Skip to main content

​
Forecast Administration

The Forecast Admin section provides tools for configuring how forecasts are generated, managing bulk data, and controlling forecast versions. These settings significantly impact forecast quality.
Forecast Administration is restricted to Admin users.

​
Accessing Forecast Admin

Navigate to Demand Forecast → Admin (or through the settings menu).

​
Admin Tabs

The Forecast Admin page has multiple tabs:
TabPurpose
SettingsModel selection and configuration
UploadBulk forecast data imports
VersioningManage forecast versions

​
Model Selection

A model is a method for generating a possible forecast. Each model uses an algorithm that defines how forecast values are calculated. Your organization can define multiple models with different algorithms and compare their outputs before selecting one as active.
For a deep dive into how each algorithm works, including mathematical formulas and parameter details, see the Forecast Algorithms page.

​
Available algorithms

Tether provides 11 forecast algorithms you can assign to models:
AlgorithmDescriptionBest forSupports auto-tune
90-Day Rolling AverageSimple arithmetic mean of the last 90 days of salesStable, predictable demandNo
120-Day Rolling Average (Recursive)Sliding window average that incorporates prior forecast values as it projects forwardVolatile products, strategic planningNo
Exponentially Weighted Moving Average (EWMA)Applies exponential decay weighting to emphasize recent salesTrending products, fashion itemsYes
Linear TrendLeast-squares linear regression projected forwardProducts with consistent growth or declineNo
Simplified SeasonalWeighted combination of weekly (80%) and monthly (20%) seasonal patternsSeasonal and holiday-driven productsYes
Simple Seasonal with Trend AdjustmentSeasonal patterns blended with a year-over-year growth factorSeasonal products with growth trendsYes
Seasonal Year-over-Year GrowthSeasonal baseline multiplied by observed year-over-year growth rateSeasonal products with strong annual growthYes
Rolling MomentumDetects acceleration or deceleration by comparing short-term vs long-term moving averagesViral, accelerating, or declining productsYes
Holt-Winters (Triple Exponential Smoothing)Captures level, trend, and seasonality with separate smoothing factorsComplex seasonal patterns with trendsYes
SARIMASeasonal AutoRegressive Integrated Moving Average for complex time seriesComplex time series with seasonal and trend componentsYes
Static BaselineDirectly uses uploaded baseline forecast values with no algorithmic processingExternal forecasts, manual imports, hybrid workflowsNo

​
Creating a model

1

Open the Models tab

Navigate to Demand Forecast → Admin and open the Models tab.
2

Click Create Model

Click Create Model to open the model creation dialog.
3

Configure the model

Provide a name and description, then select an algorithm from the list above.
4

Save

Click Save. The system calculates the model’s output for all SKU-channel combinations.

​
Selecting the active model

Only one model can be active at a time. The active model drives the official sales forecast used by downstream processes (consumption forecast, inventory forecast, and reorder calculations).
1

Compare model outputs

Use the Forecast Comparison view or click a row in the forecast table to open the model selection dialog and see outputs from all models side by side.
2

Select a model

Choose the model you want to activate and confirm your selection.
3

Forecast regenerates

The system regenerates the official sales forecast using the selected model’s algorithm.
Changing the active model triggers an immediate forecast regeneration for all SKU-channel combinations.

​
Parameter optimization (auto-tune)

Seven of the 11 algorithms support automatic parameter optimization. When enabled, the system finds the best parameter values by minimizing forecast error on historical data. Optimization methods:
  • Nelder-Mead Simplex — used for algorithms with continuous parameters (e.g., smoothing factors, weights). Iteratively adjusts parameters to minimize error.
  • Grid Search — used for SARIMA, which has discrete integer parameters (AR/MA orders). Tests all combinations within specified ranges.
Error metrics available for optimization:
MetricDescription
MSE (Mean Squared Error)Penalizes large errors heavily (default)
MAE (Mean Absolute Error)More robust to outliers
MAPE (Mean Absolute Percentage Error)Scale-independent comparison across SKUs
Per-SKU/channel optimization: Optimized parameters are stored per SKU-channel combination, so each product-channel pair can have its own tuned settings within a model.

​
Key algorithm parameters

Each algorithm has configurable parameters. Here are the most commonly tuned ones: Rolling Average:
  • window_days — number of days to average (default: 90 or 120)
EWMA:
  • window_days — window size in days (default: 60)
  • weight_factor — exponential decay factor, 0.1–0.99 (default: 0.8). Lower values emphasize recent data more heavily.
Holt-Winters:
  • alpha — level smoothing, 0.0–1.0 (default: 0.2)
  • beta — trend smoothing, 0.0–1.0 (default: 0.1)
  • gamma — seasonal smoothing, 0.0–1.0 (default: 0.1)
  • phi — trend damping, 0.8–1.0 (default: 0.95). Values below 1.0 prevent unrealistic long-term trend extrapolation.
SARIMA:
  • ar_order, diff_order, ma_order — non-seasonal ARIMA orders
  • seasonal_ar_order, seasonal_diff_order — seasonal orders
  • seasonal_period — season length in days (default: 365)
Rolling Momentum:
  • short_window_days — recent performance window (default: 30)
  • long_window_days — baseline comparison window (default: 90)
  • sensitivity — momentum reaction strength, 0.0–1.0 (default: 0.5)
For the full list of parameters, defaults, and optimization ranges, see Forecast Algorithms.

​
Forecast Settings

​
General Settings

SettingDescription
Forecast HorizonHow far ahead to forecast (in days)
GranularityForecast output is always at the day × SKU × channel level

​
Bulk Upload

​
Upload Tab

The Upload tab allows importing forecast data in bulk.

​
Template Types

Tether provides two different upload templates depending on your period configuration:
TemplateWhen to UseDownload
Standard Date TemplateDefault - when custom periods are NOT enabledDefault template download
Custom Period TemplateWhen custom time aggregations are enabledSelect “Period” format
The custom period template is only available if your organization has enabled custom time aggregations in period settings.

​
Standard Date Template

Use this template when your organization uses standard calendar dates. Required Columns:
ColumnRequiredDescription
sku_codeYesSKU identifier
channel_nameYesSales channel name
dateYesDate in MM/DD/YYYY or YYYY-MM-DD format
quantityYesForecast quantity (integer)
Example: 
sku_code,channel_name,date,quantity
SKU001,Amazon US,2025-01-01,1000
SKU002,Shopify,01/15/2025,500

​
Custom Period Template

Use this template when your organization has custom periods enabled (e.g., fiscal periods like P1-P12). Required Columns:
ColumnRequiredDescription
sku_codeYesSKU identifier
channel_nameYesSales channel name
period_labelYesPeriod name (e.g., P1, P10)
fiscal_yearYesFiscal year (e.g., FY2025, 2024)
quantityYesForecast quantity (integer)
Example: 
sku_code,channel_name,period_label,fiscal_year,quantity
SKU001,Amazon US,P1,FY2025,1000
SKU002,Shopify,P3,FY2025,500
Do not mix date and period columns in the same file. The system will reject files that contain both a date column and period_label/fiscal_year columns.

​
Upload Process

1

Download Template

Click Download Template and select the appropriate format:
  • Standard for date-based uploads
  • Period for custom period uploads (if enabled)
2

Prepare Data

Fill in the template with your forecast values. Ensure all required columns are populated.
3

Upload File

Click Upload and select your CSV file
4

Preview

Review the preview of changes:
  • New forecasts to be added
  • Existing forecasts to be updated
  • Errors to resolve
5

Confirm

Click Apply to import the forecasts

​
Upload Validation

The system validates all uploads: Common Validations:
  • SKU codes must exist in the system
  • Channel names must match existing channels
  • Quantities must be numeric and non-negative
  • Duplicate SKU + Channel + Date/Period combinations are flagged
Date Template Validations:
  • Dates must be in valid format (MM/DD/YYYY or YYYY-MM-DD)
Period Template Validations:
  • Period labels must match configured periods
  • Fiscal years must match configured fiscal years
Non-whole number quantities are automatically rounded to the nearest integer with a warning.
Invalid rows are rejected. You can fix errors and re-upload, or proceed with valid rows only.

​
Upload History

View past uploads:
  • Upload date and time
  • User who uploaded
  • File name
  • Record count
  • Status (success, partial, failed)

​
Versioning

​
What is Forecast Versioning?

Versioning maintains historical snapshots of forecasts:
Version TypeDescription
CurrentActive forecast being used
HistoricalPast versions for comparison
DraftWork-in-progress (not active)

​
Versioning Tab

The Versioning tab shows:
  • List of forecast versions
  • Version dates and descriptions
  • Actions (view, restore, compare)

​
Creating a Version

1

Go to Versioning Tab

Open the Versioning tab
2

Click Create Version

Start a new version snapshot
3

Add Details

  • Name: “Q1 2024 Plan”
  • Description: Reason for version
4

Save

Version is created and stored

​
Comparing Versions

Compare forecast versions:
  1. Select two versions
  2. Click Compare
  3. View differences by SKU/channel/period

​
Restoring a Version

To revert to a previous version:
1

Select Version

Click on the version to restore
2

Click Restore

Choose Restore action
3

Confirm

Review what will change and confirm
4

Version Active

Selected version becomes current
Restoring a version replaces the current forecast. Consider creating a version of the current state first.

​
Forecast Regeneration

​
When forecasts regenerate

The system recalculates model outputs automatically when:
  • A model definition is modified — the system immediately recalculates that model’s outputs for all SKU-channel combinations. If the modified model is the currently selected model, the official forecast is also regenerated.
  • The active model is changed — selecting a different model triggers a full forecast regeneration using the new model’s algorithm.
  • Historical data is updated — events like data syncs or manual sales history uploads trigger a recalculation of all active model outputs.
  • A daily update occurs — scheduled regeneration recalculates forecasts with the latest available data.

​
Manual regeneration

To manually trigger a forecast update:
1

Go to Forecast Admin

Navigate to Demand Forecast → Admin.
2

Click Regenerate

Click the Regenerate Forecasts button.
3

Confirm

Confirm to start the regeneration process. The system recalculates model outputs for all SKU-channel combinations using the currently selected model’s algorithm.
Regeneration recalculates algorithmic model outputs. User edits made through the Consensus Model editing workflow are preserved — they are not overwritten by regeneration.

​
Data Quality

​
Data requirements by algorithm

Different algorithms have different minimum data requirements:
AlgorithmMinimum history
Rolling Average, EWMA, Linear Trend, Rolling Momentum30–120 days (depending on window size)
Simplified Seasonal, Seasonal YoY Growth, Holt-Winters, SARIMA365+ days (at least 1 full seasonal cycle)
Static BaselineNo historical data needed (uses uploaded values)
Seasonal algorithms (Simplified Seasonal, Holt-Winters, SARIMA, Seasonal YoY Growth) automatically fall back to a 90-day rolling average if fewer than 365 days of history are available.

​
Data quality indicators

Watch for warnings about:
  • SKUs with insufficient sales history for the selected algorithm
  • Channels with sparse or missing data
  • Products with high variability that may need a different algorithm
  • Recent data gaps that could skew forecasts

​
Best Practices

  • Start with simpler algorithms (e.g., 90-Day Rolling Average) and compare against more complex ones (e.g., Holt-Winters, SARIMA).
  • Use the Forecast Comparison view to evaluate accuracy before switching.
  • Consider using different models for different product segments — stable products may perform best with Rolling Average, while seasonal products benefit from Simplified Seasonal or Holt-Winters.
  • Enable parameter optimization (auto-tune) for algorithms that support it to find the best settings per SKU-channel combination.
Keep forecasts current:
  • Ensure regular regeneration so forecasts incorporate the latest sales data.
  • Monitor data freshness — stale sales history leads to outdated forecasts.
  • Periodically review model accuracy and consider switching algorithms if performance degrades.
Use versioning effectively:
  • Create a version snapshot before switching the active model.
  • Name versions descriptively (e.g., “Before SARIMA switch Q1 2026”).
  • Compare versions during planning cycles to track forecast evolution.
Handle bulk uploads with care:
  • Validate data before upload — ensure SKU codes and channel names match existing records.
  • Always preview changes before confirming.
  • Create a version snapshot of the current forecast state before uploading.

​
Troubleshooting

​
Forecasts not updating

Possible causes:
  • No regeneration has been triggered (manual or automatic).
  • The modified model is not the currently selected model — only the selected model drives the official forecast.
  • Data sync has not completed, so the latest sales history is not yet available.
Solutions:
  1. Trigger a manual regeneration from the Admin page.
  2. Verify which model is currently selected and confirm it’s the one you expect.
  3. Check that data integrations have synced successfully.

​
Upload errors

Common issues:
  • Wrong template type (date template vs. period template).
  • SKU codes or channel names that don’t match existing records.
  • Invalid date format (use MM/DD/YYYY or YYYY-MM-DD).
  • Mixing date and period columns in the same file.
Solutions:
  1. Download the correct template for your organization’s period configuration.
  2. Verify all SKU codes and channel names exist in the system before uploading.
  3. Check for non-numeric or negative quantity values.
  4. Ensure no special characters or encoding issues in the CSV file.

​
Next Steps

Forecast Algorithms

Learn how each algorithm works, its parameters, and when to use it

Forecast Comparison

Compare model outputs and accuracy side by side

Forecast Dashboard

View and edit forecasts

Edit Log

Track forecast changes