Fixing the missed pickup at the call, not the curb

This entry describes a five-day kaizen held in June 2021 at a residential waste-collection company serving a large city on the U.S. West Coast, covering the process by which a missed residential pickup becomes a ticket and a ticket becomes a serviced customer — from the customer’s call or website report through to the truck that returns. The company is not named here, and identifying details — the city, the operation, and the vendor software — have been generalized; individual names are omitted and the customer-service, dispatch, and operations staff who redesigned the work are credited by role. The operational detail is otherwise as the team recorded it during the event.

When one of the company’s trucks passes a house without emptying the bin, the customer calls or fills out a form on the website, a customer-service representative opens a missed-pickup ticket, and the operation either works the recovery into the next day’s route or, if the customer needs it sooner, sends someone back the same day on overtime. Two departments touch every miss — the call center that takes the report and the operations group that runs the trucks — and the ticket is the object they hand between them.

By the spring of 2021 the operation was fielding on the order of 100 calls a day about misses, and had been for six months. The number carried more weight than its size — a missed pickup is the one moment a reliable service becomes visible to a customer, and a handful of them does disproportionate damage to the company’s standing. Leadership’s charter was plain: reduce the misses. Alongside it sat a second grievance the representatives heard daily — the customer expected to be told whenever a pickup was missed, and was told only about a quarter of the time. The presenting problem, as it arrived, was an operations problem: the trucks were missing houses, and the routes needed to miss fewer.

The team spent the first three days mapping the two halves of the process on the wall — the customer-service side, from the incoming call to the written ticket, and the operations side, from the route-sequencing a supervisor does each week to the key process that governs gated and locked-access pickups. Against the maps they laid two weeks of ticket data, sorted by what had actually happened.

Of 152 missed-pickup tickets in the sample, 55 were genuine misses — a truck that should have serviced a house and did not. The other 97 were something else. Forty-six were customer-service error: a pickup the driver had in fact completed, re-declared as a miss because the representative had no quick way to check the driver’s record, or a second and third ticket opened on a miss already logged. Fifty-one were customer fails — a bin set out after the truck had passed, or a call placed after the route’s service window had closed. Fewer than four in ten of the “misses” were misses.

The noisy count was a symptom, and the two maps set side by side showed of what. The department that took the call and the department that ran the trucks had no agreement between them about what a missed pickup was owed. The call-taker, with nothing firm from operations to rely on, promised whatever seemed to settle the customer — sometimes a same-day return, sometimes nothing at all. Operations, for its part, was handed a stream of tickets of every urgency and no standard, and absorbed them as it could. A miss was declared while drivers were still out and the day’s service was still unconfirmed — born, in other words, into the empty space between the two departments.

That was the pivot. No amount of pressure on the drivers could move the two-thirds of the number that had nothing to do with them; the constraint was not the trucks, and it was not the phones, but the interface that had never been drawn between them.

The customer-service and operations current-state maps, built on the wall during the kaizen. — **The current-state maps.**·Photograph: house archive·June 2021

The redesign drew that interface. The two departments agreed one plain service level: every missed pickup called in would be serviced the next day — reliably, and no sooner. The call-taker would never again promise a same-day return.

That single commitment did the work in both directions. It gave the call center a promise it could keep — a customer told exactly what would happen and when, in place of a hopeful guess — and it gave operations a target both predictable and easy to hit: the next day’s routes already passed most of the missed houses, so a recovery was a stop added to a run already being driven, no overtime and no special trip. The variability that had made the service feel unreliable was designed out, not worked harder.

The quiet finding inside this was that customers did not mind. Told plainly that their bin would be emptied the following day, as a commitment they could count on, they did not push back for the same-day service they had never really been promised. What had frustrated them was not the wait; it was the not knowing. A predictable tomorrow beat an uncertain today.

The next-day default also dissolved a whole apparatus the operation had built around disputing misses. Drivers had been recording the reason a bin went unemptied and photographing the empty curb — assembling, house by house, the case that a miss was the customer’s fault, all to justify skipping an expensive same-day return. When the team walked the process into the call center, it found the case was never made: the representatives were helpful, and whatever the driver had logged, they simply arranged the pickup. The evidence was collected and never once consulted. Once the recovery happened the next day by default, the costly same-day trip the documentation had guarded against was gone too — and with it any reason to prove fault at all. The curb photographs, the reason logs, the pushback they were meant to arm, the stress that pushback put on the representative: all of it came out, and several steps of the process disappeared with it.

What stayed was the data the operation actually used. Before a ticket was declared a miss, the representative would check the driver’s own record — not to argue fault with the customer, but to confirm the pickup had not in fact already been made, so the operation stopped opening duplicate and false tickets on work already done. The one thing in the way was named honestly: the representatives’ login to that record was too slow to use inside a live call, and the team wrote fixing it onto the plan as the dependency it was. Each genuine miss still carried a simple cause — a key or access problem, a relief-driver miss, an obstruction — but kept now for the operation’s own learning, to tell the misses it could eliminate from the ones it could not, rather than to build a case against a customer.

Most of what carried the redesign was small, and the team sorted it deliberately by effort and impact before committing to any of it. The low-effort, high-impact moves went first: the new call script and every representative trained on it inside 30 days, the website’s “report a problem” menu rebuilt so a customer picks the right commodity and location, and a revised standard operating procedure requiring a comment in the routing system on every missed ticket. The heavier items — fixing the routing-system login for good, and investigating in-cab cameras to document obstructions and contamination — were named as high-effort work and parked as dependencies, not pretended away. On the operations side the supervisors mapped and standardized the key process and the weekly route-sequencing, and committed to closing key tickets in full. Each task left the week with a named owner and a June date, and the measures that would tell them whether it held were defined and assigned the same way.

Because the week closed with the changes barely in place, the proof at report-out was one leading indicator already moving and a system built to move the rest. The after-service survey — sent to every serviced customer three days later — had risen from 81.7% in April to 88.5% in May, against the 91% the team set as its 90-day mark. The same-day overtime the operation had been paying, about $40 for every 30 minutes, was designed out rather than trimmed: the new promise never called for it. The number the operation had been chasing was, for the first time, a number that meant one thing.

What the week produced, as the team recorded it at the report-out:

A promise the routes could keep: A next-day service level agreement between the call center and operations: the representative never promises same-day; the recovery is added to a route already running the following day — a reliable promise the call-taker can make and a predictable target operations can hit at negligible marginal cost, with none of the same-day overtime (about $40 per 30 minutes) the old ad-hoc promises had triggered.
A dispute apparatus removed: The driver-side fault documentation — reason logs and curb photographs built to prove a miss was the customer’s fault and justify skipping a costly same-day trip — was scrapped: the call center had never used it, and the next-day default made the trip it guarded against unnecessary. Several process steps went with it.
After-service satisfaction: April 81.7% → May 88.5%, already moving toward the 91% 90-day target.
The count, read true: Of 152 missed-pickup tickets in the sample, only 55 were genuine misses; 46 were customer-service error and 51 were customer fails — and a quick check of the driver’s completion record before declaring a miss cut the duplicates it caused, targeted from five a day to two.
Sustaining: A daily missed/supervisor-ticket log under five a day, a ticket-life query targeting under 36 hours, and route-sequencing counts — each with a named owner.

The visible work of the week was a call script and a ticket with a reason on it. The structural move was quieter, and it was not on the trucks or in the call center. The two departments that between them owned a missed pickup — the one that took the call, the one that ran the route — had never agreed what the service was. Each had been doing its best against the other’s silence: the call-taker promising what it could not guarantee, the trucks absorbing whatever arrived. The overtime, the variability, the count no one trusted, the customer left in the dark — all of it lived in that undrawn interface. The intervention was not to push harder at either end but to write the agreement between them: one reliable next-day promise, modest enough that the routes could keep it and clear enough that the customer could rely on it. That single commitment also emptied the process of everything that had grown up around disputing a miss — the curb photographs, the fault logs no one read — because a service you simply provide the next day needs no case built against the customer you provide it to. It turned out the customer had wanted the certainty more than the speed all along.

· · ·