Churn prediction · 04Borrador

Predict churn before it shows up in the dashboard.

Models trained on telco signals — usage, billing, network experience and sentiment — score every subscriber daily, rank the drivers behind each risk, and trigger the right prevention playbook automatically.

Agenda una demo
Model scoring·Retention desk·WhatsApp·Outbound voice
El problema

Most churn is predictable weeks out — the signals are just never connected.

Usage decay, billing friction, network complaints and falling sentiment each sit in a different system. Read alone, none of them look urgent. Scored together, daily, they name the subscribers about to leave — with enough lead time to actually change the outcome.

Agente de texto

Lo resuelve en el mismo chat.

En WhatsApp, SMS y chat web autentica al suscriptor, resuelve la solicitud y envía el enlace — con contexto completo, en segundos.

Inerxia · Risk radar · daily runonline · replies in seconds
Morning run complete. 1,204 lines crossed the risk threshold overnight — 312 are new since yesterday.
07:15
Top driver this cohort: network quality drop in zone 04. Playbook: CPE check + goodwill credit.
07:15
Trigger the playbook for the high-risk group.
07:18
Done. 312 interventions queued across voice and WhatsApp, margin-capped.
07:18
Message
Playbook triggered · 312 queued
Agente de voz

Y contesta el teléfono.

Llamadas entrantes y salientes naturales que identifican la cuenta y ejecutan la acción en vivo — sin laberinto de IVR, sin música de espera.

Camila S. #77,410risk 0.83 · flagged 19d early · LIVE 01:20
Live
  • InerxiaHi Camila, this is Inerxia. We noticed a few dropped connections on your line this week.
  • CallerYeah, it's been unreliable in the evenings.
  • InerxiaI've scheduled a CPE swap for tomorrow and added a goodwill credit for the trouble.
  • CallerAppreciate that, thank you.
Root cause fixed · risk 0.83 → 0.21
Disparadores

Qué activa al agente.

Las señales se puntúan en tiempo real; la acción correcta se dispara automáticamente, dentro de tus límites.

Churn model · daily scoring
SignalsUsageBillingNetworkSentiment
0.91ranking AUC
ActionsScoreExplain driversTrigger playbook
128,402 lines scored/day · retrained weekly
lines scored daily
128K
median lead time before port-out
19d
ranking AUC
0.91
Inerxia

Míralo funcionando con tus propios datos.

Agenda una demo de 30 minutos y observa a Churn prediction resolver casos reales en vivo.