Designing Holiday and Special Event Forecasting Adjustments Using Historical Analog Days

What This Guide Covers

This guide details the engineering methodology for configuring Holiday and Special Event adjustments within Genesys Cloud WFM by leveraging Historical Analog Days. You will learn how to map non-recurring demand patterns to historical data points, apply multipliers to base forecasts, and validate the resulting volume predictions to ensure staffing accuracy during volatility.

Prerequisites, Roles & Licensing

• Licensing: Genesys Cloud WFM License (Standard, Premium, or Enterprise). The Premium tier is recommended for access to advanced forecasting algorithms and higher historical data retention limits.
• Permissions:
  • WFM > Forecast > Edit
  • WFM > Holiday > Edit
  • WFM > Special Event > Edit
  • WFM > Analog Day > View (Required to verify historical data integrity)
• External Dependencies:
  • At least 12 months of historical interaction data (voice, digital, or blended) ingested into the WFM data store.
  • A stable Base Forecast model already configured and validated for regular business days.

The Implementation Deep-Dive

1. Selecting and Validating Historical Analog Days

The foundation of any special event forecast is the selection of a valid Historical Analog Day. This is the day in the past that most closely resembles the expected behavior of the future event. In Genesys Cloud, this is not a manual selection of a date, but a configuration of logic that the forecasting engine uses to retrieve data.

Architectural Reasoning: Why Analog Days Matter

Standard forecasting models (such as Exponential Smoothing or ARIMA) rely on seasonality (weekly, monthly, yearly) and trend. They fail to account for “outliers” or discrete events that disrupt the normal pattern. A holiday or special event breaks the weekly seasonality. By injecting an Analog Day, you are effectively telling the engine: “Ignore the standard Tuesday pattern for this future date; instead, look at how the system behaved on this specific past date, then apply growth or shrinkage factors.”

Configuration Steps

1. Navigate to Workforce Management > Forecast > Analog Days.
2. Create a new Analog Day definition. You must specify:
   • Name: A descriptive identifier (e.g., Black_Friday_2023_Analog).
   • Date: The specific historical date to use as the source of truth.
   • Channel: Voice, Email, Chat, or Social. The analog day must match the channel being forecasted.
   • Shift Pattern: If your staffing model changes significantly on holidays (e.g., no overnight shifts), you must ensure the analog day reflects a similar shift availability, or the shrinkage calculations will be incorrect.

The Trap: The “Calendar Mismatch” Error

The most common failure in analog day selection is choosing a date that falls on a different day of the week than the target event, without accounting for day-of-week seasonality.

• Scenario: You are forecasting a product launch on a Wednesday. You select a historical Friday as the analog day because it had high volume.
• The Failure: The forecasting engine applies the raw volume from the Friday to the Wednesday. However, customer behavior on Fridays often differs from Wednesdays (e.g., higher abandonment rates, different call duration distributions). If your base model has a strong day-of-week seasonality component, injecting a Friday pattern into a Wednesday slot creates a distortion that the algorithm cannot smooth out, leading to overstaffing in the morning and understaffing in the evening.
• The Solution: Always select an analog day that falls on the same day of the week as the target event if possible. If that is not possible (e.g., the event is a one-time Saturday launch and you have no prior Saturday data), you must manually adjust the intra-day distribution (shape) in the post-forecast validation phase.

2. Configuring Holidays and Special Events

Once the Analog Day is validated, you must define the Holiday or Special Event entity that triggers the adjustment. In Genesys Cloud, “Holidays” are typically recurring (e.g., Christmas), while “Special Events” are one-off (e.g., a Flash Sale). The configuration logic is nearly identical, but the recurrence settings differ.

Step-by-Step Configuration

1. Navigate to Workforce Management > Calendar > Holidays (or Special Events).
2. Click Add Holiday.
3. Define the Name and Date Range.
   • Note: For multi-day events (e.g., Black Friday through Cyber Monday), create a single Holiday entry spanning multiple days. Do not create separate entries for each day unless the demand pattern changes drastically between them.
4. Under Forecasting Settings, select Use Analog Day.
5. Link the Analog Day created in Step 1.
6. Configure the Multiplier.
   • 1.0: Use the historical volume exactly as recorded.
   • >1.0: Scale up (e.g., 1.15 for 15% growth).
   • <1.0: Scale down (e.g., 0.8 for a decline).

The Trap: The “Shrinkage Blind Spot”

When applying a multiplier to an Analog Day, operators often apply the multiplier to the Volume only. They forget that Shrinkage (breaks, meetings, absenteeism) is also forecasted.

• The Failure: You set a multiplier of 1.2 on a Holiday. The engine multiplies the volume by 1.2. However, the default shrinkage profile for that day (e.g., “Regular Weekday”) remains active. If your Holiday shift pattern requires different break lengths or has different absenteeism expectations (e.g., higher no-show rates on holidays), the headcount calculation will be wrong.
• The Solution: Verify that the Shift Pattern associated with the Holiday matches the operational reality. If you are working a “Holiday Schedule” with reduced breaks, ensure the Analog Day used also had a similar break structure, or manually adjust the shrinkage percentage in the Forecast Validation step.

3. Applying Multipliers and Growth Factors

The multiplier is the lever that adjusts for year-over-year growth or specific event intensity. This is where architectural precision is critical.

Architectural Reasoning: Linear vs. Logarithmic Growth

If your contact center has experienced exponential growth, a linear multiplier may underpredict volume. Conversely, if the market has saturated, a linear multiplier may overpredict.

• Best Practice: Analyze the Year-Over-Year (YoY) growth trend for the specific channel.
  • If growth is stable at 10% per year, use 1.10.
  • If growth is accelerating, consider using a higher multiplier for future years (e.g., 1.15 for next year, 1.20 for the year after).
  • Caution: Do not compound multipliers. If you have a Base Forecast that already includes a growth trend, and you apply a Holiday Multiplier, you are double-counting growth.
  • The Rule: The Analog Day volume is “flat.” The Multiplier should account for all growth from the Analog Date to the Target Date.

Code/Config Example: JSON Payload for API Creation

If you are automating this via the Genesys Cloud WFM API, the payload structure for a Special Event looks like this:

POST /api/v2/wfm/forecast/special-events

{
  "name": "Summer_Sale_2024",
  "startDate": "2024-07-15T00:00:00.000Z",
  "endDate": "2024-07-15T23:59:59.000Z",
  "timezone": "America/New_York",
  "forecastSettings": {
    "analogDayId": "a1b2c3d4-e5f6-7890-g1h2-i3j4k5l6m7n8", // ID from Step 1
    "multiplier": 1.12, // 12% growth from analog year
    "useAnalogDay": true
  },
  "channels": [
    {
      "channelId": "voice",
      "enabled": true
    },
    {
      "channelId": "email",
      "enabled": true
    }
  ]
}

The Trap: The “Base Forecast Override” Conflict

Genesys Cloud allows you to set a “Base Forecast” multiplier and a “Holiday” multiplier. If both are active, the engine applies them multiplicatively.

• The Failure: You have a global growth multiplier of 1.05 in your Base Forecast settings. You also set a Holiday multiplier of 1.10. The final forecast is Volume * 1.05 * 1.10 = 1.155. If you intended for the Holiday to simply be 10% higher than the current base, you have now over-forecasted by 5.5%.
• The Solution: When using Analog Days, disable the global growth multiplier in the Base Forecast settings for the duration of the event, or set the Holiday multiplier to (1 + Target_Growth) / (1 + Base_Growth). This normalizes the calculation.

Validation, Edge Cases & Troubleshooting

Edge Case 1: The “Zero Volume” Analog Day

The Failure Condition: The historical analog day had zero interactions (e.g., the channel was not launched yet, or there was a system outage).
The Root Cause: The forecasting engine multiplies zero by the growth factor, resulting in a forecast of zero. This causes the staffing recommendation to drop to zero, leading to a severe understaffing crisis.
The Solution:

1. Pre-Validation Check: Before finalizing the forecast, run a “Forecast Validation” report. Filter for dates with Forecasted Volume = 0.
2. Manual Override: If the analog day is invalid, do not use an Analog Day. Instead, switch the forecasting method for that day to “Manual Entry” or use a “Similar Day” heuristic (e.g., average of the last 5 Tuesdays) and apply a conservative multiplier.
3. API Fallback: Use the GET /api/v2/wfm/forecast/analog-days endpoint to programmatically check the volume of the selected analog day before applying it to the holiday.

Edge Case 2: The “Shift Pattern Mismatch”

The Failure Condition: The analog day was a full 24-hour operation, but the holiday will be an 8-hour operation. The forecast shows high volume in the overnight hours, but the shift pattern does not schedule agents.
The Root Cause: The Analog Day provides the intra-day distribution (shape) of the volume. If the shape includes overnight peaks, but the shift pattern cuts off at 6 PM, the forecasted volume in the evening will be “unstaffable,” causing the service level to drop to 0%.
The Solution:

1. Normalize the Shape: In the Forecast Validation tool, use the “Adjust Shape” feature to redistribute the volume from the overnight hours to the active hours.
2. Alternative Analog: Select an analog day that had similar operating hours. If none exist, use a “Standard Day” analog and apply a volume multiplier, but manually flatten the intra-day distribution to match the holiday’s expected behavior.

Edge Case 3: The “Digital Channel Spike”

The Failure Condition: A holiday event drives massive traffic to Chat and Social, but the Analog Day was selected based on Voice history.
The Root Cause: Digital channels often have different seasonality and elasticity than Voice. A voice-heavy analog day may underpredict digital volume if the event is digitally driven.
The Solution:

1. Channel-Specific Analogs: Never use a Voice Analog Day for a Chat Forecast. Create separate Analog Day definitions for each channel.
2. Correlation Analysis: If you lack historical digital data for the specific event, analyze the correlation between Voice and Digital during similar past events. If Voice increased by 20% and Digital by 50%, apply a higher multiplier to the Digital forecast than the Voice forecast.

Official References

• Genesys Cloud WFM: Configure Analog Days
• Genesys Cloud WFM: Configure Holidays and Special Events
• Genesys Cloud Developer: WFM Forecast API Reference
• Genesys Cloud WFM: Validate Forecasts