Operations Analysis
UNIQLO hems pants over $20 for free, and promises it the same day. It’s one of the clearest signals of the brand’s quality-at-a-fair-price position. This paper maps that service at UNIQLO UTC in San Diego, measures its capacity, and asks what happens to the promise on a sale-day Saturday.
The processing times, staffing and inflow rates come from my own shifts as an alteration specialist, and from an interview and walkthrough with the store manager.
Flow unit
One alteration order — one pair of pants
Starts
Alteration requested at checkout
Ends
Customer collects the altered pants
Out of scope
The ~5 min checkout line, shared with the whole store
Where UNIQLO competes
UNIQLO trades trend speed and variety for standardized quality at a fair price: seasonal rather than weekly assortments, few SKUs, its own fabric tech. It sits inside the frontier Zara and H&M define, and free hemming is part of how it earns that position in-store.
Operations frontier
Cost efficiency against responsiveness, 1–10, higher is better on both.
Positions are the paper’s estimates from public pricing and analyst benchmarks — indicative of magnitude, not measurements.
| Company | Cost eff. | Speed | Rationale |
|---|---|---|---|
| Zara | 6 | 9 | 2-week design-to-shelf; moderate price |
| H&M | 8 | 7 | Very cheap, reasonably fast; high volume |
| UNIQLO | 6 | 4 | Affordable quality; slow cycles, fast in-store service |
| Gap | 4 | 4 | Mid-market; operational lag |
| Lululemon | 2 | 5 | Premium price; medium speed; differentiated on quality |
| Nordstrom | 2 | 3 | High cost, curated selection; service-led |
Where AI fits. Forecasting, inventory prediction and fit recommendations would move UNIQLO toward the existing frontier — improvement, not innovation. Automated hemming that finishes precisely in under an hour with no added labor would shift the frontier itself.
Process map
Solid cards are work; dashed ones are racks, where pants hang and nothing happens to them. The analysis is mostly about the dashed ones. Walk an order through and watch how few of the minutes are spent on the pants.
Alteration Capacity Analysis covers 11 of 12 steps.
Customer
Floor staff
Specialist
Manager
Send one lockstitch order through the service and watch where the minutes go.
Quality control and rework
The manager’s check is the last gate. A rejected hem — usually a small inseam mismatch or an uneven seam — goes back to the machine for another 15 to 30 minutes. Rework is rare, but when it happens it re-enters the busiest resource in the store.
Pricing
| Stitch | Eligible | ≥ $20 | < $20 |
|---|---|---|---|
| Lockstitch | All pants | Free | $5 |
| Blindstitch | Dress pants only | $5 | $10 |
Capacity
Each pool’s capacity is how many staff it has × 60 ÷ its minutes per order. Specialists carry 21 minutes of work per order, and there are never more than three of them.
Weighted hem time
(0.80 × 15) + (0.20 × 30) = 18 min
80% lockstitch at 15 min, 20% blindstitch at 30
Specialist unit load
1.5 + 18 + 1.5 = 21 min / order
Claim it, hem it, mark it confirmed
Orders per hour each resource pool can handle
Capacity = resources × 60 ÷ unit load. Every non-specialist pool clears even the best specialist staffing.
| Pool | Context | Unit load | Resources | Orders/hr |
|---|---|---|---|---|
| Sales associate · intake | All shifts | 8 min | ~2 at peak | 15 |
| Specialist | Weekday morning | 21 min | 1 | 2.86 |
| Specialists | Weekday afternoon/evening | 21 min | 2 | 5.71 |
| Specialists | Weekend | 21 min | 3 | 8.57 |
| Manager · quality check | All shifts | 2.5 min | 1, part-time | 12 |
| Sales associate · pickup | All shifts | 3 min | 1 | 20 |
That matches the floor. A lone weekday-morning specialist gets through 11 or 12 orders in a four-hour shift; anything past that becomes a backlog. The manager check has room on paper — its problem isn’t capacity, it’s that nobody reliably tells the manager there’s something to check.
Little’s Law · weekday afternoon
Two specialists, about five orders an hour. The system is stable — inflow sits under capacity — but only just.
5 / hr
Throughput (R)
Inflow; stable because 5 < 5.71
~120 min
Flow time (T)
Observed on weekday afternoons
~10
Orders in the system (I)
I = R × T = 5 × 2
88%
Specialist utilization
5 ÷ 5.71 — almost no slack
One order on a weekday afternoon: about 120 minutes
Theoretical flow time is the sum of hands-on steps. The rest is time spent hanging on a rack.
| Part | Minutes | Share |
|---|---|---|
| Being worked on (theoretical flow time) | 34.5 | 29% |
| Waiting in a buffer | 85.5 | 71% |
| Observed flow time | 120 | 100% |
Theoretical flow time
| Order | Hands-on steps | Minutes |
|---|---|---|
| Lockstitch (80%) | 4 + 4 + 1.5 + 15 + 1.5 + 2.5 + 3 | 31.5 |
| Blindstitch (20%) | 4 + 4 + 1.5 + 30 + 1.5 + 2.5 + 3 | 46.5 |
| Weighted average | (0.80 × 31.5) + (0.20 × 46.5) | 34.5 |
Does the process match the strategy?
The output does: professional hems, same day, checked before release. The process around it doesn’t — hand-typed intake, a notification that fails, and a radio doing the software’s job add 85.5 minutes and nothing a customer would pay for. And at 88% utilization, any bump in demand has nowhere to go.
Stress scenario
Three specialists — the most the store ever schedules — against a weekend rush observed at 10 to 20 orders an hour. The paper uses 12 as its defensible figure. Try the rest of the range.
Saturday back-to-school sale: the queue by the hour
Change the rush and the staffing. The model is the paper’s; the defaults are its assumptions.
20.6
orders waiting at 6pm
6.9
still waiting at close
+1 hr 20 min
added to the average wait
These are the paper’s inputs, and the model lands on its numbers: 20.6 at the peak, 6.9 carried into Sunday, about an hour and 20 minutes of added wait.
| Measure | Value |
|---|---|
| System capacity | 8.57 orders/hr |
| Peak backlog (6pm) | 20.6 orders |
| Left at close (9pm) | 6.9 orders |
| Average backlog, noon to close | 11.43 orders |
| Added wait (I ÷ R) | 1.33 hr |
A customer who drops off at 2pm isn’t waiting because anyone is working slowly. They’re waiting because of the queue that formed before they arrived. Push the rush toward 15 or 20 an hour and the rack can’t recover before close no matter how quiet the evening gets — at that point “same day” is a promise the system can’t keep.
Recommendations
01
Flow time — removes Buffer 2
Buffer 2 exists only because a software notification fails. Configure Waitwhile to push properly to the manager’s device, or message the manager the moment an order is marked Confirmed. Zero added work for the specialist; one whole buffer gone from every order.
02
System capacity — relieves the bottleneck
Only about 10% of staff are trained for alterations, and most of them are managers with other jobs. Training two or three floor employees lets the schedule add a fourth specialist when peaks are expected, taking weekend capacity from 8.57 to 11.43 orders an hour.
03
Intake efficiency + notification reliability
Intake takes 4–5 minutes of hand-typed fields, and the SMS opt-in is the step forgotten in a rush. Flip it to opt-out so customers are notified without anyone changing behavior, and prompt staff with the fields they need.
Until then, on a surge day
One service, four angles
This paper builds on the research from the WaitWhile redesign. Read next to the tailoring experiment, it raises a harder question: if offering alterations did bring in more demand, could the service absorb it?
More work inOperationsDataStrategy
Next Project
← All workLike work like this? Tell me what you're building.
Say hi →