This article provides an overview of changes and additional features released on or before 6th January 2026

The changes made in this release concern Puzzel Workforce Management.

Forecasting Enhancements

Confidence & Accuracy Scoring

We have added overall gradings for confidence & accuracy to the instances tab.

• Confidence scoring is based on how closely correlated your source data is.  The more similar your source data, the more confidence you can have in the forecast.
• Accuracy scoring is based on comparisons between the forecast instance & the actual data for that campaign week.

AI Summaries

For each available score, you can see a further description of confidence or accuracy, with suggestions on how you might improve further forecasting.
A range of statical measures, made at the same times that we created the overall scores, are sent to an LLM to produce the summary.

You can access the summaries using the stars button or from the data menu.

https://help.puzzel.com/product-documents/user-guide/puzzel-wfm/planneradmin-portal/forecasting/forecast-scores-summaries

Server Side (Manual) Forecasting

In itself, this will give a performance boost to creating a forecast.  More importantly, it adds some features that had previously only been available in auto forecasting:

Low Volume Queues

It can be very tricky to forecast for queues that have low volumes.  Interval aggregations will often leave you with even lower overall volumes than expected.  The interval forecast volumes for each queue/day will now be scaled to equal the average daily volume across the source data, with the extra contacts being distributed to the intervals that are most likely to receive an extra contact.

Public Holidays & Non-Standard Contact Patterns

When you expect a day's contact arrival pattern to be different from usual, such as a public holiday, you should add a tag to that day for the relevant queues before creating a forecast instance.

• When forecasting for a day/queue with a tag, we will try to use the data from the last 12 days that have the same tag, no matter which source weeks were selected
• If there is no historical data with the associated tag, the data from the selected source weeks will be used as if there was no tag on that day/queue

Default set of weeks selected

We will now pre-select up to 12 recent weeks as source data.
You can still add & remove weeks as you see fit.

https://help.puzzel.com/product-documents/user-guide/puzzel-wfm/planneradmin-portal/forecasting/creating-a-forecast-using-the-wizard

Removal of Old Adherence Pages & Reports

The old adherence pages & reports have now been removed, leaving only what were the preview pages & reports.

The old adherence report endpoints will remain in the API for a while longer.

Bug Fixes

• Shift & Activity report counts 10 minute activity as 15 minutes
• Adherence score on Adherence Now! is actually the Straight score
• Adherence Activity Report results duplicated
• Customers in USA see Adherence page data separate to the time line