Case Study: AI/ML SOLUTIONS

Real-Time Speech Port Out to Reduce Churn Rate of Large Telecommunications Provider

One of the largest telecommunications provider partnered with Spearhead to use AL/ML solutions with NLP deep learning to figure out why the client's current churn prediction models kept being missed and why customer's who were porting out were labeled as low-risk.

Challenges

Client relies on state-of-the-art churn prediction models to keep its customer’s voluntary churn rates on account and line levels below 1.2%. The existing churn models flag customers as churn risk for retention offers and proactive reach out.
However, customers are still churning via porting out to competitors, despite being labeled as low-risk by the churn prediction models resulting in missing the Client’s target churn rates.

Solutions Offered

From detailed analysis, Spearhead discovered that most of the port-out to competitors is driven by customers calling and talking with agents. Still, there are no indications in their conversions to being dissatisfied, words of disconnect or the like, or mentioning of competitors/offer but yet ported out their phone services within an average of 12 days after the calls. Those customers are flying under the churn models risk radar.
Spearhead AI/ML solution involved processing conversation transcripts using NLP deep learning and other ML techniques to identify callers with a high probability to disconnect from words that are not indicative of disconnect or otherwise dissatisfaction, for example, whether on pricing or service quality.
The solution successfully identified this cohort risk 83% of the time and reduced the Client’s overall churn rate to its desired level.