12 Demand Planning KPIs to Improve Forecasting in 2026

Demand planning is driven by data, but in practice it’s also shaped by real-world constraints and market shifts. 

Whether it’s working around global tariffs that force production moves, or managing dead stock that clogs your main warehouse, forecast accuracy depends on the demand planning KPIs used to guide your decisions.

Ahead, you’ll learn the 12 essential KPIs every planner should use to prioritize top-selling SKUs, improve availability, and ensure every item in the warehouse is destined for the customer’s doorstep. 

What is demand planning?

Demand planning is a prediction of what a company plans to sell in the future. It involves determining the demand for a product or service and building plans to support it. 

For example, when you look back at your sales history, you can see the volume and products sold. But you still need to know what you’ll be able to sell in the future. Planners have to predict which products they plan to sell, how many, where to sell them, and how to process them through the supply chain. 

In general, a demand planning flow looks like this:

• Data gathering. Pull sales, inventory on hand, and lead times. Unified data here helps you learn exactly what is happening across your online storefront and point-of-sale system in real time. 
• Baseline forecast. Build a baseline demand forecast using historical data and seasonal trends.
• Demand review and validation. Adjust the model to account for external factors like marketing pushes, global tariffs, and changing buyer sentiment.
• Planning. Based on your forecast, build out purchase schedules, inventory allocation, and fulfillment capacity. This includes tiering locations, like Tier A for flagship stores that carry the full assortment, versus Tier C and D for factory outlets. 
• Optimization. Track sell-through rate. If a product is moving faster than expected, increase ad spend and re-orders to avoid stockouts. 

Demand planning is a core inventory management process for businesses across retail, B2B, and professional services. It’s different from demand forecasting, which predicts what customers might buy based on sales patterns, seasonality, and upcoming product launches. 

What are demand planning KPIs?

Demand planning KPIs are measurable values that show how effectively you’re achieving your objectives. They aren’t used to punish supply planners or dangle a carrot ahead of them for endless improvement. Instead, they share information so you can find uncertainty and bias in planning, and thus prioritize high-selling items and refine your process. 

Some KPIs focus on forecast quality, such as MAPE or Bias, to indicate whether your plan is reliable. Others focus on execution outcomes, such as fill rate or sell-through, to determine whether you actually delivered what was promised. Every demand planning KPI has a threshold and an action, including what will change, who will do it, and by when. 

The most important demand planning KPIs

1. Forecast error

Forecast error is the starting point for measuring the quality of your forecasts. It’s a fundamental metric that explains if your forecast is drifting, and serves as an input for more complex accuracy and bias calculations. 

For Tier A, or top-selling SKUs, even small unit errors impact total revenue. For Tier C SKUs, higher error margins are tolerated as long as they don’t lead to stockouts or dead inventory. If a SKU shows large errors in the same direction for two to four cycles, it’s a process issue like a missed promotion or shift in channels. 

In categories like footwear, error often hides in the size range. If sizes six through eight drive 80% of your business, but you continue to forecast and buy size 11s that represent only 1% of sales, you’re creating future dead stock.

The formula is: 

Forecast error = actual sales – forecast sales

In practice, say a retailer forecasts $10,000,000 in sales for a new outerwear line, but it generated only $7,500,000. The forecast error is $2,500,000, which means the company over-invested more than $2 million into inventory now sitting in a warehouse. 

If that same line was forecasted for $10,000,000, but sold $13,000,000, the $3 million error represents missed opportunity and likely thousands of frustrated customers facing out-of-stock notices. 

2. Forecast accuracy

Forecast accuracy measures the deviation between what your model predicted and the actual outcome. For established product lines, a target of 90% is achievable since you have years of data—there are few reasons for a big miss. If you’re expanding into new territory or launching a new product, an 80% rate is considered a win. Forecast accuracy often dips during events like BFCM and targets should be adjusted to account for volatility. 

The formula is:

Forecast accuracy % = 1 – (|actual – forecast| / actual)

For example, say an electronics retailer is forecasted to bring in $5,000,000 in revenue, but the actual sales come in at $4,200,000. The forecast error is $800,000. Following the formula: $1 – (800,000 / 4,200,000) = $0.809, or 80.9% accuracy.

3. Mean absolute percentage error (MAPE)

MAPE is a common evaluation metric to assess the accuracy of a forecasting model. It measures the average magnitude of error as a percentage of the actual values. 

The formula is: 

MAPE = (1 / n) * Σ [ |actual – forecast| / |actual| ] x 100

Where:

• n: The total number of data points or forecasts.
• Actual: The real, observed value for a given period.
• Forecast: The predicted value for that same period.
• |Actual – forecast|: The absolute difference between the actual and forecast.
• Σ: The sum of these individual relative errors.

If you have a core product that consistently sells 100 units a week, and your MAPE is 10%, you know your inventory risk is consistently within a 10-unit margin. You can set very tight safety stock levels and free up capital. Use MAPE as a diagnostic tool for your stable, always-on SKU where sales rarely hit zero. 

4. Symmetric mean absolute percentage error (sMAPE)

Standard MAPE can be biased because it treats over-forecasting differently than under-forecasting. sMAPE symmetrizes the error by averaging the forecast and actuals in the denominator, making it more reliable for newer products with fluctuating volumes.

The formula is:

sMAPE = (1/n) * Σ [ |actual – forecast| / ((|actual| + |forecast|)/2) ] x 100

Where:

• n: The number of data points or observations.
• Σ: Summation symbol (add up the results for each data point).
• |Actual – forecast|: The absolute difference between the actual value and the predicted value.
• ( (|Actual| + |forecast|) / 2 ): Half the sum of the absolute actual and forecast values, making the error symmetric.

When launching a new SKU, your sales might jump from five units to 50 units in a week. sMAPE prevents these early low-volume misses from making your forecast model look like a total failure. It provides a more balanced view of your progress.

5. Weighted absolute percent error (WAPE)

WAPE is your best single scoreboard KPI because it prevents a tiny SKU with a huge percentage swing from distorting your performance data. It calculates the total error across all SKUs and divides it by total actual sales.

The formula is:

WAPE = (Σ |actual – forecast|) / Σ actual

Where:

• Σ |actual – forecast|: Sum of the absolute differences between the actual demand and the forecasted demand for each time period.
• Σ Actual: The sum of all actual demand over all periods. 

Say you sell two items. Item A sold 1,000 units against a forecast of 1,100, a 100-unit error. Item B sold 10 units against a forecast of 20—a 10-unit error. Without weighting, both look like 10% to 50% misses. WAPE combines them to show that your total volume error is only about 10.8%

6. Mean forecast error (MFE) 

If forecast error tells you how much you missed by, forecast bias tells you which direction you are heading in.

• A positive bias means you’re under-forecasting and selling more than you plan for. Under-forecasting leads to stockouts, emergency freight costs, and lost revenue.
• A negative bias means you’re over-forecasting and overestimating demand. Over-forecasting can lead to dead inventory, high storage fees, and eventually deep discounting that affects your margins.

The formula is:

Σ(forecast – actual) / n

Where:

• Σ(forecast – actual): Sum of the difference between forecasting and actual values. 
• n: The number of periods. 

Say a consumer electronics brand tracks a flagship laptop over four months. It forecasts 1,000 units per month but has seen sales of 1,200, 1,150, 1,100, and 1,050. The sum of the errors is -500.

Using the formula, the MFE is -125 units, indicating an under forecasting bias and suggesting the team is being too conservative. 

7. Tracking signal

Tracking signal identifies when forecasts are consistently too high or too low over time, even if your MAPE looks acceptable. It compares your cumulative forecast error against typical error size to ensure that normal volatility doesn’t trigger a false alarm. Planners usually escalate if the signal crosses a predefined threshold for two consecutive planning cycles. 

The formula is:

Tracking signal = cumulative forecast error / mean absolute deviation (MAD)

For example, you’re tracking a high-volume SKU over six months. The MAPE is a healthy 10%, but the tracking signal has climbed to +5.2. Despite the good accuracy percentage, the positive tracking signal means an under-forecasting issue that requires a review of the demand model before a major stockout. 

8. Order fill rate

Order fill rate is an important customer experience KPI that measures the percentage of demand fulfilled immediately from available inventory. It’s a precursor for delivery speed because you can’t fill an order on time if the inventory wasn’t fillable in the first place. 

Retailers prioritize first-pass fill rates, which means the order was fulfilled in full from the initially assigned location without substitutions or splits. High fill rates are correlated with high customer satisfaction, as it suggests customers received their orders without delay.

The formula is:

Order fill rate % = (units shipped on first pass / units ordered) x× 100

In practice, say a beauty brand receives orders for 1,000 units of a trending serum. Due to a lag in inventory sync, it has only 850 units available at the main distribution center. The remaining 150 have to be back ordered or split-shipped from a secondary location. The fill rate is 85%.

The customer will eventually get the product, but the 15% miss represents higher shipping costs and risk of order cancellation. 

9. Perfect order rate

The perfect order rate is another demand forecasting KPI that your customers feel. It measures the probability that an order is taken correctly, allocated immediately, delivered on time, and arrives undamaged. 

To improve this metric, you have to categorize why an order wasn’t perfect. Some common points include carrier-side delays, pick/pack errors, and inventory allocation failures.

The formula is: 

Perfect order rate = (on-time % x complete % x damage-free % x accurate docs %)

Each individual component of perfect order rate can look good on its own, but when applied with the formula show broad operational gaps. For example, say a retailer delivers 95% of orders on time, 98% complete, 99% damage-free, and 99% with accurate documentation. 

Each department looks successful, however, the perfect order rate is 91.2%. That means 9 out of 100 customers still have a less-than-ideal experience. 

10. Inventory turnover

Inventory turnover measures how many times your stock is sold and replaced over a specific period. It indicates how efficiently you’re using capital. 

A high inventory turnover rate reduces holding costs like insurance, security, and obsolescence. However, turnover rates aren’t static—a healthy number for a seasonal, trending fashion catalog will look different from an evergreen catalog of stable basics. 

The formula is:

Inventory turnover = cost of goods sold (COGS) / average inventory

For example, an apparel brand has a COGS of $4,000,000 and carries an average inventory value of $1,000,000. Its turnover is 4.0.

If the brand implements better demand planning and reduces average inventory to $800,000 while maintaining the same sales volume, its turnover improves to 5.0. A change like this frees up $200,000 in cash flow that was previously locked up in warehouse stacks. 

11. Sales-to-inventory ratio

This demand forecasting KPIs measures whether inventory is in balance with your sales velocity. If the ratio is too high, your capital is sitting on shelves. If it’s too low, you’re at high risk for a stockout. 

Low turnover in bulky categories, like knee-high boots in large boxes, can physically max out your back-of-house capacity. When turnover slows, the cost is a lack of physical square footage to receive new, high-turning products.

When sales-to-inventory ratio is flat, it means inventory is accumulating and tying up capital that could be used for marketing, payroll, or faster-moving products. In general, calculate sales-to-inventory (or stock-to-sales ratio) by category or vendor for a more accurate reading. For a good baseline, US Census data shows the total inventory-to-sales ratio for 2025 was 1.38.

The formula is:

Stock-to-sales ratio = ending Inventory / sales for the period

As an example, say your footwear brand has $500,000 in ending inventory for a boot category that generated $100,000 in sales for the month. The sales-to-inventory ratio is 5.0. If your goal is 3.0, you’re overstocked by $200,000. 

Before liquidating, the team can use multi location inventory reports to see if that stock is just parked at the wrong warehouse. If one location is bloated while another is lean, you can balance the nodes before resorting to discounts. 

12. Gross margin return on investment (GMROI)

GMROI brings profitability into the conversation for demand planners. It answers the question, “For every $1 tied up in inventory, how much gross margin is coming back?” Considering GMROI prevents planners from chasing metrics on products that don’t contribute to the bottom line. You can use it as a tool to decide whether to reorder, reprice, or liquidate a SKU. 

High sell-through at an outlet, for example, doesn’t always equal a healthy GMROI. If you are transferring aged inventory to clearance centers and selling it at a negative margin just to alleviate physical capacity in flagship stores, your GMROI on those items is failing. Successful demand planning identifies these redundant SKUs (like buying 20 styles of sweaters when only 10 are needed) before they drain margin for the outlet channel.

For an accurate GMROI, use consistent landed costs for your average inventory. If you don’t allocate inbound freight and duties correctly, your GMROI will look better than it really is.

The formula is:

GMROI = gross margin / average inventory cost 

For example, you’re reviewing two SKU categories. Category A has a high sell-through rate but low margins, resulting in a GMROI of 1.2. Category B has slower sales but much higher margins, resulting in a GMROI of 2.5.

If you follow GMROI to make decisions:

• Category B. Focus on improving discoverability through better merchandising or email flows. This category doesn’t need discounts, as the inventory is already very profitable. 
• Category A. Here you have a margin problem, not a demand one. Renegotiate supplier costs or adjust pricing versus ordering more product, since as of now it provides a poor return on invested capital.

Get full visibility into your supply chain with Shopify

To improve your supply chain management, demand planning is the heart of it all. And Shopify is here to help. 

With inventory reports like sell-through rate and ABC analysis, you can see what products are moving and where to adjust. For scaling brands, Shopify also leverages ShopifyQL for custom KPI modeling and integrates with major ERPs like NetSuite to sync demand with supply in real time. When you unify your operations with Shopify, you can eliminate dead stock and fulfill customer demand with absolute precision.