# Setting Up Forecast Workbench

Forecast Workbench uses statistical models with historical sales data to forecast demand into the future. It uses R modeling techniques and can either automatically select the best model or allow you to choose one manually.

## What's Already Set Up

Claret seeds some data for new tenants. Before starting, check what already exists:

* **Calendars** — Default calendars and calendar definitions have been created.
* **Hierarchy Structures** — Default Finished Goods Item Hierarchy and Customer Group Hierarchy structures exist.

## Prerequisites

* Review [Global Settings](https://docs.claret.app/using-claret/getting-started/global-settings) to confirm calendars and timing intervals are configured
* If you've set up Sales Collaboration, most master data is already done
* **Historical sales data is required** — the statistical models need history to generate forecasts

## Setup Steps

Forecast Workbench requires the following master data to be configured:

| Step | Data                     | Description                                  | Instructions                                                                             |
| ---- | ------------------------ | -------------------------------------------- | ---------------------------------------------------------------------------------------- |
| 1    | Item Hierarchy           | Define how your finished goods are organized | [Item Hierarchy](https://docs.claret.app/master-data/item-hierarchy)                     |
| 2    | Items                    | The products you forecast                    | [Items](https://docs.claret.app/master-data/items)                                       |
| 3    | Customer Group Hierarchy | Define how your customers are organized      | [Customer Group Hierarchy](https://docs.claret.app/master-data/customer-group-hierarchy) |
| 4    | Customer Groups          | Who you sell to                              | [Customer Groups](https://docs.claret.app/master-data/customer-groups)                   |
| 5    | Item @ Customer Groups   | Which items are sold to which customers      | [Item @ Customer Groups](https://docs.claret.app/master-data/item-customer-groups)       |
| 6    | Sale Types               | History and Forecast types                   | [Sale Types](https://docs.claret.app/master-data/sale-types)                             |

### Application Settings

Forecast Workbench has module-specific settings that must be configured:

1. Go to Settings > Application Maintenance > [Application Settings](https://docs.claret.app/application-maintenance/application-settings)
2. Configure the Forecast Workbench settings, including the default timing interval and forecast parameters

{% hint style="warning" %}
Ensure Application Settings for Forecast Workbench are configured before using the module.
{% endhint %}

## Loading Data (Required)

Forecast Workbench requires historical sales data to generate statistical forecasts:

1. Go to Settings > Transactional Data > [Sales](https://docs.claret.app/transactional-data/sales)
2. Import historical sales data against your Item @ Customer Groups
3. Ensure you have sufficient history for meaningful forecasts (typically 12+ months)

## What's Next

* [Forecast Workbench](https://docs.claret.app/sell/forecast-workbench) — detailed usage and model configuration
* [Forecasting Models](https://docs.claret.app/sell/forecast-workbench/forecasting-models) — understand the available statistical models

## Verification

To verify Forecast Workbench is set up correctly:

1. Navigate to Sell > Forecast Workbench
2. You should see your Items and Customer Groups listed on the left
3. Select an Item and Customer Group combination
4. Click the configuration (blue) button to set up forecast parameters
5. Set your timing interval, history date range, and forecast horizon
6. Click the Play button to run the model
7. You should see a chart with history and forecast data

**If no data appears:**

* Verify Item @ Customer Group combinations exist
* Verify historical sales data has been imported
* Verify Application Settings for Forecast Workbench are configured

## Common Questions

### How much history do I need?

More history generally produces better forecasts. We recommend at least 12 months of data, but the models can work with less.

### What if I have gaps in my history?

The Missing Imputation setting in the forecast configuration tells the model how to handle gaps — you can choose to use zeros, averages, or other methods.

### Can I override the statistical forecast?

Yes. After running the model, you can manually override values in the data table below the chart.