Lone Star’s predictive modeling helped AT&T optimize their converged IP service delivery platform, cutting system costs and risks by 40x through accurate demand forecasting and capacity planning.
The Issue
AT&T had committed to a converged IP service delivery platform (SDP) for its U-verse offering.
The approach was both new and untested at scale.
AT&T did not understand “when, why, where, or how” the system would break.
Due to uncertainty, AT&T was installing four times the nominal estimate for the required systems.
The risk was unacceptable, and so was the massive overspend on CAPEX.
Multiple AT&T departments needed a common framework and understanding.
The Solution
Predictive Model 1 for Demand: Lone Star built a model predicting demand in the face of marketing/rollout uncertainty and complexity.
Predictive Model 2 for Demand Impact: A second model predicted system interface impacts based on demand. Peak hour and peak day were predicted over a 9-month horizon. This predicted SDP infrastructure and CAPEX requirements.
The Results
Accurately Predicted AT&T Future: Nine months in advance, we accurately predicted system demand within 10%.
40x Improvement Achieved: Enabled AT&T to dramatically reduce systems cost and risk – a 40x improvement!
