UPS Between a Workforce Exit and a Network Payoff
On July 28, UPS put two operating profits beside each other. Its second-quarter earnings release reported $930 million under generally accepted accounting principles and $2.1 billion after company adjustments. Inside the difference sat a physical network changing shape and a workforce exit that had already acquired a price.
UPS recorded an $891 million after-tax transformation charge, equal to $1.05 per diluted share. It said the charge consisted primarily of employee separation costs associated with workforce reductions under its completed Driver Choice Program. In the U.S. domestic business, reported operating profit was just $16 million. Adjusted operating profit was $1.188 billion.
CEO Carol Tomé described the same quarter as the completion of the company’s Amazon volume glide-down and related network reconfiguration. UPS had planned to reduce volume from its largest customer by more than half from 2024 levels by June 2026. It was also closing building operations, consolidating sorts, installing automation, reducing or redeploying capacity across vehicles and aircraft, and redesigning processes.
Those events often get compressed into a simple account of technology replacing workers. UPS’s disclosures connect the reset to several causes. Customer volume changed. Network capacity changed. Package-handling tasks changed. A union contract shaped the driver exit. Costs and expected benefits appeared on different accounting clocks.
Other companies will face the same reporting problem. An operator announces automation, a lower workforce, and a savings target in one presentation. Investors connect the three. Employees hear that their work has been engineered out. Managers receive new staffing targets while the equipment, package mix, and service requirements are still moving.
A credible workforce case needs a bridge between those statements. It should begin with demand, follow the work through facilities and tasks, record each worker path, and end with service and financial outcomes. UPS offers enough public detail to build that bridge. It also leaves enough unanswered to show where inference must stop.
Reported profit carried the transformation bill
UPS reported consolidated revenue of $22.8 billion for the second quarter. Revenue rose from the prior year, and the company raised its full-year revenue outlook to about $91.2 billion. The headline can therefore sound like a company moving cleanly from a restructuring period into growth.
UPS’s second-quarter release carried the bill. It reconciled $930 million in reported operating profit to $2.102 billion in adjusted operating profit. The $1.172 billion pre-tax adjustment was assigned to Network Reconfiguration and Efficiency Reimagined. After tax, the adjustment was $891 million.
Adjusted figures can help an operator compare recurring package economics across periods distorted by a large separation program. UPS said on the second-quarter call that it had made one-time Driver Choice payments, although it did not publish the remaining payment schedule. Payroll taxes and consulting fees still consume resources. A building can generate a sale gain while requiring a lease exit, equipment transfer, and local operating cutover.
Domestic results make the tension visible. The segment produced $14.93 billion in revenue, up 6 percent, while revenue per piece rose 9.3 percent. Reported operating margin was 0.1 percent. Adjusted margin was 8 percent. A manager evaluating daily package flow needs the adjusted view to see operations after the special charge. A board evaluating capital and labor commitments needs both.
UPS says transformation costs exceed ordinary work to improve profitability and are not expected to drive incremental revenue. Its initiatives change organizational layers, processes, technologies, and the business portfolio across multiple periods. That definition explains the adjustment. It also describes a management program whose boundaries can move as new buildings or workforce reductions are identified.
UPS reported about $1.2 billion of program benefits in the first half and expected about $3 billion for the full year. Its “benefits” label spans network consolidation and end-to-end process redesign. It is a separate measure from cash returned to shareholders, profit attributable to automation, or the present value of a driver exit. UPS also maintained an expected capital expenditure budget of about $3 billion for 2026.
Put the figures on one page and four separate measures appear: a reported expense, an adjusted operating result, a calculated program benefit, and capital spending. Each answers a different question. Combining them into one automation return would produce a number with no stable denominator.
A transformation owner should label every figure by period and accounting treatment. The $891 million after-tax charge belongs to the second-quarter income statement. The roughly $1.2 billion of benefits covers six months. The roughly $3 billion full-year benefit remains an expectation. Capital expenditures cover the entire company, rather than only this program.
That labeling also protects operating teams. If a finance deck treats the entire adjustment as a one-time past event, managers may receive an immediate run-rate target while local transfers and service work continue. If operations treats the charge as proof that savings have arrived, the company can miss a difficult period between an employee’s departure and a stable route design.
Amazon volume set the size of the network
UPS’s network reset began with packages. In its first-quarter Form 10-Q, the company said it expected to complete a reduction of more than 50 percent in volume from its largest customer, measured against 2024 levels, by June 2026. Tomé named the completed Amazon glide-down in the July earnings release.
A package network has large fixed commitments. Buildings, sort windows, trailers, aircraft, routes, and driver schedules are coordinated around density. Removing a large customer’s volume can leave a sort with less work while the conveyor, lease, and supervisory structure remain. Sending the remaining packages through fewer nodes can restore density, but the cutover changes travel, shift timing, and exception handling.
UPS described Network Reconfiguration as an expansion of its Network of the Future initiative. The filing connects planned customer-volume declines with reductions in facilities, vehicles, aircraft, and workforce. It also connects the program with automation and operational sort consolidation. These are interacting inputs, not interchangeable explanations.
During the first quarter, UPS closed operations in 23 leased and owned buildings. Twenty-two were permanently closed by March 31. The company had identified another 27 buildings for closure during 2026 and said further volume reviews could identify more.
On its second-quarter conference call, UPS updated the first-half total to 45 building closures. It also said the operational workforce was down nearly 30,000 positions from the first half of 2025. That comparison includes Driver Choice reductions but does not assign the entire change to the program.
Building counts supply a portfolio view. They omit jobs affected at each location, the portion transferred to another facility, and the service effect. A closed daily operation could move packages and employees elsewhere. A permanent facility closure can eliminate positions while creating work at a receiving hub. Geography determines whether a transfer is practical for a worker.
From the loading dock, a transfer is a commute and a new shift before it is a capacity number. A preloader may be able to follow work to another building and lose the arrangement that made an early shift possible. Another may stay at the receiving building and inherit unfamiliar package flow. UPS has not published those local outcomes, although each can surface later as absence, overtime, or turnover.
A larger base appears in UPS’s fourth-quarter 2025 results. The company said it had reduced its operational workforce by about 48,000 positions during 2025. That total included 15,000 fewer seasonal positions. It also closed daily operations at 93 leased and owned buildings.
A position count describes capacity rather than a count of people who lost jobs. Seasonal positions that were never filled, attrition, voluntary separations, transfers, and involuntary layoffs reach different people through different processes. A single transformation total obscures those paths.
Demand should come before the automation claim. Suppose a facility processed 100 units of work and one customer supplied 20. If that customer’s contribution falls by half, ten units disappear before any machine changes a task. Consolidating the remaining 90 units into another facility can reduce duplicate shifts. Automation might then change staffing within the receiving operation. Each step needs its own baseline.
This sequencing preserves the effect on workers while locating management responsibility more accurately. A job lost after a customer decision is still lost. Commercial concentration, contract strategy, network design, capital deployment, and workforce policy all contributed to the outcome.
It also changes the alternatives available. A volume decline can be answered with new sales, lower capacity, or a different service mix. A task automation can be answered with reassignment, lower staffing, shorter cycle time, or higher volume. A site closure raises commuting and transfer questions that no productivity model resolves.
For UPS, revenue per piece rose while volume mix changed. That supports a commercial story about higher-quality revenue while leaving route, hub, and shift stability unanswered. The network payoff arrives only when the remaining system can move those packages at the promised service level with sustainable labor.
RFID removes scans from the route
One part of the task change is unusually tangible. In April, UPS said its RFID sensing technology had reached all U.S. package delivery vehicles and facilities, as well as packages moving through more than 5,500 UPS Store locations. The company reported investing more than $100 million in the system.
RFID can identify a package without requiring the same manual scan at each point. UPS says its teams receive an alert when a package appears to be in the wrong delivery vehicle. Facility systems can create additional visibility as packages move through the network. UPS presents the technology as a way to reduce loading errors and improve customer information.
Matt Guffey, UPS’s chief commercial and strategy officer, described the rollout as a move from scanning to sensing. That phrase captures the customer promise: packages report their movement with less manual intervention. It also comes from the executive responsible for the strategy, so measured delivery performance must test the promise.
This is direct evidence of task automation. A manual scan can disappear. The release reports product coverage and company claims, without a controlled comparison of labor hours, headcount, injury rates, missed deliveries, or cost per piece before and after the rollout. Workforce attribution needs those additional measures.
Removing one scan may save seconds and attention across millions of packages. It can also move work into exception handling. Someone has to respond when the sensor and physical package disagree, replace damaged labels, manage device failures, and decide whether a truck should wait. The normal path gets shorter; the unusual path becomes more important.
That shift changes a supervisor’s work. Staffing plans should distinguish time removed from the normal flow from time added to exception queues, training, and system maintenance. A daily average can hide a burst of alerts near dispatch, when a few minutes have more service value than the same minutes in the middle of a sort.
Workers need the operating rule behind the alert. Does a driver unload and search the vehicle? Can a loader correct the issue before departure? Who records a false positive? If the system catches more errors because it sees more, an initial increase in recorded exceptions can indicate better detection instead of worse work.
At dispatch, a saved scan and a new alert have different time values. Five routine scans removed early in a shift may be welcome. One unresolved package warning when a vehicle is due to leave can hold the route or send the wrong parcel out. Averages should preserve that timing instead of treating every minute as exchangeable.
UPS also places AI in the network. In June, the company said it was combining RFID with AI-powered tracking and a real-time digital twin covering facilities, air and ground movement, and package flows. Its planning tools model volume, weather, and transportation delays. This establishes a deployed planning layer, while leaving the workforce share of any resulting savings unassigned.
Calling the reductions “AI layoffs” would attach an unsupported cause and obscure the demand change that set the network’s size. Workforce attribution requires evidence about where each system ran, which task changed, how long the change persisted, and what happened to labor and service in comparable operations.
A useful deployment record would pair each facility’s activation date with package volume, manual scan events, exception minutes, staffing hours, injury reports, on-time departures, and delivery errors. It would mark facility consolidations and customer-volume changes in the same timeline. Without those controls, an improvement after deployment may belong to a cleaner package mix or a smaller network.
Frontline workers can improve that record. Drivers and preloaders know which alerts prevent a miss and which ones arrive too late to help. Mechanics and technicians see device failures that an executive dashboard may classify as downtime without showing the operational workaround. Their evidence turns a product rollout into a work design.
Exception work can be more consequential and more stressful. Fewer repetitive scans can reduce strain, while tighter dispatch windows can increase it. The company should measure both before describing the remaining role as “higher value.”
7,500 applications moved through bargaining
Driver Choice moved through a different system. UPS offered the voluntary separation program to full-time U.S. drivers in the first quarter. Its May filing expected about $1.2 billion in separation costs and related payroll taxes during 2026, with most recognized in the second quarter. The Q2 release later assigned $1.1 billion of expected full-year adjusted expense exclusions to Driver Choice, revising the earlier program estimate.
Teamsters challenged how the offer was introduced. In its April 5 account of the settlement, the union said it had filed national grievances after UPS pursued the program without agreement. The union said those grievances caused the company to withdraw the offer in 13 states before negotiations produced national terms.
Teamsters General President Sean O’Brien framed the settlement as enforcement of the contract and seniority rights. UPS framed the agreement as a nationwide option that had drawn strong employee interest and would be approved according to seniority and business need. These accounts describe the same mechanism from different seats: one emphasizes a bargained constraint, the other a workforce option inside a network plan.
Under the settlement, drivers who accepted receive a $150,000 early-retirement payment. Offers covered long-haul feeder drivers and regular package-car drivers, followed seniority, and were capped at 7,500 across job classifications nationwide. UPS agreed to avoid another severance program during the current national master agreement, which expires July 31, 2028.
UPS issued its own statement confirming the agreement. The two accounts agree that an agreement was reached. The union’s account supplies the dispute, cap, payment, and contract protections from labor’s perspective.
That public ceiling became a company-reported acceptance count. On the first-quarter conference call, Tomé said applications exceeded available places and UPS accepted 7,500. On the second-quarter call, UPS said about 80 percent of participants departed during the quarter and that it had made the one-time Driver Choice payments.
Multiplying 7,500 by $150,000 produces $1.125 billion, close to the program expense discussed by UPS. The company’s accounting also includes related payroll taxes. Locations, center-level effects, and the positions removed after each departure remain undisclosed.
An eligible driver can decline a voluntary payment. The surrounding network may still close facilities, change routes, or reduce other positions. Accepting $150,000 can be rational for a worker near retirement and inadequate for someone with years of expected earnings, health coverage, and pension decisions ahead.
For an eligible driver, the choice lands inside a household calendar. Pension eligibility, family health coverage, debt, and the odds of finding another job at a similar wage can change the value of the offer. Public materials reviewed for this article include no account from a participant, so they reveal the employer’s national capacity decision more clearly than the worker’s calculation.
Nor is a voluntary program identical to an involuntary layoff. The choice, seniority process, negotiated cap, and contract restriction materially change the worker’s position. Reporting should preserve those distinctions while acknowledging that the employer designed the program to reduce the workforce.
Bargaining exposes a limit to unilateral operating design. Technology teams can install sensing equipment and network planners can model fewer sorts. A collective agreement determines how a driver exit can proceed. The settlement turned a rollout disputed in 13 states into nationwide written terms covering payment, seniority, and a 7,500-person cap.
Labor rights and contract terms belong inside the automation economics. A plan that assumes 7,500 immediate exits before bargaining has a faulty schedule even if its engineering assumptions are sound.
Through July 2028, another severance offer is unavailable as an unrestricted lever if volume falls again or a facility changes. Redeployment, attrition, scheduling, sales, and negotiated changes may carry more weight.
For remaining drivers, the program changes seniority lists, route bids, vacation coverage, and the distribution of experience. A national separation expense leaves local outcomes unanswered: whether a center lost the people who knew its difficult routes, or whether younger drivers gained stable assignments. Local operations must reconcile the national program with the work that remains.
The workforce case begins with demand
An executive team can connect these moving parts in a one-page workforce bridge. The artifact should sit beside the financial reconciliation, rather than inside a separate HR appendix. Each line needs a baseline, an observed change, an owner, and a date for review.
| Bridge field | UPS evidence available now | Missing operating evidence | Decision owner | Review trigger |
|---|---|---|---|---|
| Demand baseline | Planned reduction of more than 50% in largest-customer volume from 2024 levels; Amazon glide-down described as complete | Package volume by affected sort, route, day, and service level | Commercial and network planning | Volume differs from the approved plan or replacement demand arrives |
| Network capacity | 45 building operations closed in H1; Q1 filing identified 27 further 2026 closures at March 31 | Transfers, receiving-site utilization, route miles, aircraft and vehicle changes | COO and regional operations | Service misses, overtime, or capacity exceeds threshold |
| Task technology | RFID across U.S. vehicles and facilities; more than $100 million invested; manual scan removed in parts of flow | Activation dates, minutes removed, exception labor, false alerts, injury and error effects | Technology and operations | Exceptions or dispatch delay erase normal-path savings |
| Worker path | Driver Choice completed; negotiated $150,000 offer; UPS reported 7,500 accepted applications and about 80% departing in Q2 | Locations, remaining departure timing, positions eliminated, transfers, attrition, involuntary exits | HR, labor relations, and local operations | Staffing plan changes or a protected contract right is implicated |
| Financial timing | $891 million after-tax Q2 charge; about $1.2 billion H1 benefits; about $3 billion full-year benefits expected | Cash payment dates, recurring savings by source, implementation spend by program | CFO and transformation office | Benefit misses forecast or excluded cost recurs |
| Service and safety | Revenue per piece and segment margin disclosed | On-time performance, damage, misloads, miles, overtime, injuries, customer loss by affected operation | Operating leaders | Two review periods deteriorate after cutover |
| Worker capability | System changes create exception, maintenance, and transfer work | Training hours, certification, bid outcomes, vacancies, proficiency, retention | Frontline managers and workforce leaders | Exception queue grows or critical experience leaves |
Start with the demand row and preserve the forecast that approved the redesign. Actual volume can then reveal a commercial forecast miss without erasing the original assumption. At the next level, map each closed operation to its receiving site, equipment action, last sort, and employee path. A portfolio count such as 23 buildings will never show whether the work moved cleanly.
Review one affected route family each month. Commercial leaders bring planned and actual volume; the facility manager brings sort hours, missed departures, and overtime. Technology brings activation and exception data. Labor relations records the worker paths. Finance reconciles expense dates with recurring savings. Contradictory numbers are the work of the meeting.
Name tasks in units a worker recognizes. “Automate preload” is vague; “remove a handheld scan before vehicle loading and add response to a wrong-vehicle alert” can be timed. Keep worker paths mutually exclusive, so an accepted voluntary separation never reappears as attrition and an unfilled seasonal slot stays outside the laid-off employee count.
Customer experience supplies the hard stop. Volume per labor hour can improve while late departures rise, and cost per piece can fall while overtime migrates to another building. A customer lost during a rough cutover makes the remaining network look more efficient because fewer packages remain. Reopen the plan when service, exceptions, safety, or overtime crosses its trigger.
Clearer public categories can coexist with protected operating detail. Investors need to see which benefit has been realized and which remains expected. Employees need the local consequence: a departure, transfer, changed task, training requirement, or shift.
Exit costs hit the ledger ahead of route savings
UPS reported $1.8 billion in costs to date for Network Reconfiguration and Efficiency Reimagined as of June 30, including $1.2 billion incurred during 2026. It expected the initiatives to conclude by 2027. For full-year 2026, the company expected adjusted operating expense to exclude $1.3 billion to $1.5 billion of costs, mainly employee separation and third-party consulting fees. It assigned $1.1 billion of that range to Driver Choice.
Those disclosures make the current charge visible. They provide less detail about the composition of the $1.2 billion in first-half benefits. Network actions can reduce compensation, building expense, transportation, and purchased services. Process changes can prevent cost from growing. The same program can incur a consultant fee now and remove a recurring expense later.
A sound benefit ledger identifies the source, start date, and counterfactual for each item. Workforce savings should use actual departures and loaded compensation. Facility savings should begin after operations cease and include transfer costs. Productivity should use comparable volume and service, with technology support and exception labor included.
Avoid using the annual benefit target as proof for every local action. A region may exceed its labor plan because demand was lower than forecast, while another misses because a receiving building needed overtime. The total can meet plan while the operating thesis is wrong in both places.
Adjusted operating profit is also a decision view, rather than a cash ledger. UPS explicitly says its adjusted measures should supplement GAAP and may not be comparable with similarly named measures at other companies. A board should keep the reconciliation visible when deciding whether the next network investment earned funding.
Worker investments need their own line because exit is only one response to task change. Walmart’s 2026 Jobs Spotlight supplies a disclosure contrast from another physical network. Walmart said it had met its goal of helping associates move into 100,000 in-demand roles ahead of schedule, with more than half of those associates advancing through promotions.
Walmart expects the total to reach 200,000 by the end of year three and reported more than 1,000 associate-to-driver transitions. The two companies face different demand, and the Walmart figures are self-reported. Its publication still demonstrates a choice to put internal worker paths beside an automation story.
UPS’s public disclosures make the separation pathway unusually concrete. The next layer would show transfers, training, technician demand, and roles created at receiving facilities. That information would help employees evaluate options and help investors distinguish a shrinking denominator from a more capable network.
One transformation can contain both outcomes. Some positions disappear because volume and capacity fall. Some tasks disappear inside jobs that remain. New exception, maintenance, analytics, and planning work may appear. Count each change before aggregating it into a workforce reduction or a savings target.
Dispatch after the glide-down
Service and recurring cash performance will establish the network payoff. A receiving building must handle transferred packages, an RFID warning must arrive in time to correct a load, and the route must still leave on time. Vacations, weather, equipment failures, and peak days will test the smaller system more severely than an average shift.
UPS entered the second half with higher revenue guidance and a stated completion of the Amazon glide-down. Domestic revenue per piece rose 9.3 percent. Package-level service, overtime, safety, and customer retention will determine whether those commercial signals survive the changed network.
Public materials reviewed for this article contain no account from a participating driver or a manager at a receiving facility. Their experience remains a material gap because local route knowledge, commuting feasibility, staffing depth, and exception work sit outside the national charge.
One date is especially concrete: July 31, 2028. Until then, the Teamsters settlement bars UPS from offering another severance program. If volume slips again, management will have to work with the people and capacity left in the network, find another commercial or operating response, or return to the bargaining table.
UPS has booked the workforce exit. The next reset will need a different answer.