A mid-sized e-commerce company ships 800,000 orders a year. Its transport bill is roughly $3.2 million a year (illustrative figures). The operations director wants a TMS, and the TMS selection decision falls to the CFO. The vendor quotes $180,000 a year for a subscription. The IT lead says, "We can build it ourselves for $400,000." The CFO asks the only question that matters: over three years, which one is actually cheaper — and by how much?
Most build-vs-buy debates never get past that first exchange, because nobody runs the full numbers. Let us run them.
TMS Selection: The Three-Year TCO, Build vs Buy
The table below is a worked example — assumed figures for a company of this size, labelled as such, to show the method. Replace every number with your own quotes before deciding anything.
| Cost item, 3-year total | Build in-house | Buy off-the-shelf |
|---|---|---|
| Development / licence & subscription | $400,000 build | $540,000 ($180,000/yr) |
| Implementation & carrier integration | $120,000 | $90,000 |
| Hardware / cloud hosting | $90,000 | Included |
| Maintenance & support team | $270,000 | Included in subscription |
| Training | $30,000 | $45,000 |
| Upgrades & new features | $150,000 | Included |
| Three-year total | $1,060,000 | $675,000 |
In this example, buying wins by about $385,000 over three years. But the table is not the answer — it is the starting point. Two things usually flip it. First, the build estimate is almost always low: internal builds run late, and every quarter of delay is a quarter of paying the old manual process. Second, the subscription quote is almost always low too, because it excludes the three hidden costs below. Run your own table twice — once with the vendor's numbers, once with your team's honest numbers — and compare the gap, not the headline.
The Three Hidden Costs Everyone Forgets
Carrier integration. A TMS is only worth something when it talks to your carriers — rating, labels, tracking, proof of delivery. Each carrier connection takes real engineering time, and carriers change their APIs without asking you. In the example above I allowed $90,000–$120,000 for integration; companies with six or more carriers regularly spend double that in year one. Ask the vendor, in writing, which carriers are included in the subscription and what each additional connection costs.
Training and process change. A TMS changes how dispatchers work, not just what screen they look at. Budget for two things: formal training, and four to eight weeks of lower productivity while the team learns the new flow. The second one never appears in a vendor quote and it is usually the bigger number. A team of ten dispatchers running 10% slower for six weeks costs more than the training course itself.
Upgrades. With a bought system, upgrades are the vendor's problem — until the vendor deprecates the version you run on, or charges for the module you actually need. With a built system, upgrades are your problem forever: every carrier API change, every tax rule change, every new warehouse is a ticket in your backlog. Over three years, I have seen in-house systems where upgrades alone cost more than the original build.
TMS Selection: Who Should Build, and Who Should Just Buy
The rule is simpler than vendors admit. Buy, without hesitation, if you ship fewer than roughly 500,000 orders a year, run fewer than 50 vehicles or carrier accounts, and your transport flows look like everyone else's. Your needs are standard, the market has already solved them, and building a standard product is paying to reinvent a wheel.
Consider building only when at least two of these are true: your transport logic is genuinely unusual — say, temperature-controlled multi-drop routes with customer time windows that no packaged TMS models well; you already have a strong internal engineering team with spare capacity; or transport is your core competitive advantage, not a cost centre. Even then, build only the unusual part. Buy the standard core — rating, dispatch, tracking — and build the one module that makes you different. The hybrid route is where most sensible companies land, and it is the option vendors mention last.
One more test. Ask your team: if the two engineers who would build this leave next year, who maintains it? If there is no good answer, you do not have a build option. You have a future crisis with a project plan attached.
When you sit through a vendor demo, ignore the polished screens and watch three things. First, ask them to connect one carrier you actually use, live, and run a shipment from rating to label — not a recorded video. Second, ask what happens when a carrier's API goes down: does the system fail loudly, or does it fall back to a backup rating path? Third, ask how your data leaves the system — if you switch vendors in three years, can you export every shipment, invoice and proof of delivery in full? Vague answers on any of these three mean the money you saved on the quote will come back as cost somewhere else. The cheapest demo is the one that survives these three questions.
The Break-Even Formula
Here is the calculation to take into the meeting. Let B be the three-year cost of building, S the three-year subscription cost, and H the three-year hidden costs of buying (integration, training, upgrade extras). Buy wins when S + H < B. The break-even point is where they are equal.
Work it as a sensitivity, not a single number. Take the vendor's subscription and raise it 20% — vendors raise prices. Take your build estimate and raise it 30% — builds overrun. Take the hidden costs and double them — they always grow. If buying still wins under all three stresses, the decision is made. If the answer flips under stress, you do not have a decision yet; you have a negotiation. Go back to the vendor with the stressed numbers and ask what they will fix in the contract: price lock, included integrations, upgrade guarantees.
In the example at the top, buying won by $385,000 on headline numbers. Stress it — subscription up 20% ($108,000 more), hidden costs doubled (another $90,000) — and buying still wins by about $187,000. That is a decision you can defend in a board meeting. If your numbers do not survive the stress test, do not sign.
The cheapest TMS is not the one with the lowest quote. It is the one whose three-year total, honestly counted, is lowest — and whose bill you can still explain in year three.
By Lao Zhou, Logistics Cost Analyst







