Jerome NazarioHow I build and ship real products
← Back

TransWealth Staffing Analysis

Data Analyst (Freelance) · Data-backed headcount cut · 2 min read

TransWealth Fleet & Parking Management runs valet operations that move a daily average of 302 cars. In Q1 2026 they faced the question every operations manager dreads asking out loud: if we cut the driver roster by 47.5%, does service actually survive?

They brought me in as a freelance data analyst to answer it with arithmetic instead of instinct. The deliverable wasn't a dashboard — it was a management review report a leadership team could act on, with the math shown.

The question under the question

Headcount decisions usually get argued with anecdotes — the one bad Saturday, the one idle Tuesday. The real question is structural: how many cars can a smaller team actually move, hour by hour, before quality slips? That's a capacity-modelling problem, and it has a numeric answer.

The method

Four steps from gut feeling to defensible number.

Demand traced from real daily volume, the reduced roster's throughput modelled from cycle times per station, then the model pushed past reality until something broke.

Step 1
Trace demand
Count the real day, hour by houra daily average of 302 cars — arrivals, retrievals, and the surge windows between them
Step 2
Model capacity
Compute the reduced team’s ceilingcycle times per driver per station → how many cars the smaller roster can actually move
Step 3
Stress test
Push the model past realitythe cut roster clears more than 2.5× current daily demand before service degrades
Step 4
Flag the risk
Find where it breaks firstnot the drivers — the entrance/ticketing station is the first bottleneck in a surge
Python + Excel

The model runs in Python; the deliverable lands in the format management already reads.

302 cars/day

Every scenario is anchored to the operation's actual daily average, not a hypothetical.

What the model said

The headline finding was blunt: even after cutting nearly half the drivers, the remaining team's throughput clears more than 2.5× the current daily demand. The reduction wasn't a gamble — the old roster had been sized for a peak that the data says never arrives.

The reduced team could absorb 2.5× today's volume before service quality degrades — the staffing cut wasn't a risk to defend, it was over-capacity finally measured.

But the model also found the failure point nobody was arguing about. Cars don't queue for drivers — they queue at the entrance and ticketing station, the single point every arriving vehicle passes through. In surge windows, that's where the line forms first.

The bottleneck

The constraint wasn't the people being cut.

Driver capacity holds with room to spare; the entrance/ticketing station is what breaks first under surge load. Cutting drivers was safe — ignoring the gate would not have been.

Driver capacity

Even at 47.5% fewer drivers, the roster moves more than 2.5× the daily volume. Headcount was never the constraint the operation thought it was.

holds at 2.5× demand — with room to spare
Entrance & ticketing

One station meets every arriving car. During peak surges the queue forms here first — before any driver is even close to busy.

the true bottleneck during surge windows
Counter-intuitive

The analysis defended the cut and redirected attention to the part of the operation the cut couldn't fix.

What management got

The report closed with three operating recommendations, each tied to a number the team can watch:

Monitor weekly volume against the model

The 2.5× margin is real but not infinite — a standing weekly check catches demand drift long before it becomes a service problem.

Set surge staffing thresholds

Defined volume trip-points that trigger extra coverage, so peak days are handled by rule instead of by panic.

Rework the entrance/ticketing flow

Operational adjustments at the gate — the one station where the queue forms first and the customer feels it most.

The outcome: a 47.5% staffing reduction that leadership could sign off on with the math in front of them — and a watch-list for the one bottleneck that headcount was never going to solve.

Tech stack
Client engagement — the full report stays private