# Statistical vs Sales Forecast Blend Calculator

> Source: https://netray.co/tools/forecast-blend-statistical-vs-sales-calculator
> Publisher: Netray (netray.co), AI and integration engineering for ERP and manufacturing systems.
> Section: Free Tools

## Short answer

Weight the statistical forecast and the sales or customer forecast by how accurate each has been, and see the consensus number and the gap between sources.

## Category

supply-chain

## Audience

- Demand planner
- Sales operations
- S&OP lead

## Tool

**Kind:** calculator

**Inputs:**

#### Statistical forecast

**Key:** stat

**Type:** number

**Default:** 5200

**Min:** 0

**Step:** 50

**Unit:** units

#### Sales forecast

**Key:** sales

**Type:** number

**Default:** 6100

**Min:** 0

**Step:** 50

**Unit:** units

#### Recent MAPE of statistical forecast

**Key:** mape_stat

**Type:** number

**Default:** 22

**Min:** 1

**Step:** 1

**Unit:** %

#### Recent MAPE of sales forecast

**Key:** mape_sales

**Type:** number

**Default:** 35

**Min:** 1

**Step:** 1

**Unit:** %

#### Weighting mode

**Key:** mode

**Type:** select

**Default:** 0

**Options:**

##### Inverse error (automatic)

**Value:** 0

##### Manual statistical weight

**Value:** 1

#### Manual statistical weight

**Key:** manual

**Type:** slider

**Default:** 60

**Min:** 0

**Max:** 100

**Step:** 5

**Unit:** %

**Help:** Used only in manual mode.

**Outputs:**

#### Automatic statistical weight

**Key:** w_auto

**Expr:** (1 / mape_stat) / (1 / mape_stat + 1 / mape_sales) * 100

**Format:** percent

#### Statistical weight used

**Key:** w

**Expr:** if(eq(mode, 1), manual, w_auto)

**Format:** percent

#### Blended consensus forecast

**Key:** blend

**Expr:** stat * w / 100 + sales * (100 - w) / 100

**Format:** integer

**Primary:** true

#### Sales minus statistical

**Key:** gap

**Expr:** sales - stat

**Format:** integer

**Description:** Positive means sales is more optimistic.

#### Gap as percent of statistical

**Key:** gap_pct

**Expr:** gap / stat * 100

**Format:** percent

**Primary:** true

**Chart:**

#### Blended forecast as sales MAPE worsens

**Type:** line

**X:** mape_sales

**From:** 5

**To:** 80

**Steps:** 30

**Series:**

- blend

**Disclaimer:** Inverse-error weighting assumes errors of similar bias. Review bias separately and keep a record of why a manual override was chosen.

## Content

**Intro:** Most demand planning teams hold two numbers, a statistical forecast and a view from sales. Averaging them 50/50 is easy but ignores which one has been right. This calculator weights each by its recent accuracy, or by a split you set, and shows the gap between them.

**Sections:**

#### How to use it

**Body:** Enter the two forecasts for the same item, period and level. Add the recent MAPE of each, measured over the last three to six cycles at the same lag. In automatic mode the weights are proportional to one over MAPE, so the more accurate source gets more weight. Choose manual mode when you want to apply a policy such as 70/30 for mature items.

**Bullets:**

- Compare at the same lag and level
- Use at least 3 cycles of accuracy history
- Document each manual override
- Check bias separately from error

#### The method

**Body:** Combining forecasts usually beats the average single source because errors partly cancel. Inverse-error weighting sets the statistical weight to (1 over MAPE_stat) divided by the sum of (1 over MAPE_stat) and (1 over MAPE_sales). The blend is then stat times the weight plus sales times one minus the weight. This is a pragmatic stand-in for optimal weights, which also need the correlation between errors.

#### Example

**Body:** Statistical 5,200, sales 6,100, MAPE 22% and 35%. Weights are 61.4% and 38.6%, so the blend is 5,547, and the gap is 900 units or 17% of the statistical figure. If sales accuracy improves to 20% the weights flip toward sales and the blend moves to about 5,690. The chart shows how much a deteriorating sales forecast loses weight.

#### Common mistakes

**Body:** Blending at a level where one source has no information, for example a sales forecast that is really a target. Using MAPE from different horizons for the two sources. Letting a consistently biased source drive the blend rather than correcting its bias first. And not tracking forecast value add, which is the real test that the human input improves on the statistical baseline.

**Cta:** Netray can measure statistical and sales forecast accuracy from your ERP and S&OP data and keep these weights up to date for each product family.

**Faq:**

#### Is a 50/50 blend a bad idea?

**Answer:** Not always, but it ignores track record. If one source is consistently twice as accurate, equal weights give away accuracy. Weighting by recent error is simple and usually better.

#### Why use MAPE and not bias?

**Answer:** MAPE measures size of error and is easy to compare. Bias shows direction and should be tracked too: a consistently optimistic source should be corrected before it is blended.

#### What is forecast value add?

**Answer:** The change in accuracy caused by a step in the process, such as a sales override on the statistical forecast. If the step makes accuracy worse, it adds no value.

#### Can I use more than two sources?

**Answer:** Yes. The inverse-error idea extends to any number: each weight is one over its error divided by the sum of those values.

## Search terms

forecast blending calculator, consensus forecast calculator, statistical vs sales forecast, weighted forecast combination, inverse error weighting forecast, forecast combination accuracy, sales input demand plan, consensus demand plan weights

---
Talk to an engineer: https://netray.co/contact?intent=erp_ai | Machine catalog: https://netray.co/agent/catalog.json