Glossary · P
Predictive routing

Definition
Predictive routing uses a model to assign contacts to the agents with whom a chosen KPI, such as first contact resolution, is likely to be achieved best.
Classic routing asks: who is free and has the right skill? Predictive routing adds a question: with whom is this contact likely to reach the best result, measured by a KPI that you choose?
#How it works
- Choose a KPI. For example first contact resolution, conversion or satisfaction after the contact.
- Learn from the past. A model analyses which combinations of request, context and handling led to good results.
- Rank. For each new contact, the available agents are sorted by expected result.
- Measure. The effect is checked against a control group that is distributed the classic way.
The model does not replace the rules of the ACD; it works within them. Skills, languages and permissions remain mandatory, and the prediction only chooses between suitable people.
#What it requires
- One clear KPI. Trying to improve everything at once improves nothing measurably.
- Enough data. The model needs enough past contacts per request.
- Wait-time thresholds. The best person is no use if the customer waits too long for them. Beyond a threshold, the next suitable person takes the contact.
- Honest metrics. Measuring what matters, rather than what is easy to count, yields better results.

#Limits and fairness
A model that always favours the same people overloads them. Good methods therefore balance the load within the prediction too and keep an eye on it.
Because the method evaluates the results of individual agents, in Germany it falls under the works council's co-determination. Purpose, KPI and level of analysis should be agreed before go-live.
#Predictive routing in Univerx
In Univerx, predictive routing assigns each contact to the person most likely to achieve the best result for the chosen KPI. It works within call distribution and its rules. Which KPIs suit the purpose is discussed in Measuring customer experience without vanity metrics.
See Univerx on your own example.
In a demo we show the platform along your channels and requests, from AI agents to video, workforce scheduling and analytics.
Online, in German or English, with no preparation on your side.











