proposed demand plan, showing each adjustment separately so that the reasoning is visible. This is the document taken to the review.

5. Hold the consensus meeting. Sales, marketing, finance and operations review the proposed plan by exception: the largest items, the biggest changes, the highest risks. They challenge assumptions, agree on adjustments and record the decisions. See consensus demand planning.

6. Approve and hand off. The agreed plan is signed off by the accountable executive and passed to supply planning, procurement, production and finance. The plan is locked for the cycle, and any later change goes through a defined route. This hand-off is the point at which the plan becomes purchasing and production decisions.

7. Measure and learn. After the period closes, compare the plan with actuals, calculate accuracy and bias at the right levels, and review the largest misses to find root causes. Feed what you learn into the next cycle: cleaner data, better models, corrected assumptions. See forecast error root cause analysis.

A sample monthly calendar

The exact days depend on how quickly your systems close the month. The point is to fix the calendar and keep it. A cycle that slips by a week each month soon becomes out of date.

A worked example: top-down split

Many companies forecast at a product family level, where the signal is steadier, and then spread the result across items. Suppose the family forecast is 12,000 units for next month, and the recent sales mix is item X at 50 percent, item Y at 30 percent and item Z at 20 percent (hypothetical). The split is:

• Item X: 12,000 × 0.50 = 6,000
• Item Y: 12,000 × 0.30 = 3,600
• Item Z: 12,000 × 0.20 = 2,400

Check: 6,000 + 3,600 + 2,400 = 12,000. The mix should be updated as it changes. If item Z is being discontinued, its share must be redistributed, and a planner might move half of its volume to item Y and assume the rest is lost: Z's demand falls to zero, Y receives 1,200 units, which makes Y 4,800, and 1,200 units are lost from the family total, bringing it to 10,800 (6,000 + 4,800 = 10,800). The family total changes because of a decision, and that change should be recorded with a reason. Top-down splits are quick and stable, but they hide item-level surprises, so check a sample of items against bottom-up views. For large assortments, see SKU-level forecasting.

Roles in the process

The demand planner owns the calendar, the data, the baseline, the proposed plan and the measurement. The sales leader owns customer-level inputs and is accountable for the commercial assumptions. The marketing lead owns promotion and launch plans. Finance reconciles the plan to the budget and flags financial implications. Operations or supply planning reports constraints and receives the plan. The executive owner, often the head of supply chain or the COO, approves it and settles disputes. In small companies several of these roles are held by the same few people, but the responsibilities still need to be written down.

What good looks like at each step

Data is closed on time and anomalies are documented. The baseline is produced the same way each month, with changes to the method recorded. Inputs arrive in a standard format by a deadline. The proposed plan shows baseline and adjustments separately. The meeting is short, focused on exceptions and ends with decisions recorded. The approved plan is published once, and everyone refers to the same version. The review of last month's accuracy is a standing part of the next meeting.

Failure points

1. Late inputs. If sales delivers numbers the day of the meeting, nobody can challenge them. Set deadlines, and show who is late.
2. Over-long meetings. Reviewing every item wastes time. Review by exception and by value.
3. Undocumented overrides. If changes are not logged with reasons, you cannot learn which ones helped.
4. Multiple versions of the plan. Everyone must work from the same published version.
5. Treating the meeting as a negotiation over targets, not a review of expectations.
6. No feedback loop. Without measuring last cycle's error, the same mistakes repeat.
7. Skipping the cycle during busy periods, which is when it is most needed.

Scaling the process

A company with a hundred items can run this in a spreadsheet and a one-hour meeting. A company with thousands of items needs segmentation, so that the most valuable and most volatile items get attention and the stable tail is forecast automatically and reviewed only when it breaks tolerance. It also needs workflow tooling for collecting inputs and recording changes. See planning automation for deciding what to automate.

Common mistakes

1. Making the process too heavy at the start. Begin with the top items and a short meeting.
2. Using the sales target as the plan.
3. Not tying the plan to supply and finance after the meeting.
4. Forecasting at a single level and ignoring the others.
5. Never retiring bad overrides.

FAQs

What are the steps of the demand planning process?

Collect and clean data, build a statistical baseline, gather market intelligence, assemble the proposed plan, hold the consensus review, approve and hand off, and measure results.

How long should the monthly cycle take?

Most mid-sized companies complete it in about ten working days, from closing actuals to publishing the plan.

Who should attend the demand review?

The demand planner, sales and marketing leads, finance and operations or supply planning, with an executive who can resolve disputes.

What is a top-down forecast?

A forecast made at an aggregate level, such as a product family, then allocated to items using a mix. It is stable but can hide item-level changes.

How do I know the process is working?

Plans are published on schedule, accuracy and bias are tracked and improving, overrides are documented and reviewed, and purchasing uses the plan.