Closed-loop feedback management that turns every unhappy customer vote into a callback promise on a 48-hour clock, and measures whether the customer comes back happy
No unhappy customer left without a call back, and every recovery measured.
See one case, screen by screen ↓Closed-loop feedback management means following up every unhappy customer one by one, not just reporting the score. Lumina Utilities asks every customer to rate the contact on a happiness meter — by app, website, kiosk or an SMS after the call. About 1,400 votes a month come back Unhappy or Neutral, and behind many of them a care agent has already promised: “Someone will call you back within 48 hours.” Those promises live in call notes and spreadsheets, so nobody can say how many were kept.
The survey tool reports the score and the CRM holds the case, but nobody holds the promise: who said what, by when, whether the call was made and whether the customer came back happy. Detractors chase the utility with a second and third call, each one costing money and goodwill, and the frontline teams never learn which callbacks they could have avoided by fixing it on the first call. Measured from call notes for July and August 2026, Lumina made 58% of promised callbacks inside 48 hours, and nobody asked the customer again.
Estimated split of an officer's working time on one promise, in hours, by hand and with the solution. The clock that matters to the customer is the 48 hours.
Six agents read each unhappy vote, prepare the fix and watch every 48-hour clock; the recovery officer calls and closes the loop.
Omar Rahimi lives at 14 Cedar Lane, Harbour Point, and takes electricity and water from Lumina Utilities. His September electricity bill is $412.59 — the third estimated bill in a row. Here is what happens next, screen by screen, in the working solution.
Omar calls about a $412.59 bill for a two-bedroom flat. Tariq sees that July, August and September were all estimates, and that a reading was submitted but never applied. He raises it with billing and makes the promise every contact centre makes. The 8:52 call is kept with its transcript, the key lines marked.
Omar: “I sent my reading in the app on the twenty-eighth, with a photo.” · Tariq: “Someone will call you back within 48 hours.”
Omar votes Unhappy by SMS link: “Third estimated bill in a row… Nobody listens.” Staff professionalism Happy; speed of service and information quality Unhappy. The vote reader joins the vote to Tariq's call by its link ID, and the promise keeper opens a promise due Tue 13 Oct, 09:14, owned by Hana Qasim on billing and power. On Tuesday morning the votes screen shows it has read 412 votes since 07:00, each in about 4 seconds.
Set aside, with a reason: “Same customer, same remark, 3 minutes earlier · joined to the promise above”
At shift start the keeper spreads 27 promises due today across 7 officers, nobody over 30. At 07:02 Leila Hamdan's promise broke at hour 48 because her owner, Sami Okafor, is off sick, so it passed to the recovery lead, Rania Aziz. Then it proposes three moves: Ivan Petrov to Aisha Noor, because Tomás Ruiz starts after the promise is due; Grace Mbeki's next try on the mobile added on 3 Oct; and Omar kept with Hana, whose brief and credit approval are ready.
“Ivan Petrov is due at 11:40, but Tomás Ruiz starts at 13:00 · Move to Aisha Noor”
Omar's photo reading of 50,212 kWh arrived on 28 Sep at 19:42, after the 26 Sep read window, so billing rule BR-4.2 held it and September went out on an estimate of 2,140 kWh. Actual use from 27 Jun is 2,824 kWh against 4,040 kWh billed on estimates. September rebilled is $178.15, a credit of $234.44. That is above Hana's $150 authority, so it went to Daniel Osei, billing analyst, who approved it at 08:05.
“Credit above $150 sent for approval · Approved by Daniel Osei, billing analyst, at 08:05”
Hana's ledger opens on 34 open promises, sorted by time left: 2 broken at the top, then the amber rows, then the ones on track. Each row has its 48-hour rail from the vote to the deadline. At 08:14 Omar's row turns amber, and the keeper's run log notes that the brief and the approved credit are ready.
“Omar Rahimi turned amber at hour 47 · 1 hour left · Hana Qasim told; brief and approved credit are ready”
Hana opens with the first line of the brief and tells Omar what is already done: the reading found, September now $178.15 and $234.44 credited, the next read set to actual. The 4:10 call joins the journey as contact 3. She logs the outcome as resolved on the call, with the root cause tagged: submitted readings not applied — bills stay on estimates, 38 voices this month.
Omar: “Oh. You actually called.”
Hana closes the promise as kept at hour 47:02. The outcome recorder links the root cause to the Voice of Customer challenge, updates team 3's lesson and sends the re-rating SMS with the approved wording at 08:31, fifteen minutes after the call. At 08:44 Omar re-rates Happy. The customer's own rating sits beside the solution's reading of the remark, never instead of it.
Omar, re-rated Happy: “Thank you for actually calling back. Bill fixed.”
The lessons screen shows unhappy and neutral votes by team and cause; team 3's estimated bills stand out at 82. Of 1,402 promises, 211 could have been fixed on the first call. The coaching note for Marisol Vega's team says 38 promises were for a submitted reading never applied, and in 31 the care agent could have applied it on the call.
“Customers re-rated Happy after the fix 72% of the time — the fix works; it just comes two days late.”
Each cause has a playbook: what to check, the remedy, who may give it and what to say. Rania owns version 6. Testing it on 5–11 Oct replays 331 closed promises: the same remedy as the officer gave in 309 (93%), no credit above authority proposed, and 6 outage-notice cases with no remedy step — a suggestion sent to Rania.
PB-01: “We found your reading from the 28th. Your bill is now $178.15 and the difference is back on your account.”
From 14 Sep to 13 Oct: 11,860 votes, 1,436 Unhappy or Neutral, 1,402 promises opened, 1,356 called back, 812 re-rated and 498 Happy. 97.0% of closed promises were kept inside 48 hours (1,327 of 1,368), against 58% before the ledger went live on 7 Sep. Re-rated Happy by cause runs from 72% for estimated bills to 50% for outages and notices.
“498 customers said they were happy after the callback — a number the survey report never had.”
The same unhappy vote, seen by the five people who own it — what their week looked like, and what it looks like now.
Six agents read each unhappy vote, prepare the fix and watch every 48-hour clock; the recovery officer calls and closes the loop.
Reads every happiness-meter vote and remark, decides whether it needs a callback — Unhappy always, Neutral with a remark or a second contact in 30 days — and joins it to the call, chat or email it rates by its link ID.
Joins the contact, the vote and the account's last 90 days — bills, readings, payments, work orders and outages at the premise — into one journey, in time order.
Opens each promise due 48 hours after the vote, assigns an owner by skill and shift, warns the owner at hour 40, turns the row amber at hour 47 and passes it to the recovery lead at hour 48, logging every action on the ledger.
Reads the journey, finds the cause in the account data, picks the recovery playbook, works out any rebill or credit and writes three lines: what went wrong, what is ready, what to say first.
Logs the outcome, the remedy and the root cause, links the case to the Voice of Customer challenge, sends the re-rating SMS 15 minutes after the call and records the answer; a re-rating of Unhappy opens a second promise to the lead.
Each week groups a team's promises by cause, finds the ones the care agent could have fixed on the call using the playbooks, and drafts a short note for the team's manager with counts and one huddle topic.
Makes every call; a billing analyst approves credits above $150. The agents propose; a named person decides.
Ask about a promise, a customer, a team or the month. Every answer cites its source.
Which promises break in the next hour?
Five, all amber: Priya Nair (Ravi Menon, due 08:41), Kofi Asante (Lina Park, due 08:52), Marta Novak (Yousef Karimi, due 09:03), Ahmed Rashid (Chloe Martin, due 09:09) and Omar Rahimi (Hana Qasim, due 09:14). Every brief is ready. Two more have already broken and sit with Bilal Haidari and Rania Aziz. I have filtered the ledger to amber.
Why did Omar Rahimi get an estimated bill?
He sent a photo reading of 50,212 kWh on 28 Sep at 19:42, but it arrived after the 26 Sep read window, so rule BR-4.2 held it for validation and September went out on an estimate of 2,140 kWh — the third in a row. On the actual reading September is $178.15; Daniel Osei approved the $234.44 credit at 08:05.
Which team promised callbacks it could have fixed on the first call?
Team 2 had the most (41 reconnections that should have followed a payment), then team 3: 38 submitted readings that were never applied, which is Omar Rahimi's case. In 31 of team 3's, the care agent could have applied the reading on the call. The coaching note for Marisol Vega is drafted.
How many customers did we win back this month?
498 of the 812 customers who answered the re-rating said Happy after the callback (61%), 201 Neutral and 113 Unhappy. 97.0% of closed promises were kept inside 48 hours, against 58% before the ledger.
11 screens from the working solution, on its sample data. Pick one to see it large.
Which votes open a promise, how long the clock runs, who may give what and how the customer is asked again are settings, versioned and approved by the recovery lead.
| Setting | Default | Choose from |
|---|---|---|
| What opens a promise | Unhappy always · Neutral with a remark or a second contact in 30 days | Each rule On · Off; a Happy vote gets a thank-you message only |
| Set aside | 3 rules | Empty remark · staff test accounts · duplicates within 10 minutes |
| Callback window | 48 hours from the vote | Warn the owner at hour 40 · amber on the ledger at hour 47 · red at hour 48, when it passes to the lead |
| Quiet hours | 21:00–07:30 | No calls; the clock keeps running |
| Attempts before a letter | 3 | At least 2 hours apart; then a letter and SMS, and closed as unreachable (On · Off) |
| Authority | Officer up to $150 | Billing analyst above $150, before the call — Daniel Osei and 3 others · goodwill payment of any amount, Rania Aziz |
| Re-rating | SMS 15 minutes after a made callback | Approved wording · one reminder after 24 hours · if re-rated Unhappy, open a second promise to the lead (On · Off) |
| Assignment | By skill, at most 30 open per officer | Billing · water · power · digital · connections; no promise due while its owner is off shift — the keeper proposes a move first |
Every figure is labelled: a target the solution is built to, an estimate, a typical published result, or a proven one.
“demo” = seen in the working solution, on its sample data · “target” = the design goal, measured in the live solution · “estimated” = our estimate: 1,402 promises a month × 12 = 16,824 a year × 39 points more made on time (58% → 97%) × 2.2 repeat contacts each × $9.54 a call ≈ 14,400 calls and $138K a year; the hours split is an estimate too · “typical” = published figures (E Source: $9.54 average cost per utility call, more than $2.3M a year on repeat contacts at large utilities, and, at one utility, 90%+ of customers resolved first time satisfied against 59% of those who had to call back). Lumina Utilities, its people, customers, accounts and amounts named on this page are fictional — characters and sample data in the working solution.
It is following up each unhappy or neutral customer one by one — calling them back, fixing what went wrong and asking them again — instead of only reporting the score. Closed-Loop Feedback Management turns every such happiness-meter vote into a promise on a ledger: promised, due, made, outcome and re-rated.
The vote reader reads every vote and joins it to the call it rates. The promise keeper opens a 48-hour promise with an owner by skill and shift, and the recovery preparer finds the cause in the bills, readings and work orders and prepares the remedy and the words. The customer recovery officer makes the call, and 15 minutes later the customer is asked to rate the utility again.
It works after the survey. It takes the happiness-meter or survey votes you already collect from the app, website, kiosk and SMS, and holds what survey tools and CRMs leave out: the promise, its clock, the callback and whether the customer came back happy.
Every promise has a visible clock. The owner is warned at hour 40, the row turns amber at hour 47 and at hour 48 it passes to the recovery lead. No promise falls due while its owner is off shift — the keeper proposes a move first; in the sample month, 97.0% of closed promises were kept inside 48 hours, against 58% before.
The inner loop is the case-by-case follow-up with each customer; the outer loop is fixing the causes found across many of them. The ledger and the journey are the inner loop. Root-cause tags, the lessons for each frontline team and the outcomes by cause are the outer loop.
The recovery preparer works out the rebill or credit from the account and the playbook. An officer may give up to $150; anything above goes to a billing analyst for approval before the call, and the approval is shown on the case. In the sample case, a $234.44 credit was approved at 08:05 and the call made at 08:16.
Yes. The agents never call a customer and never change account data; they prepare and propose. The officer decides what is said and closes the promise, a billing analyst approves credits above $150, and the recovery lead approves every change to the clock, authority and playbooks.
The Agentic Solution Engine builds and deploys it from your requirements and documents — your callback rules, authority limits, recovery playbooks and a sample of past votes and call notes — and it goes live once every quality gate has passed. We will walk you through it on your own unhappy votes first.
Every voice read, and each recurring problem named with an owner and a fix.
1-min demoComplaint Root Cause AnalysisEach complaint trend proved with system data, owned, and fixed at the cause.
1-min demoCustomer Satisfaction Index ReportingEvery move in the score explained by name, with the customers’ voices behind each line.
1-min demoWe'll run Closed-Loop Feedback Management on a sample of your own survey responses and call notes.