Demand Planning KPIs: What to Measure and How Often

Demand planning KPIs tell you whether the forecast is accurate, whether the process adds value and whether the plan is turning into service and cash, and a short, well-chosen set is far more useful than a wall of charts. Too many planning teams either track nothing beyond a headline accuracy percentage or drown in dozens of measures that nobody acts on. This guide lists the twelve KPIs that matter most for a consumer product company, gives the formula, the right frequency and the owner for each, and works through a small example in which a forecast with zero bias still has a 12 percent weighted error. It is part of demand planning and inventory optimization, follows what a demand planner does and the demand planning process, and leads to statistical forecasting and consensus demand planning. Operational measures like these also appear in operations KPIs for supply chain and planning leaders, and the cash effects are covered in cash flow management.

How to choose KPIs

Measure what you will act on. A KPI is worth tracking if someone owns it, if it can be moved by a decision, and if it is reviewed on a regular schedule. Group the measures into three families: forecast quality, process quality and business outcomes. Report each at the level where decisions are made, and avoid averaging across items of very different size, which hides problems.

Forecast quality

1. Weighted absolute percentage error (WAPE). The sum of absolute errors divided by the sum of actuals. Weighting by volume stops tiny items from dominating the result. Frequency: monthly, by item, category and total. Owner: demand planner. See WMAPE.

1. Bias. The sum of forecasts minus the sum of actuals, divided by the sum of actuals. A positive value means over-forecasting. Bias tells you the direction of errors, which accuracy alone does not. Frequency: monthly. Owner: demand planner. See forecast bias.

1. MAPE. The average of the absolute percentage errors by item. Simple but distorted by small items and zeros. Use with care. Frequency: monthly. See MAPE.

1. Accuracy at the lead time. Forecast error measured against the forecast made one lead time ahead, because that is the forecast that drove the purchase. A forecast that is accurate a week before the sale is of little use if the order was placed twelve weeks earlier. Frequency: monthly. Owner: demand planner.

Process quality

1. Forecast value added (FVA). The change in accuracy contributed by each step in the process compared with a naive forecast. It shows whether overrides and consensus help or hurt. Frequency: quarterly, or monthly for large steps. Owner: planning lead. See forecast value added.

1. Plan stability. How much the forecast for a given future month changes between cycles. Large swings make supply planning chaotic. Frequency: monthly. Owner: planning lead.

1. Cycle adherence. Whether the calendar was met: inputs on time, meeting held, plan published on schedule. Frequency: monthly. Owner: planning lead.

1. Exception rate. The share of items flagged for review that actually needed action. Too many false flags waste time. Frequency: monthly. Owner: demand planner.

Business outcomes

1. Service level or fill rate. Units or lines shipped complete on time divided by units or lines ordered. Frequency: weekly. Owner: operations. See in-stock rate.

1. Inventory turnover and days of supply. Cost of goods sold divided by average inventory; and the days of demand covered by stock. Frequency: monthly. Owner: supply chain and finance. See inventory turnover and days of supply.

1. Excess and obsolete inventory. The value or percentage of stock above a defined coverage or with no recent demand. Frequency: monthly. Owner: supply chain. See excess and obsolete inventory.

1. Cash tied up in inventory. The inventory balance and its change, linked to days inventory outstanding. Frequency: monthly. Owner: finance. See days inventory outstanding.

A worked example: zero bias, 12 percent error

Consider five items for one month (illustrative):

• WAPE: 60 / 500 = 12 percent
• Bias: (500 - 500) / 500 = 0 percent

Check: forecasts sum to 120 + 70 + 55 + 180 + 75 = 500; actuals sum to 100 + 80 + 50 + 200 + 70 = 500; absolute errors sum to 20 + 10 + 5 + 20 + 5 = 60. The total forecast was exactly right, so bias is zero, yet every item was wrong and the average item error was 12 percent. If you looked only at the total, you would conclude the forecast was perfect. If you looked only at bias, you would miss that stock would be too high on items A, C and E and too low on B and D. The lesson is to track both accuracy and bias, and to look at the level where stock is held.

Service level example

Suppose customers ordered 5,000 units in a week and 4,860 were shipped complete and on time. The fill rate is 4,860 / 5,000 = 97.2 percent. The shortfall is 140 units, or 2.8 percent. Review the shortfall by item to see whether it came from a few items (a planning problem you can fix) or was spread across many (a capacity or execution issue).

Turnover and days of supply example

If annual cost of goods sold is $24 million and average inventory is $4 million, turnover is 24 / 4 = 6.0 times a year, and days of supply are 365 / 6 = 60.8 days. If the target is 50 days, inventory would need to fall to 24 million × 50 / 365 = $3.29 million, a reduction of about $0.71 million. Check: 24,000,000 × 50 / 365 = 3,287,671, and 4,000,000 - 3,287,671 = 712,329. That is the cash that better planning could release, if service levels hold.

How often to review

How to present KPIs

Show trend over time, not only the latest month. Show the measure at the level that matters, and flag the items that drive most of the error. Keep one page for executives and a deeper view for the planning team. Put the definition of every KPI in a short dictionary so that everyone calculates it the same way.

Setting targets

Base targets on your own history and segment. Stable items can be forecast far more accurately than promoted or new items, so a single accuracy target for everything is misleading. Set targets that improve over time, and tie service and inventory targets together, because pushing one without the other just moves the problem.

Common mistakes

1. Reporting accuracy only at the total level.
2. Ignoring bias.
3. Measuring against the wrong lag, after the order has already been placed.
4. Too many KPIs with no owners.
5. Rewarding accuracy without regard to service and inventory.
6. Changing definitions without notice.

FAQs

What are the most important demand planning KPIs?

Weighted forecast error, bias, forecast value added, service level, inventory turnover or days of supply, and excess inventory.

How often should demand planning KPIs be reviewed?

Service and exceptions weekly, forecast quality and inventory measures monthly, and process measures such as forecast value added monthly or quarterly.

What is a good forecast accuracy?

It depends on the product mix and level of detail. Compare against your own history and a naive forecast, and set targets by segment.

Why track bias as well as accuracy?

Bias shows whether errors lean in one direction, which causes either chronic excess or chronic shortage, even when overall accuracy looks acceptable.

Which KPI shows whether planning adds value?

Forecast value added, which compares each step with a naive forecast.