One year, 158,000 calls, and 333 qualified prospects handed live to a tax-resolution firm's closers.
A leading US tax-resolution firm started with one Kusp agent on past-due accounts in October 2025. Twelve months later, three outbound use cases and a Spanish inbound line run on Kusp, monthly call volume is up 63×, and qualified callers reach a licensed closer while they're still on the line.
Twelve months, one flagship account.
From pilot to production volume
Outbound dials per month, Oct 2025 to Sep 2026
Conversation → result, by use case
All-time, live conversations only (silent connects excluded)
Positive: a promise to pay, a scheduled callback, or active interest. Qualified: meets the firm's debt and situation criteria for a closer, from a list of cold, months-old enquiries.
Hardship goes to a human
Share of conversations escalated to a human specialist, last six weeks
The agent doesn't try to collect from someone in hardship. It stops, and a person picks up.
This wasn't headcount that needed hiring. It was revenue sitting in piles nobody worked.
The firm had three piles its team couldn't get to: the under-$500 delinquent bucket nobody prioritised, contracts that signed but never funded, and tens of thousands of past enquiries, leads already paid for in marketing, that had gone cold.
“We have a lot of humans that are taking on raw call volume. Pay for a bot to do it because the bot's gonna be more consistent. It's not going to get insulted, it's not going to have a bad day of the week.”
54,780 leads the firm had already paid for, worked again.
Liz, our re-engagement agent, calls people who once asked the firm for help and never signed. She screens them against the firm's criteria in plain conversation, and when someone qualifies she transfers them live to a closer. No callback queue, no lost afternoon.
Qualified callers transferred live, per week
Sep 1 to Oct 12, 2026 · *four dialing days
Volume up 7×, quality held.
Scaling a cold list usually dilutes it. Here the share of conversations that qualified stayed in the same band while weekly dials grew from 4,320 to 30,150.
Real cases, not curiosity.
The median caller handed to a closer reported about $13,000 to $16,000 of IRS debt, in their own words on the call.
A record week.
In the week of October 6 the agent averaged about 6,600 dials and 18 live transfers a day.
Every call, one click away.
What happens to a qualified caller
357 qualified callers · Sep 1 to Oct 15, 2026
Share of qualified callers transferred live
Dormant-lead re-engagement, by month, 2026 (October to the 15th)
Call explorer
Every call one click away: recording, transcript, and every extracted field
What's real here: the KPI strip, the qualified-caller split and the transfer trend are production figures from the firm's call records, October 2025 to October 15, 2026. Call identifiers are synthetic.
We don't ship and disappear. We read the calls, fix what they show, and prove it.
Every change is staged on a copy of the agent, regression-tested, promoted with a rollback point, and then watched on every production call by an automated monitor.
Transfer-first close
Qualified callers now go straight to a closer instead of into a callback queue. The share transferred live went from about one in three to nearly nine in ten.
The decline audit
We reviewed every caller the agent had turned away and found about 11% of qualified volume declined at the boundary. A decision gate now catches those calls before they're lost.
Wider qualification, safely
At the firm's request, the debt threshold moved from $5,000 to $4,000. The change passed its full regression suite before launch and shipped with a live monitor watching for misfires.
Recording notice, fleet-wide
Every agent now opens with a recording notice, applied to inbound and outbound together so the fleet never drifts apart.
We earn the next use case. We don't big-bang.
- October 2025
Past-due recovery goes live
One agent, one use case: the delinquent accounts nobody prioritised. Dashboard and twice-weekly ops reviews from day one.
- December 2025
Warm-lead recovery
Added once past-due had cleared its own ROI bar: contracts that signed but never funded.
- March–April 2026
The first proof point
7,860 dials, 880 conversations, 23% ending in forward motion. That became the first edition of this case study.
- May 2026
Spanish, around the clock
A 24/7 Spanish inbound line, pulled by the firm, not pushed by us.
- July 2026
Dormant-lead re-engagement
An outbound agent with inbound and callback twins, working the firm's back catalogue of enquiries.
- September–October 2026
Production scale
Around 60,000 dials a month, a transfer-first close, a decline audit, and a wider qualification band.
Benchmarked against the vendor they had before.
“It's reciting back the case IDs perfectly. Able to have a full-blown conversation with custom fields and just being able to quote to a customer, hey, you owe $212.14. Instead of us having to use fixed values, is going to be really cool for us.”
“It's very clear who the winner is in this one.”
Dials and conversations come from the firm's production call records, October 15, 2025 to October 15, 2026. A conversation is a connected call where someone spoke; silent connects don't count.
Live transfers are calls handed to a closer and confirmed by the telephony record. Costs are per-minute usage only and exclude the platform fee.
Debt figures are what callers reported on the call. We don't claim dollars collected or revenue closed: that happens in the firm's own systems after the handoff, and we only publish what we can see.