Build, Buy, or Outsource? The Total Cost of Owning an Employee Transportation Program

A transportation ops manager costs about $143,000 a year fully loaded. A 55-passenger motorcoach rents for roughly $93 an hour plus $4.10 a mile. A custom dispatch tool, built in-house, has a better-than-even chance of finishing late and over budget. Three numbers, three different ways to run an employee transportation program, and the question every Operations VP and CFO eventually faces: build it, buy a contract, or buy software and keep the program in your own hands.

Most procurement models answer that question with one of two cost lines: vehicles or vendor fees. Both are visible. Both are easy to drop into a spreadsheet. Both leave out the labor your own team will spend, the optimization savings you forgo, the price of switching when the first choice goes wrong, and the compliance exposure that lands on your desk regardless of who drives. For an employer running 200 to 20,000 people across one or many sites, those omitted lines often decide which model is actually cheaper over five years.

This is the upstream decision. Once you know which of the three models fits your situation, the 2026 buyer’s guide to shuttle management software covers how to score specific platforms, and the data-driven guide to reducing corporate transportation costs runs the multi-year NPV math on a single shuttle decision. Here, the job is narrower and earlier: a vendor-neutral framework for the corporate shuttle build vs buy choice, scored across six cost lines, so you can take a defensible total-cost-of-ownership comparison to your own finance team. What this framework does not do is pick a vendor for you, or pretend any one model wins everywhere. None do.

What “total cost of ownership” actually covers for an employee transportation program

Start with why the spend exists at all. Mean one-way commute time hit 26.8 minutes in 2023, and 69.2% of US workers still drove alone to work (Census Bureau, ACS 1-Year Estimates). That structural fact - one worker, one car, two trips a day - is what a shuttle, vanpool, or managed-commute program exists to bend. Whether you do it with your own buses, a charter operator’s coaches, or a platform that orchestrates either, the goal is the same: move more people in fewer vehicle-miles, reliably enough that they leave the car at home.

Employee transportation total cost of ownership is the full annual cost of running that program, not the sticker price of the vehicles or the line on the vendor invoice. Six categories carry almost all of it.

• Operations labor, fully loaded: planners, dispatchers, schedulers, and (if you own vehicles) drivers, including benefits and overhead, not just base wage.
• Vehicles or contract fees: capital, fuel, insurance, and maintenance for an owned fleet, or per-vehicle-hour and per-mile fees for a contracted one.
• Software: the dispatch, routing, rider-app, and reporting layer - licensed, built, or bundled into a contract.
• Optimization savings (a negative cost): the vehicles and miles you avoid because routes are consolidated rather than padded.
• Switching cost and lock-in: what it costs to leave - contract notice periods, re-bidding, data migration, re-onboarding.
• Risk and compliance: insurance liability, driver-hours rules, accessibility, and the reporting burden that survives any model.

The reason TCO matters more than the invoice is that the three models load these six lines very differently. Building in-house pushes nearly everything onto your own labor and capital. Outsourcing collapses labor and capital into a single fee but quietly inflates the switching-cost line. The software-managed model splits the difference, trading a license fee for control over the optimization and switching lines. Price only vehicles and fees, and all three look roughly comparable. Price all six, and they diverge by millions over a multi-year term.

Two figures anchor the cost inputs that follow, and both come from the IRS for 2026: the business standard mileage rate is 72.5 cents per mile (IRS newsroom), and the pre-tax commuter benefit limit sits at $340 a month for transit and $340 for parking ($4,080 a year each), per IRS Publication 15-B. Keep those consistent if you also model a do-nothing reimbursement baseline. They are the floor the three active models compete against.

Running shuttles in-house: the build model and the labor it hides

Building means you own the program end to end. Owned or leased vehicles, your drivers or contracted ones, your dispatchers, your spreadsheets or your own software. Control is total. So is the labor bill, and procurement models routinely understate it because they price the wage, not the worker.

A transportation, storage, and distribution manager earned a median annual wage of $102,010 in May 2024, the BLS reported through its Occupational Outlook Handbook. That is the wage. The worker costs more. For private-industry employees, benefits ran 29.9% of total compensation as of December 2025 (BLS, Employer Costs for Employee Compensation), which lands the fully-loaded multiplier near 1.4x on wages - consistent with the long-standing MIT rule of thumb of 1.25 to 1.4 times base salary. Apply it: that $102,010 manager costs roughly $143,000 a year all in. For unionized or public-transit operators, where benefits reach 38.3% of compensation, the multiplier runs closer to 1.6x.

One manager is rarely the whole labor line. A program that dispatches across multiple shifts needs coordination labor that the org chart forgets. Dispatchers (excluding police, fire, and ambulance) earned a median $48,880 in May 2023, near $23.50 an hour. If you own the vehicles, drivers are the largest labor line of all: transit and intercity bus drivers earned a median $57,440, and school-bus-class drivers $47,040. Layer the 1.4x multiplier on each, then multiply by the headcount a 24-hour, multi-site operation actually requires, and the build model’s labor line alone can clear $500,000 before a single bus turns a wheel. That number never appears on a vehicle quote.

Then there is the software trap. Plenty of operations teams decide they will just build the dispatch and optimization tool themselves, because the routing logic “isn’t that hard.” It is. McKinsey, working with the University of Oxford across more than 5,400 projects, found that large IT projects run 45% over budget and 7% over time while delivering 56% less value than predicted - and that 17% go so badly they threaten the organization’s existence. The size effect is the part that bites here: small software projects succeed most of the time, while large custom builds rarely do. A workforce-routing platform with rider apps, HRIS integration, driver tooling, and live dispatch is a large build. The McKinsey dataset is from 2012, but it remains the most-cited benchmark for exactly this failure mode, and nothing since has reversed the direction.

The build model wins in a narrow band: a single site, a stable schedule, abundant in-house ops capacity, and a long enough horizon to amortize the fixed labor. Microsoft has run its own Connector network in the Puget Sound region since 2007, scaling to 22 routes served by 74 buses and, by the company’s 2014 accounting, eliminating more than 4.2 million single-occupancy vehicle trips and 59 million miles of travel. That is a build that pencils. It also belongs to one of the largest employers on earth, with the headcount density and engineering bench to justify owning the whole program. Most employers do not have that.

Outsourcing to a charter operator: the per-hour model and its hidden lines

Hand the whole thing to a charter or motorcoach operator and the cost structure inverts. No drivers on your payroll, no buses on your balance sheet, no dispatch software to maintain. You pay a fee, the operator runs the service, and your labor line shrinks to whoever manages the contract. For employers without in-house ops capacity, that simplicity is the entire appeal.

The fee is where it gets less simple. Charter pricing is built from a per-vehicle-hour rate plus a per-live-mile charge, a minimum number of billable hours, and add-ons for second drivers and positioning. Real published contracts show the shape. A University of Nevada, Las Vegas rate card priced a 55-plus-passenger motorcoach at $92.70 an hour plus $4.10 per live mile, a 30-passenger mini-coach at $93.97 an hour plus $4.43 per mile, a second driver at $65 an hour, and a three-hour minimum (2022 pricing). Smaller shuttle work runs cheaper: the City of Opa-locka, Florida, paid an incumbent operator $54 an hour for a community shuttle route, spending about $210,000 to $211,000 a year on it.

Compare those private rates to the fully-loaded public benchmark and the gap explains a lot. The national weighted-average operating expense for fixed-route motor bus service was $181.88 per vehicle revenue hour in 2023, computed from the Federal Transit Administration’s National Transit Database across more than 1,200 reporting agencies. Commuter bus mode ran higher, at $262.94. That figure is not a contradiction of the $54-to-$93 charter rates - it is the difference between a private operator’s quoted hourly fee and a transit agency’s all-in cost with overhead, maintenance, and administration bundled in. The lesson for a build-vs-buy model: a charter quote is the operator’s price before your own oversight cost, while the NTD number is what the same service costs once every overhead line is counted. Your true outsource cost sits between them.

Two lines hide inside a clean per-hour quote. The first is deadhead - the time and miles a vehicle travels out of revenue service, garage to first pickup and last drop-off back to garage. The FTA’s Charter Service rule explicitly contemplates that customers can be billed for positioning time, cautioning only that they should not pay for “inordinate hours of deadhead time.” Finance models that price service hours and ignore positioning understate the bill, sometimes badly, on routes far from the depot. The second is escalation. Opa-locka’s contract built in a clause that “a 3% increase will be acceptable after the expiration of the term,” which is the outsource model’s quiet ratchet: rates climb every renewal, and the longer the relationship, the more you pay per hour for the same service.

Outsourcing fits employers with variable or seasonal demand, no appetite to manage drivers, and routes a contractor already knows. It fits less well when you run a permanent, dense, multi-shift program, because then you are paying an operator’s margin on a service you have enough volume to optimize yourself - and you are accepting the contract terms that make leaving expensive. More on that below.

Managing it with software over your fleet or vetted operators: keeping the program in your hands

A third model sits between building everything and outsourcing everything: run a management platform on top of either your own vehicles or vetted third-party operators, and keep the program design - routes, schedules, rider rules, reporting - in-house. The vehicles can be yours, a contractor’s, or a mix. What you own is the orchestration layer that decides where they go and who rides.

What that layer buys, in cost terms, is optimization headroom - and the headroom is large because the baseline is so wasteful. The 2017 National Household Travel Survey found that 80% of work trips were single-occupant; only 10% carried two or more people (FHWA / Oak Ridge National Laboratory). Every empty seat on a commute is a vehicle-mile that did not need to exist. How much of that can consolidation actually recover? The cleanest peer-reviewed measurement comes from a PNAS study of 150 million New York City trips, which found that with modest increases in passenger inconvenience, cumulative trip length could be cut by 40% or more through optimized pooling (Santi et al., 2014). That was taxi data, not a shuttle, so treat it as evidence of the mechanism rather than a guaranteed shuttle result. The direction is the point: matching riders and routes algorithmically removes vehicle-miles that padded, manually-planned schedules leave on the table.

Tech employers in San Francisco built exactly this pattern, even though the public attention went to the buses. Google, Apple, Genentech, and others run commuter shuttles that, in aggregate, carried about 17,000 daily boardings on average weekdays, the SFMTA reported in its 2015 program evaluation. Nearly half of surveyed riders - 47% - said they would otherwise drive alone. The companies design the programs and own the rider experience; the driving is contracted to operators like Hallcon, WeDriveU, and Bauer, with employers paying the city $9.35 per stop event (SFMTA Commuter Shuttle Program FAQs). That is the software-over-operators model in the wild: the employer keeps program control and demand data, the operator supplies vehicles and drivers, and a coordinating layer ties them together.

The honest cost case requires naming this model’s own downsides, because it is not free. You pay a software license, which is a real recurring line. The platform has to integrate with your HRIS and your operator’s dispatch, and integration engineering is rarely zero. And if the platform is bound to a single fleet or a single operator, you have simply moved the lock-in from a charter contract to a software contract. The version of this model that avoids that trap is fleet-agnostic: a platform that works with an owned fleet or any vetted operator means you can change vehicles or vendors without changing the system of record. Ryde’s smart employee commuting platform is built around that fleet-agnostic design, running over an employer’s own fleet or contracted operators rather than locking the program to either. Whether you end up choosing Ryde or another platform, the structural test is the same: does the software let you change who drives without rebuilding the program?

The six cost lines, side by side

Here is the framework as a comparison table. Each model loads the six TCO lines differently; the table makes the trade explicit rather than declaring a winner.

A couple of cells deserve a second look. Optimization savings is the only line that can be negative - it reduces total cost rather than adding to it - and it is the line the build and outsource models capture least. And the switching-cost line is the one that flips the whole comparison over a five-year horizon, which is why it gets its own section below.

A worked TCO example for a 1,200-person multi-site employer

Numbers make the framework usable. Take a manufacturing employer running 1,200 people across two sites on three shifts, with a commute long enough that drive-alone is the default - the structural case the ACS 26.8-minute, 69.2%-drive-alone backdrop describes. Assume the program needs roughly 8 vehicles in service to cover the shift peaks. What does each model cost in a year?

Build first. Labor is the heavy line: one ops manager at ~$143K fully loaded, two dispatchers at ~$68K each loaded ($48,880 median × ~1.4), and eight drivers at ~$80K each loaded ($57,440 median × ~1.4) lands near $850,000 in labor before vehicles. Add eight owned or leased buses with fuel, insurance, and maintenance, and the all-in build cost for this site clears well past $1.5M annually - with a one-time software build on top if you go that route, carrying the McKinsey overrun risk. You own everything, including the failure modes.

Outsource second. Eight vehicles running, say, 2,400 revenue hours a year at a blended $80 an hour comes to about $192,000 in hourly fees, plus per-mile and deadhead charges that can add 20% or more, plus the 3% annual escalator compounding over the term. Call it $250,000 to $300,000 in year one, climbing each renewal. Far lower than build on paper - because you are renting capacity, not owning it, and the operator’s optimization upside is theirs, not yours.

Software-managed third. You pay a platform license plus integration, run the same eight vehicles (owned or contracted), and capture the consolidation savings yourself. If optimized routing lets you cover the same shift peaks with six vehicles instead of eight - well within the PNAS-documented range for the mechanism - you have removed a quarter of the largest variable cost, whichever model supplies the buses. The license and integration are real adds; the avoided vehicle-miles and the retained demand data are the offsets.

For a real, fully-costed reference point, the University of Minnesota’s campus transit program published its FY2023 numbers: 2.25 million riders, $6.18M in total annual cost, $2.74 cost per passenger, and $10.74 per mile across 575,865 miles operated. That is a campus circulator serving students and staff, not a pure employee shuttle, so use it as a transit-cost yardstick rather than a like-for-like comparable. The discipline it models is the one that matters: a real program publishes cost per rider and cost per mile, and any TCO comparison worth taking to finance ends in those two numbers, not in a vehicle quote.

Switching cost and lock-in: the line that decides it over five years

Most build-vs-buy spreadsheets compare year-one cost and stop. That is the expensive mistake, because the cheapest model in year one is frequently the most expensive to leave in year four. Lock-in is, in the GOV.UK definition, the condition “where switching from one technology or provider to another is difficult, time consuming and disproportionately expensive.” It is a cost even when it never shows up as an invoice line, because it removes your ability to re-bid the price down.

The outsource model carries the heaviest version. Charter contracts run on multi-year terms with renewal options and long notice windows. Salem State University’s shuttle and charter RFP set a base term plus two one-year options, capped at five years, required 120 days’ advance written notice of intent not to renew, and demanded that bidders have at least five years of operating history. Read those three terms together and the switching cost is obvious: you are committed for years, you must give months of notice to exit, and the five-year-history bar limits how many operators can even bid to replace the incumbent. Add Opa-locka’s 3% annual escalator and the price ratchets up across exactly the period you are least free to leave. The convenience of outsourcing is real; so is the cost of being unable to walk away from it cheaply.

Building avoids contract lock-in - you own the buses and the team - but substitutes sunk capital lock-in. Eight buses and a custom software build are not assets you unwind in 120 days; they are a decision you live with until they depreciate. The software-managed model is the only one of the three where lock-in is a design choice rather than a structural feature. A fleet-agnostic platform lets you change operators, add or retire vehicles, and re-bid the driving without rebuilding the system that runs the program. A single-fleet-bound tool gives that advantage away. So the test, again, is structural: can you change who drives without changing the system of record? If yes, the switching-cost line stays low. If no, you have bought lock-in under a different name.

How to decide: a scoring rubric for your situation

No single model wins for every employer, so score your own situation rather than copying someone else’s answer. Rate each factor 1 to 5 for how strongly it points toward a given model, then read the totals.

Lean toward building when you have abundant in-house operations capacity, a single dense site, a stable long-term schedule, headcount high enough to amortize fixed labor (think thousands at one location), and a strategic reason to own the program outright. Microsoft’s Connector is the archetype. If you are not Microsoft, score this model honestly and it usually loses on the labor and software-build lines.

Lean toward outsourcing when demand is variable or seasonal, you have no appetite to employ drivers, your routes are ones a contractor already serves, and your horizon is short enough that the escalators and lock-in never fully bite. Outsourcing is also the right first move for a pilot, where you want to test demand before committing capital - the reasons shuttle pilots stall at week six are worth reading before you scope one.

Lean toward software over fleet or operators when you run a permanent, multi-shift, multi-site program, want the optimization savings to accrue to you rather than a contractor, need centralized reporting across sites, and value the freedom to change vehicles or vendors without rebuilding. This is the model that fits the most common enterprise case: enough scale to justify owning the program design, not enough reason to own every bus and build every tool.

The factors that move the score most are headcount per site, number of sites, in-house ops capacity, expected program lifespan, and how much control over rider experience and data the program needs. Long lifespans and multiple sites push toward owning the orchestration layer; short horizons and single sites pull toward a contract. Run the six-line TCO for each model with your own numbers before you decide - and weight the switching-cost line heavily, because it is the one a year-one comparison hides.

Common questions on build, buy, and outsource decisions

Is it cheaper to outsource shuttle services or run them in-house?

In year one, outsourcing is almost always cheaper than building, because you rent capacity instead of buying buses and hiring drivers. Over five years the answer can flip: charter contracts carry 3% annual escalators and multi-year lock-in (as in the Opa-locka and Salem State public contracts), while a build amortizes its fixed labor and capital. The decision turns on program lifespan and scale, not on the year-one quote alone.

How much does it cost to run a corporate shuttle in-house?

The labor is the line most teams underprice. A transportation operations manager runs about $143,000 fully loaded - a $102,010 median wage (BLS, 2024) times roughly 1.4 for benefits and overhead - before you add dispatchers (median $48,880) and drivers (median $57,440), each loaded the same way. For a multi-shift, multi-site program needing eight or so vehicles, the in-house labor line alone can clear $850,000 a year, on top of vehicle capital, fuel, insurance, and maintenance. Custom software, if you build it rather than license it, adds the large-IT overrun risk McKinsey documented: 45% over budget for 56% less value than predicted.

What costs are included in employee transportation TCO?

Six lines: operations labor (fully loaded, wages plus benefits and overhead), vehicles or contract fees, software, optimization savings (a negative cost when routes consolidate), switching cost and lock-in, and risk and compliance. Most procurement models price only vehicles and fees and miss the other four - which is exactly where the three models diverge.

Should you build or buy shuttle management software?

Buy it, in almost every case, unless software is your core business. Large custom IT builds run 45% over budget and deliver 56% less value than predicted (McKinsey / University of Oxford, across 5,400+ projects), and a workforce-routing platform with rider apps, integrations, and live dispatch is a large build. Licensing a fleet-agnostic platform also keeps your switching cost low, since you can change operators or vehicles without rebuilding the system that runs the program.

What is vendor lock-in and how do you avoid it in a transportation program?

Vendor lock-in is when switching providers is “difficult, time consuming and disproportionately expensive” (GOV.UK). In employee transportation it appears two ways: a single multi-year charter contract with notice windows and escalators, or a software tool bound to one fleet or operator. You avoid it by choosing a fleet-agnostic management layer that lets you re-bid the driving and swap vehicles without changing your system of record - so the switching-cost line stays low across the whole program lifespan.

Where to start before you pick a model

Two decisions sit underneath everything above. First, price all six TCO lines, not the two that fit easily on a quote - the labor, optimization, switching, and compliance lines are where the three models actually separate. Second, weight the switching-cost line for a five-year horizon, because the model that is cheapest in year one is often the one you cannot afford to leave in year four.

This week, pull your own numbers: fully-loaded ops headcount, current vehicle or contract spend, the seats running empty at peak, and the contract notice period you are bound by today. Run the six-line comparison for each model against those figures. If the math points toward keeping the program in-house while letting software do the routing and reporting, the smart employee commuting platform is built for exactly that fleet-agnostic case - and when you are ready to score specific platforms, the 2026 shuttle software buyer’s guide picks up where this framework ends. Start with the numbers; let them, not a vendor demo, choose the model. See how Ryde fits at goryde.com.

Sources

• US Bureau of Labor Statistics, Transportation, Storage, and Distribution Managers, Occupational Outlook Handbook (median $102,010, May 2024).
• US Bureau of Labor Statistics, Dispatchers, Except Police, Fire, and Ambulance, OEWS 43-5032 (median $48,880).
• US Bureau of Labor Statistics, Bus Drivers, Occupational Outlook Handbook (transit/intercity median $57,440; school-bus $47,040).
• US Bureau of Labor Statistics, Employer Costs for Employee Compensation - December 2025 (benefits 29.9% of private-industry compensation; 38.3% state and local government).
• US DOT, Federal Transit Administration, National Transit Database Metrics (motor bus $181.88/VRH in 2023; commuter bus $262.94).
• University of Nevada, Las Vegas Purchasing, Royal Coach Tours Contract Pricing (motorcoach $92.70/hr + $4.10/mile, 2022).
• City of Opa-locka, FL, Community Shuttle Service RFP #22-1213100 ($54/hr; 3% annual escalator).
• Salem State University, Shuttle and Bus Charter Service RFP 2021-07 (5-year ceiling; 120-day notice; 5-year-history bidder bar).
• US DOT, Federal Transit Administration, Charter Service Final Rule, Federal Register Vol. 73 No. 149 (deadhead definition; 2008).
• Santi et al., Quantifying the benefits of vehicle pooling with shareability networks, PNAS 111:13290-13294 (up to 40%+ trip-length reduction, 2014).
• FHWA / Oak Ridge National Laboratory, 2017 National Household Travel Survey - US Travel Profile (80% single-occupant work trips).
• University of Minnesota Parking & Transportation Services, Transit Programs Annual Report, FY2023 ($2.74/passenger; $10.74/mile).
• McKinsey & Company / University of Oxford, Delivering large-scale IT projects on time, on budget, and on value (45% over budget; 56% less value; 2012).
• GOV.UK, Central Digital and Data Office, Cloud guide for the public sector (vendor lock-in definition; updated November 2023).
• US Census Bureau, Commuting Characteristics by Sex, ACS 1-Year Estimates, Table S0801 (26.8-minute mean commute; 69.2% drive-alone, 2023).
• Internal Revenue Service, IRS Sets 2026 Business Standard Mileage Rate at 72.5 Cents Per Mile (72.5 cents/mile, 2026).
• Internal Revenue Service, Publication 15-B, Employer’s Tax Guide to Fringe Benefits ($340/month transit and parking commuter limit, 2026).
• Microsoft, Making commuting easy, enjoyable, and sustainable (Connector: 22 routes, 74 buses; 4.2M+ trips eliminated, 2014 cumulative).
• San Francisco Municipal Transportation Agency, Commuter Shuttles Pilot Program Evaluation Complete (~17,000 daily boardings; 47% would otherwise drive alone, 2015).
• San Francisco Municipal Transportation Agency, Commuter Shuttle Program FAQs ($9.35 per stop-event fee; permitted operators).