A wealth-management firm · Financial services · Conversational AI
45,000 clients had quietly stopped investing. The fix wasn't a campaign — it was a conversation that knew exactly when to hand over to a human.
Client identities withheld by design. Engagements delivered as lead across previous consulting roles; figures are each engagement's findings and committed design targets.
The situation
A regional market downturn, product gaps, and uneven digital service had left an investment house with tens of thousands of dormant accounts — clients who hadn't left, but had quietly stopped. Mass email had already failed them once. The diagnosis reframed the problem: dormancy isn't one condition, it's several different stories — a broken auto-payment, a season of personal financial strain, a head turned by a competitor's quarter — and each story needs a different conversation, not a louder campaign.
What the diagnosis found
Three dormancies, not one
Interaction analysis split the dormant base into administrative lapses (an auto-pay that silently failed), emotional withdrawal (financial strain nobody had asked about), and competitive drift (a rival's returns looked better last quarter). Each demanded its own tone, depth, and endpoint.
The trust boundary is the design
Clients accept AI that is transparent about being AI, warm without pretending to feel, and quick to fetch a human when the conversation turns emotional. The escalation moment isn't a failure of automation — it is the product working.
The advisors needed convincing too
Algorithm avoidance is real: people prefer human judgment even when the model is right. Adoption was designed for — advisors review and shape AI outputs, keeping ownership — rather than assumed.
The redesign
Two channels, one memory
Voice AI and WhatsApp running in parallel over a shared CRM and sentiment engine — every greeting personalized from the client's actual history, every reply parsed for tone as well as content.
A conversation per dormancy story
The admin lapse resolves inside the AI flow in minutes — permission, fix, confirmation. The strained client gets empathy and a no-pressure human check-in. The drifted client gets a straight comparison and an advisor call. Same engine, three journeys.
Handoff with full context
When escalation triggers, the advisor receives the whole conversation, the sentiment trail, and the suggested next step — so the human never asks the client to repeat themselves. The transition feels like a team, not a transfer.
A dashboard that tunes the machine
Response, drop-off, and conversion rates feed back into scripts, timing, and escalation thresholds continuously — the outreach gets more human with every conversation it has.
What the blueprint committed to
45,000
dormant clients addressable through the voice + WhatsApp engine
3
distinct dormancy stories, each with its own conversation design
100%
of emotional moments routed to a human advisor — by design, not by accident
AI handles scale and logic; humans handle nuance and trust. The handoff between them is the experience.
The first principles at work
Recognize your own journey in this?
Tell Exie what feels similar — the same diagnostic method maps your version of it. Free, and you keep the map.
Next case · A premium outdoor-gear brandRetail that follows the customer outdoors.