SLA Prediction Model

Service Level Agreements (SLA) with the customer is common in most of the service operations. SLA’s might vary depending on the business need and many a times the Non Compliance to SLA’s might lead to huge penalties as well as customer dissatisfaction.

This topic provides an approach for predicting the SLA compliance in advance and helps in effectively managing the SLA to avoid noncompliance as well as penalties. This is just an approach the model can be adjusted based on the business rules, past trends, SLA compliance parameters etc.

Scenario: The team resolves the incident tickets received from the customer/automated tools. The project has a committed SLA with the customer on the “Time to Resolve the Issue”. And this Time to Resolve SLA is different for different severity of the tickets. Each issue reported is associated with a knowledge article that is been referred by the analyst to resolve the issue.

Model Approach:

·        Understand the SLA commitments

o   Time To Resolve – Time it took to resolve the issue reported from the issue received i.e., Time To Resolve = Issue resolved time – Issue reported time

o   SLA for Time Resolve

§  Severity 3 – 240 Minutes

·        Assess the parameters that might affect the SLA

o   Parameters that might impact the Time To Resolve includes

§  Severity

§  Issue type

§  Shift

§  Week Day/ Week End

§  KB Article

·        Analyze the issue categories

o   By Problem type

o   By KB article

·        Gather the data for the past 3-6 months (Depending on the volume and issues coverage)

o   Below is the  sample data captured for the Time To Resolve SLA

·        Understand the Time To Resolve distribution with respect to the above parameters as highlighted in second bullet

o   Time To Resolve

§  Distribution with respect to

·        Confirm the parameters that are impacting the Time To Resolve

o   Time To Resolve

§  Above table as observed there is a significant variation in the distribution from Morning shift to Night shift

§  Confirmed parameters include

·        Issue Type

·        Shift

·        Baseline the Time To Resolve distribution with respect to the above parameters

o   Time To Resolve – Baselines

·        Develop the Model to be referred/used for predicting the SLA compliance. This can be customized based on the business need