Improvement in retention among customers actively porting out, against the prior benchmark.
Solutions · Customer retention
Catch the leaving, in the moment it happens.
A customer does not churn on a spreadsheet at quarter-end. They churn the minute they start a port-out, a cancellation, or let a renewal slide. The AI reaches them in that window, in their language, hears the real reason, and either saves them on the call or hands the ones that need a person to a specialist.
At-risk book · liveSimulation
at-risk reached0
saved on call0
to a specialist0
saved by AIto a retention specialist
"Retention is not a campaign you run once a quarter. It is a conversation you have in the sixty seconds a customer decides they are done."The retention doctrine · the save is a moment, not a mailer
The operator's truth
You cannot save a customer with a discount email. You save them in the conversation.
The moment of leaving is short, it is emotional, and it almost never reaches a human in time. Calling everyone by hand is too slow and too costly, so most at-risk customers are only noticed once they are already gone.
The AI works that moment at scale. It reaches the at-risk customer while the intent is fresh, separates a price complaint from a service complaint, answers the real one, and pulls in a person the instant the case needs judgement or a bigger offer.
The save sequence
From the signal to leave, to the reason to stay.
Four moves, run in the window that decides it. The AI owns the first three. The cases that need a human ear or a bigger lever move to a specialist, already briefed.
Move 1 · The signal
Notice the intent to leave, the instant it forms
A port-out request, a cancellation started, a renewal date approaching, a usage drop. The trigger fires the outreach automatically, so the customer hears from you inside the window, not a week after they have gone.
Trigger-based outreachNear real-time dialIn the customer's language
Moves: at-risk contact rate
Move 2 · The reason
Hear why, not just that
Price, service, a competitor's offer, or a grievance. The AI separates the real driver from the surface complaint, because a customer leaving over network quality needs a different answer from one leaving over money.
Reason detectionSentiment readGrievance capture
Moves: reason-for-churn clarity
Move 3 · The right offer
Answer the reason, on the call
When it is price, a tailored offer within your rules. When it is service, a guided route to a fix or a self-help channel. The lever matches the reason, so the save is genuine rather than a blanket discount that trains customers to threaten to leave.
Rule-bound offersGuided resolutionRetention nudges
Moves: save rate, margin held
Move 4 · The person
The high-value save, handed to a specialist
A marquee account, a distressed customer, or a save that needs authority beyond the AI's remit routes to a retention specialist mid-call, with the reason and history attached, so they open on the offer, not the interrogation.
Warm handoffFull contextOffer authority
AI → human, in seconds, with context
Moves: high-value retention
Beyond any one industry
Wherever a customer can walk away.
The save motion started in telecom, where the leaving moment is a visible port-out. The same brain now runs it wherever churn hides a moment of intent, and gets sharper the more books it works.
Telecom
The port-out, intercepted
Subscribers who have begun switching numbers reached mid-process, the reason heard, and a matched offer made before the port completes.
Moves: port-out save rateSubscriptions & OTT
The cancel click, answered
Customers hitting cancel or lapsing on payment re-engaged at once with a pause, a downgrade, or the right incentive, instead of a silent churn.
Moves: involuntary + voluntary churnInsurance
The lapse, caught early
Policyholders drifting toward non-renewal engaged before the grace period closes, the barrier surfaced, and the policy kept in force.
Moves: lapse rateLending & cards
The account going quiet
Attrition signals, from a foreclosure enquiry to a dormant card, worked with a retention conversation before the balance leaves for a competitor.
Moves: balance attritionD2C & commerce
The repeat buyer slipping
Customers past their usual reorder window re-engaged with a timely, personal nudge, turning a one-time buyer back into a repeat one.
Moves: repeat-purchase rateMemberships & broadband
The renewal on the fence
Members and connections approaching a decision reached in their language, the friction resolved, and the relationship renewed on the call.
Moves: renewal rateThe numbers this motion is judged on
Reached in the window, and kept.
Vi (Vodafone Idea) ran this on customers who had already begun porting out, across seven circles, in Hindi. The numbers below are that programme.
Retention uplift4.1x
Contact rate58%
Of port-out customers reached and engaged by the voicebot, after redials.
Depth of engagement66%
Of callers stayed on past 15 seconds, the point where a real conversation begins.
Cost to engageA fraction
Of the human outbound cost, which is what made reaching this segment viable at all.
What retention leaders ask before signing
How does the AI know a customer is about to leave?
It listens to your systems for the moment of intent. A port-out request, a cancellation started, a renewal date, a usage or payment drop; each is a trigger you configure, and each fires the outreach automatically so the customer is reached inside the window that decides the save.
Will it just discount everyone and destroy my margin?
No. Offers are bound to your rules and matched to the reason. A service complaint is routed to a fix, not a discount; price objections get the specific lever you allow. The point is a genuine save, not training customers to threaten to leave.
What happens with high-value or distressed customers?
They reach a person. Marquee accounts and emotionally charged cases route to a retention specialist mid-call, with the reason and history attached, so the human opens on the offer rather than re-asking what happened.
Does this work outside telecom?
Yes. The save motion is the same wherever churn hides a moment of intent, from an insurance lapse to a cancelled subscription to a dormant card. The triggers change; the sequence does not.
How do you price this?
Consumption on handled volume, plus a success fee on customers saved above your baseline. No seats, no shelfware; the full construct is on the outcome pricing page.
What proof exists on books like ours?
Ask for the reference call. Vi (Vodafone Idea) lifted retention 4.1x on customers who had already begun switching away. Bring your churn triggers and book, and we will baseline the save rate before we start.
Judge it on your book