Build or buy: the question almost every growing shipper eventually asks.
At some point it gets said out loud in a leadership meeting. Why are we paying someone to manage our freight when we could hire a team and do it ourselves?
It tends to surface the same way. Freight spend has grown faster than the strategy managing it. A provider disappointed: opaque fees, service that drifted after signature, reporting that answered their questions instead of yours. A board asked where the margin is. And the instinct that follows is a sound one: bring it inside, where you can see it and control it.
It is the right question. It deserves a real answer rather than a brochure. So this page prices the in-house build from public data, shows the arithmetic, and names the three cases where building wins outright, including the band where the honest answer argues against us.
The short answer
$0.7M – $1.5M
Argus white paper · September 2026
Build or Buy: What a Mid-Market Freight Function Actually Costs
- The full six-role cost model, per role, fully loaded
- The technology line: subscription, implementation, integration, payback
- All eleven questions to answer before you commit either way
- Complete sources and method appendix
Build or buy: the question almost every growing shipper eventually asks.
At some point it gets said out loud in a leadership meeting. Why are we paying someone to manage our freight when we could hire a team and do it ourselves?
It tends to surface the same way. Freight spend has grown faster than the strategy managing it. A provider disappointed: opaque fees, service that drifted after signature, reporting that answered their questions instead of yours. A board asked where the margin is. And the instinct that follows is a sound one: bring it inside, where you can see it and control it.
It is the right question. It deserves a real answer rather than a brochure. So this page prices the in-house build from public data, shows the arithmetic, and names the three cases where building wins outright, including the band where the honest answer argues against us.
Argus white paper · September 2026
Build or Buy: What a Mid-Market Freight Function Actually Costs
- The full six-role cost model, per role, fully loaded
- The technology line: subscription, implementation, integration, payback
- All eleven questions to answer before you commit either way
- Complete sources and method appendix
The short answer
$0.7M – $1.5M
First-time inclusion, Gartner® Magic Quadrant™ for Fourth-Party Logistics
Start here
Most insourcing business cases are answering the wrong question.
Sit with enough teams drafting one and a pattern emerges. The plan is less a strategy than a verdict.
Somewhere behind it is a provider that disappointed. Opaque fees. Service that drifted once the contract was signed. Reporting that answered the provider’s questions instead of yours. The conclusion writes itself: if outsourcing means this, we’ll do it ourselves.
But look closely at what actually burned most of these teams and it is rarely outsourcing. It is a specific layer of it. Consider an execution provider: a 3PL with warehouses to fill, or a broker whose margin sits in the spread between your rate and the carrier’s. Its economics improve as more of your activity flows through what it owns. That is precisely the misalignment the insourcing case is reacting to.
Gartner draws the distinction cleanly: third-party providers handle tactical work like storage, shipping and fulfillment, while fourth-party providers act as strategic managers coordinating across multiple 3PLs and partners on the shipper’s behalf. Their shorthand is hard to improve on: the 3PL is the logistics muscle, the 4PL is the logistics brain. Gartner also names the confusion directly, describing significant terminology fatigue across the industry, with many 3PLs now marketing 4PL-like services.
That blur is where the error lives. Not in distrusting a provider that earned it, but in the generalization that follows.
Write down what your insourcing case is actually asking for.
The insourcing case wants
Control over decisions
Strategy, trade-offs and budget stay with your leadership. The partner executes to your priorities and reports to you.
The insourcing case wants
Transparent economics
Management fee fully separated from transportation cost. No margin hidden in your rates. No assets to feed.
The insourcing case wants
Ownership of the data
Your lanes, rates and history in dashboards built for you, not locked inside a provider's platform.
The insourcing case wants
Accountability for results
Documented savings contractually guaranteed to exceed the fee. Accountability most execution providers will not put in writing.
The insourcing case wants
An advocate with no side interests
Carrier-agnostic scorecards. The carrier that performs gets more volume, and no affiliated fleet is waiting in line.
The number
What the function actually costs, from federal data.
How we got the number, and why we built it ourselves
The honest answer depends on what you are actually building. So we priced three versions of it. The gap between them is where most insourcing business cases quietly fall apart.
Tier 1 · The floor
$476K – $1.05M
$440,300 Four-person core: manager, analyst, sourcing agent, audit specialist
$795,800 Six-person core: adds a second analyst, procurement leadership, TMS administrator
$36K–$130K TMS subscription plus implementation amortized over five years
Tier 2 · Credible mid-market
$0.7M – $1.5M
+$108K–$216K Extended-hours coverage: one to two track-and-trace coordinators
+$25K–$75K Freight bill audit tooling
+$25K–$75K Rate intelligence and benchmarking subscription
+$20K–$75K Carrier vetting and compliance tooling
+$40K–$80K Recruiting and ramp, amortized
Tier 3 · Enterprise grade
$1.5M – $3.0M
+$454K–$486K True 24/7 track-and-trace: 4.2 FTE per around-the-clock seat
+$150K–$500K Licensed enterprise TMS plus complex ERP integration
+$100K–$250K Audit, rate intelligence and carrier compliance as a full stack
+22% Senior and metro wage tier on the core team
+$264K Dedicated BI analyst, second TMS admin, compliance analyst
Here is the category error inside the cost model itself. Most insourcing business cases price Tier 1 and budget for Tier 3 outcomes. They approve a team and a TMS, then expect around-the-clock coverage, audited invoices, benchmarked rates, documented carrier vetting and SKU-level reporting. The distance between what gets approved and what gets expected is one to two and a half million dollars a year, and it does not appear anywhere on the business case.
Where these numbers come from, and where they don't
Salaries are sourced. Every headcount figure traces to BLS median wages with the published 1.43× benefits multiplier. The senior tier applies a modeled 22% metro premium, because a mid-market shipper hiring a procurement lead in Detroit, Chicago or Dallas does not hire at the national median.
The 24/7 line is arithmetic. Staffing a single around-the-clock seat requires roughly 4.2 full-time equivalents: 168 hours divided by 40. Not a benchmark. Division.
Three tooling lines are planning assumptions, not benchmarks. Freight audit tooling, rate intelligence and carrier compliance software are custom-quoted, and the per-invoice costs published online trace almost exclusively to companies selling audit software. We would not reproduce numbers we cannot stand behind, so we have shown these as ranges to plan against and labeled them as such. Price them with your own quotes and the model gets better.
Tier 1 remains a defensible floor, and if your business case shows a smaller number than ours, check which tier it priced before you conclude we are inflating. The question that matters is not what the cheapest version costs. It is what the version that delivers your target savings costs, and whether that clears at your freight spend.
Above roughly $14 million, cost stops being the argument. Four other things decide it.
At $25 million of freight a mid-market build is 2.8% to 6.0% of spend. A CFO can approve the low end of that. Which means the honest case against insourcing at the top of the mid-market is not affordability. It is that the function you can afford cannot reach the outcomes you are budgeting for.
One
Timeline
Logistics leadership searches regularly run past 90 days per role. TMS selection adds an RFP cycle. Implementation runs six to nine months at mid-market scope. Team hired, system live, processes stable: 12 to 18 months.
As of August 2026: diesel at $5.652/gal, up $1.94 year over year. Dry van spot and contract rates at parity. Truck postings down more than a quarter. Contract rates forecast ~8% above 2025 and rising through at least mid-2027.
Two
Talent
The in-house model has a single point of failure the org chart does not show: the labor market. Now apply that to a function three to five people deep.
When your procurement lead resigns, and statistically someone will, the carrier relationships, the rate history and the working knowledge of your lanes walk out during a search that regularly exceeds 90 days.
Logistician demand projected to grow 17% through 2034 (BLS). Supply chain turnover runs about a third above pre-pandemic levels (Gartner). If that departure lands mid-bid cycle, the cost is not a vacancy. It is a year of rates.
Three
Liability
This risk did not exist eighteen months ago. On May 14, 2026, the Supreme Court held unanimously in Montgomery v. Caribe Transport II, LLC that negligent carrier-selection claims are not preempted by federal law.
Carrier selection became a litigable duty rather than a best practice. A clean safety rating is not a defense. Documented, defensible vetting is. And somebody has to own it.
$604 million. Dallas County jury verdict, July 2026, in the first major post-Montgomery trial, allocating 23% of fault to the broker that arranged the load. Under appeal. The carrier held a Satisfactory federal safety rating reaffirmed three months earlier.
Four
Position
Here is the ceiling on even a flawless internal build: your team negotiates with the buying power of one company. That is the ceiling, and no amount of talent raises it.
Internal teams bid freight when the calendar or a crisis says so. A portfolio operation benchmarks continuously, because somewhere in the book something is always being bid.
Survey research puts TMS users at roughly 6% freight savings, with expectations of 5–15%. Among managed transportation users, 32% reported 12% or more. The difference is not magic. It is position.
The arithmetic
The math, by freight spend.
| Annual freight spend | Mid-market build $0.7M – $1.5M |
Enterprise grade $1.5M – $3.0M |
Verdict |
|---|---|---|---|
| $5M | 14.0% – 30.0% | 30.0% – 60.0% | Disqualified. The function costs several times the savings. |
| $10M | 7.0% – 15.0% | 15.0% – 30.0% | Disqualified at any credible scope. |
| $15M | 4.7% – 10.0% | 10.0% – 20.0% | Marginal. Only the leanest build clears, and the leanest build under-delivers. |
| $25M | 2.8% – 6.0% | 6.0% – 12.0% | Affordable at mid-market scope. Cost stops being the argument. |
| $50M | 1.4% – 3.0% | 3.0% – 6.0% | Affordable. Position and timeline decide it. |
| $75M | 0.9% – 2.0% | 2.0% – 4.0% | Affordable. Position and timeline decide it. |
| $200M+ | 0.4% – 0.8% | 0.8% – 1.5% | The internal build genuinely pencils, at full enterprise scope. |
Threshold one
Below roughly $14M, the build is disqualified on cost alone.
A credible mid-market function consumes more than 8% of spend, more than the bottom of the entire savings range it exists to capture. At $5 million it costs several times the savings before a single carrier is procured. No CFO approves that, which is exactly why much of the mid-market has historically run freight on spreadsheets and hope.
The threshold moves with scope: the leanest credible build crosses 8% at about $9M, the full mid-market build at about $19M. $14M is the midpoint, and the honest answer for most shippers in this band.
Threshold two
Above roughly $14M, cost stops being the argument, and we won't pretend otherwise.
At $25 million a mid-market build is 2.8% to 6.0% of spend. The low end of that is affordable. The case against it from there is everything above: 12 to 18 months to capability in a rising market, a function three to five people deep against a 17% demand curve, a new and unbudgeted liability function, and a permanent ceiling of single-company buying power.
And note what the second column does to this. Enterprise-grade capability does not clear 8% of spend until roughly $37M, which is why the outcomes are usually budgeted long before the function that produces them can be afforded.
The honest counter-case
When insourcing is the right call.
Credibility requires naming the cases where the build wins. There are three. If you are in one of them, build. As the section below covers, though, that rarely means building for everything.
01
Scale
North of roughly $200 million in annual freight spend, the fixed cost of a full internal function amortizes into defensibility, and dedicated internal leadership pays for itself. Above that band the internal build genuinely pencils for your core network. The section below is about everything that isn't your core network.
02
Freight as the business
If transportation is your product rather than a cost center supporting your product, the capability belongs inside. Do not outsource the thing you sell.
03
Genuinely simple networks
A single-mode, few-lane, stable network may need a strong analyst and a disciplined benchmarking cadence. Not a department, and not a 4PL.
If you are in the $10 million to $100 million band with multi-mode complexity, the arithmetic is the problem. Not your team’s talent. The arithmetic.
If you are above $200 million
At enterprise scale the question changes. It doesn't disappear.
The build pencils at your volume. We just said so, and we meant it. But “own the function” and “own every mode, geography, program and event” are different claims. The second is where capable enterprise logistics teams quietly overextend.
Your team is built for steady-state core freight. It is staffed, budgeted and measured against the lanes that run every week, and it is usually very good at them. What it is not built for is the work that arrives sideways: a program with different economics than your core network, a geography where you have volume but no infrastructure, a mode you touch too rarely to build leverage in, or an event that needs surge capacity and tooling for two quarters rather than two headcount forever.
Hiring permanent capacity for a temporary or specialized need is the same mistake this page documents at mid-market scale. At your volume it just costs more, and it is harder to unwind.
Four shapes a scoped engagement takes
Illustrative scopes, not case studies. The documented results further down this page are full-scope mid-market engagements.
|
Scope By
Program
|
A flow with economics unlike your core network: reverse logistics, core returns, aftermarket and service parts, recalls, or a sustainability program with its own reporting burden. Example shape: battery core returns consolidated across a North American dealer and distributor footprint, where the freight is low-value, high-touch, and nothing like the inbound network your team was built around. |
|
Scope By
Geography
|
A region where you have real volume but no local infrastructure, carrier depth or language coverage. Cross-border and Mexico domestic OTR is the common one. Argus has run operations from Querétaro since 2010, which is the kind of presence that takes an internal team years to replicate for one region of a network. |
|
Scope By
Mode
|
A mode you touch often enough to spend real money on and rarely enough that you never build negotiating position or benchmarking depth in it. Flatbed, temperature-controlled, expedited, drayage, project cargo. Your core team optimizes what it lives in; the outlier modes get managed reactively and priced accordingly. |
|
Scope By
Event
|
Work with a beginning and an end: a large-scale procurement event across hundreds of lanes, a network redesign, a facility move, a post-acquisition integration of two carrier bases. These need tooling, analyst capacity and market data for two or three quarters. Building that capacity permanently to run it once is how run-rate cost gets added to solve a project problem. |
What stays yours
Everything you built the function for.
Strategy, network design, carrier relationships on your core lanes, budget ownership and your team. A scoped engagement is additive, not a handover, and it does not put a provider between your leadership and its own network. If it ever reads that way in a proposal, push back. Ours included.
Why the scoping works
The tooling is already built and already paid for.
The TMS, the freight audit engine, SKU-level cost visibility at 99.7% accrual accuracy, and continuous benchmarking across a $2.5B+ portfolio exist whether your engagement is one program or a whole network. You are not funding a build. And because Argus is non-asset and carrier-agnostic, there is no owned capacity waiting to be filled with your volume.
So the enterprise version of this question is not build or buy. It is which parts of this you actually need to own — and whether the answer is the same for your core network, your outlier modes, your reverse flows and your next procurement event. It usually isn’t.
“But we'd lose control of our data.”
“Execution can be outsourced. Data control cannot.”
We are not going to argue with that. It describes real outsourcing arrangements accurately: data owned and controlled by providers, access that is limited, delayed or filtered, inconsistent formats across partners. Any provider who tells you otherwise is selling.
But notice what it actually indicts. The failure mode is a fragmented estate of execution providers, each holding a slice of your data in its own format, with no obligation to hand any of it over in a usable state. That is not an argument for building a department. It is an argument for an orchestration layer with contractual data terms, which is what the same body of Gartner research recommends.
So the right question in any evaluation is not whether you will lose control of your data. It is what the contract says. Ask for these in writing, of any provider you evaluate, ours included:
- Who owns the data, in the contract, in plain language.
- What fields you receive, in what format, at what latency.
- What happens to your rate and lane history at termination.
- Whether data access is priced separately, and if so, why.
And the fair worry underneath this section deserves an answer too. Handing off execution does not mean losing the muscle. Strategy stays yours. Network design, service trade-offs and budget ownership belong to your leadership under any model worth signing. What a 4PL removes is the layer below: procurement mechanics, daily execution, audit, exception management, reporting infrastructure.
The question that actually decides it
Do you need to own this capability, or do you need the outcome?
Those are different questions with different price tags. The first is a hiring plan. The second is a specification, and it can be met without a 12-to-18-month build. At mid-market scale the answer covers the whole function. At enterprise scale it covers whichever parts your team was never built for.
| In-house build | Argus 4PL engagement | |
|---|---|---|
| Annual run rate | $0.7M–$1.5M modeled at credible mid-market scope; $1.5M–$3M at enterprise grade | Self-funded: documented savings contractually guaranteed to exceed the fee |
| Time to capability | 12–18 months | ~60-day ERP integration across SAP, Oracle and 20+ platforms; first-quarter stabilization |
| Downside if it underperforms | Sunk cost. Restart the build. | Contractual guarantee: documented savings must exceed the fee |
| Key-person risk | Concentrated in 3–5 hires | Spread across a dedicated, named account team and documented process |
| Buying power | One company's volume | $2.5B+ in freight under management; continuous benchmarking |
| Carrier vetting liability | Yours to build, staff and document | Documented vetting on every carrier, as a portfolio function |
| Cost visibility | Invoice totals, unless you build more | SKU-level landed cost by product, lane and mode at 99.7% accrual accuracy |
| Market intelligence | Purchased subscriptions, one vantage point | Live signal across a multi-vertical portfolio |
Argus is non-asset and carrier-agnostic by design, and scoped to the mid-market rather than the enterprise accounts the 4PL category was built around. That is a narrower position by choice, and it is the reason the economics work at a scale where they otherwise wouldn’t. The operating principle is capability, not dependency: everything documented, your team trained, processes and reporting that would keep working if we walked away.
Documented outcomes
What the model produces, industry-blinded.
Tier 1 Automotive Supplier
42% Premium freight reduction
92% Reduction in freight theft claims
99.7% Accrual accuracy, up from under 90%
60 days Full SAP integration
Global Consumer Electronics Brand
19% Rate reduction
99.4% OTIF, up from 91.2%
$3.1M Territory savings
87% Chargeback reduction
Health & Beauty Manufacturer
22% Rate reduction
$5.25M Direct savings
34.2% Freight-to-sales reduction over two years
77% Chargeback reduction
Argus contractually guarantees that documented client savings exceed our management fee. Across the book, clients recover an average of 2.7x the fees they pay. Documented freight spend reductions run 8–22%. Savings are the result of the analysis, integration and management that follow. They are measured and reported, not promised on day one. Average client relationship: 13-plus years.
Before you commit either way
Eleven questions to answer on paper.
With numbers rather than adjectives. Four of them are here. The full set, and the method behind them, is in the paper.
- What is our true annual freight spend, by mode, and how confident are we in that number?
- If our transportation lead resigned tomorrow, what walks out the door, and how long to replace it?
- What is our 12-to-18-month cost of waiting: the margin leaking while a build stands up, at this year's rates?
- Whichever path we choose: is the accountability written down, and what happens if the results do not come?
On this page
The argument
- The headline cost range and the method behind it
- The category error, and the five-point specification
- The full percentage-of-spend table
- The two thresholds: where cost disqualifies the build, and where it stops mattering
- The three cases where building wins
- Four contract questions to ask any provider
In the paper · 21 pages
The worksheet
- The six-role cost model: every role, median base and fully loaded
- The technology line broken out: subscription, implementation, per-integration cost, payback period
- The full August 2026 market table: six indicators, each sourced and dated
- The workforce risk data, with sample sizes
- All eleven questions, and how to price the build with your own data
- Complete sources and method appendix, including where our sources have a commercial interest
Straight answers
Questions we get in this conversation.
What does it cost to build an in-house freight function?
It depends on what you are building, so price it in tiers. A team and a TMS alone runs $476,000 to $1.05 million a year. A four-person core costs approximately $440,300 fully loaded and a six-person core approximately $795,800, calculated from U.S. Bureau of Labor Statistics median wages with the published 1.43× benefits multiplier. A credible mid-market function that can actually audit invoices, benchmark rates, document carrier vetting and cover extended hours runs $0.7 million to $1.5 million. Built to enterprise grade, with true 24/7 coverage, a licensed TMS with ERP integration and a dedicated compliance function, it runs $1.5 million to $3 million. Most insourcing business cases price the first tier and budget for the third.
At what freight spend does building in-house make financial sense?
Below roughly $14 million in annual freight spend, the build is disqualified on cost alone: a credible mid-market function consumes more than 8% of spend, exceeding the bottom of the savings range it exists to capture. The exact threshold moves with scope: the leanest credible build crosses 8% at about $9 million, the full mid-market build at about $19 million. Between roughly $14 million and $75 million the build becomes affordable at mid-market scope, but the decision shifts to timeline, talent concentration, carrier-vetting liability and buying power. Enterprise-grade capability does not clear 8% of spend until roughly $37 million. North of roughly $200 million, the internal build genuinely pencils.
How long does it take to stand up an internal transportation function?
Twelve to eighteen months is the realistic span. Logistics leadership searches regularly run past 90 days per role. TMS selection adds an RFP cycle, and published implementation timelines run six to nine months for mid-market deployments and nine to eighteen months or more at enterprise scope. The cost is not only the wait. It is what the freight market does during it.
What's the difference between a 3PL and a 4PL in this decision?
A third-party logistics provider handles tactical work: storage, shipping, fulfillment. A fourth-party provider acts as a strategic manager, coordinating across multiple 3PLs, carriers and partners on the shipper’s behalf. Gartner’s shorthand: the 3PL is the logistics muscle, the 4PL is the logistics brain. The distinction matters here because most insourcing business cases are reacting to a bad experience with an execution provider, then concluding that all outsourcing works that way. Argus is a non-asset, carrier-agnostic 4PL, which means no warehouses to fill and no margin hidden in your rates.
How does the Supreme Court's carrier-selection ruling change the build decision?
On May 14, 2026, the Supreme Court held unanimously in Montgomery v. Caribe Transport II, LLC that negligent carrier-selection claims are not preempted by federal law, making carrier selection a litigable duty rather than a best practice. In July 2026 a Dallas County jury returned a $604 million verdict in the first major trial that followed, allocating 23% of fault to the broker that arranged the load; that verdict is under appeal. The practical consequence for an in-house build is a function most insourcing business cases never contemplated: continuous authority verification, insurance monitoring, safety scoring and fraud screening, documented well enough to survive a deposition. That is tooling plus headcount plus a retention policy, sitting on a team that already has a day job.
Won't we lose control of our data if we outsource?
That is the strongest objection, and it describes real outsourcing arrangements accurately. But the failure mode it names is a fragmented estate of execution providers, each holding a slice of your data in its own format with no obligation to hand it over usable. The answer is contractual, not structural: ask who owns the data in plain language, what fields you receive in what format at what latency, what happens to your rate and lane history at termination, and whether data access is priced separately. Ask it of every provider you evaluate, including Argus.
When should we build instead of partnering?
Three cases. Scale: north of roughly $200 million in annual freight spend, dedicated internal leadership pays for itself. Freight as the business: if transportation is your product rather than a cost center supporting your product, the capability belongs inside. Genuinely simple networks: a single-mode, few-lane, stable network may need a strong analyst and a disciplined benchmarking cadence rather than a department or a 4PL.
We spend well over $200 million on freight and have a full internal team. Is there any reason to talk to a 4PL?
Yes, but not for what you already do well. At that volume your core network belongs inside, and this paper says so plainly. Where a scoped 4PL engagement earns its place is the work your team was not built around: a program with different economics than your core flows (reverse logistics, core returns, service parts, recalls), a geography where you have volume but no local infrastructure such as cross-border or Mexico domestic OTR, an outlier mode you spend real money on but never build negotiating position in, or a finite event like a large procurement bid, network redesign or post-acquisition carrier integration. Those are capability spikes, and hiring permanent headcount to cover a spike is the same mistake this page documents at mid-market scale.
Can a 4PL engagement be scoped to one program, region or mode rather than our whole network?
Yes. A scoped engagement covers a defined slice (one program, one geography, one mode, or one event with a start and end date) while strategy, core network design, budget ownership and your team stay where they are. It works because the infrastructure is already built: the TMS, the freight audit engine, SKU-level cost visibility at 99.7% accrual accuracy and continuous benchmarking across a $2.5B+ portfolio exist whether the engagement is one lane set or an entire network, so you are not funding a build to get access to them. Argus is also non-asset and carrier-agnostic, so there is no owned warehouse or fleet capacity that benefits from absorbing your volume.
Do I have to talk to sales to get the paper?
No. The paper arrives by email and it is yours to use in your business case whichever way you decide. We will follow up once. If a conversation is useful you’ll tell us; if the paper convinces you to hire three people and buy a TMS license, that is a legitimate outcome and the paper says so in Section 10.
A freight operation that has grown faster than the strategy managing it.
That describes most of the mid-market, and it is why this question keeps coming up. The paper prices both paths honestly. Then you decide.
Already know you want the numbers from your own network? Request a complimentary freight analysis. Documented, yours to keep, no obligation. Above $200 million in spend, it can be scoped to a single program, region or mode rather than the whole network.