Training & Setup
Breadcrumbs

Smart Forecaster: Edge Cases & Adjustments

This article covers scenarios where Smart Forecaster may behave differently than expected, along with guidance on when to contact your Client Success Manager (CSM). 

Bulk Uploads and Sudden Registration Jumps 

Smart Forecaster handles bulk registrations differently depending on whether they are part of your historical pattern. 

Recurring Bulk Uploads 

If bulk uploads are part of your historical registration operations and occur at roughly the same point in the registration cycle each year, the model will account for them. The adjustment occurs once that week's data closes out. 

One-Time Bulk Uploads 

One-time bulk uploads without historical precedent will distort the forecast. The model has no pattern to reference, so it cannot distinguish this spike from organic growth.  For reference, bulk uploads that will cause the most noise in the modeling will occur later in the campaign (within 8 weeks) and represent 3% or more shifts in net attendee lift vs. the similar time period with no upload. 

Recommended action: Before performing a one-time bulk upload, notify your CSM with the business reason. Bear Analytics will determine options to accommodate. 

Major external factors 

Certain external factors sit outside the model's knowledge base. These include: 

  • Bulk registrations outside normal patterns 

  • Pandemic-related closures or restrictions 

  • Sudden changes affecting travel 

  • Major economic disruptions 

The model cannot anticipate or automatically adjust for these scenarios. There is no manual override in Bear IQ. 

Recommended action: If you are aware of an external factor likely to impact your registration trajectory, notify your CSM. Interpret the forecast with that context in mind. 

 

Holidays and Date Shifts 

Smart Forecaster normalizes based on days out from the event, not absolute calendar dates. Weekends and holidays fall on different dates each year, but their position relative to your event date is what matters to the model. 

Testing and client experience show that shifting event dates by a few weeks from one calendar year to the next does not materially impact forecast accuracy. Additionally, because forecasting is conducted on a weekly basis, each calculation window inherently includes at least two weekend days, which smooths out day-of-week variations. 

 

Segmented Forecasts 

Some organizers want to forecast specific segments—for example, excluding exhibitor booth staff badges to focus on attendee registrations, or forecasting by registration type, demographic, or persona. 

 

Current Availability 

Segmented forecasts are available in a limited format as a customization. Contact your Client Success Manager to learn more. 

 

Key Consideration 

Segmenting reduces the data pool—both historical and current-year registrations. Smaller data pools reduce forecast sensitivity and accuracy. This is the primary limiting factor, not the forecast model itself. 

 

When to Contact Your CSM 

Reach out to your Client Success Manager in the following situations: 

Scenario 

What to Include 

One-time bulk upload planned 

Business reason, approximate volume and timing 

Known external disruption 

Nature of disruption, expected impact 

Extend forecast beyond 20 weeks 

Registration window length, desired time horizon 

Segmented forecast request 

Segment definition (badge type, registration code, etc.) 

 

Related Article 

For an overview of how Smart Forecaster calculates weekly goals and forecast ranges, see Smart Forecaster: How It Works