
Build, Buy and the Option Nobody Prices
The comparison is usually between an internal build costed optimistically and a product costed accurately. The third option, doing nothing differently, is rarely priced at all.
- Build versus buy comparisons systematically understate build cost by omitting maintenance, integration ownership and staff turnover.
- The status quo is an option with a cost, and it is almost never priced in these comparisons.
- The right comparison is over several years, since the cost profiles differ in shape rather than only in size.
- Reversibility has value and rarely appears in the analysis at all.
- A published estimator can frame the range before scoping conversations begin.
Every finance function considering AI runs some version of this comparison, and most of them run it badly in the same three ways.
The build option is costed as a development project rather than as a permanently owned system. The buy option is costed accurately, because a vendor has quoted it. And the option of continuing as you are is treated as free, which it is not.
Fixing those three things does not predetermine the answer. It just makes the comparison describe reality.
Why build costs get understated
Not through optimism about the development, which teams are increasingly realistic about. Through omission of everything that happens afterwards.
A built system needs maintenance as ERPs change and interfaces are updated. It needs somebody who understands it, which becomes a recruitment and retention problem the moment that person leaves. It needs security patching, monitoring and someone answering questions about it at close. And it accumulates a backlog of small improvements that nobody has budgeted for.
Those costs are real, recurring and rarely in the comparison. A build costed for year one against a subscription costed for year one is comparing a project to an ongoing service.
The correction is straightforward: cost the build over the same period as the alternative, including the people who will own it in year two and year three.
What belongs in a complete comparison
Four cost categories that a proper comparison includes and a typical one omits.
All four apply to a build and only some apply to a purchase, which is the asymmetry that makes a naive comparison misleading rather than merely incomplete:
The third option
Build versus buy is a false binary, because continuing as you are is always available and is what happens by default when a decision stalls.
It has a cost, and pricing it changes the shape of the comparison.
What it costs today
Hours spent on the process, contractor and overtime spend at peak, error correction, late reporting and the opportunity cost of experienced people doing assembly work. All measurable, and rarely measured.
It is the real baseline
Both alternatives should be compared against this rather than against each other. A change that is expensive against a build may be obviously worthwhile against the status quo.

Comparing over the right period
The cost profiles differ in shape, which is why a single year comparison misleads regardless of which option it favours.
Three years is usually the right horizon.
| Option | Year one | Years two and three |
|---|---|---|
| Build | Development effort, concentrated and visible | Maintenance, ownership, integration changes and key person risk, all recurring and easy to omit |
| Buy | Subscription plus implementation | Subscription, with renewal exposure and less control over the roadmap |
| Status quo | Current run cost, already in the budget | The same, plus growth in volume without a corresponding change in cost per unit |
The status quo column is the one most comparisons leave blank, which makes the other two look like a choice between two costs rather than two investments.
Reversibility has value
A rarely discussed dimension is how expensive each option is to exit, and the differences are substantial.
A subscription can generally be ended, with the pain being migration and process disruption. A built system is not exited so much as maintained or abandoned, and abandoning it strands the investment and leaves the process needing a replacement urgently.
That matters most where the requirement is genuinely uncertain, which is common in this category. Where you are unsure what you actually need, the reversible option carries a value that a pure cost comparison does not capture, and it is worth stating explicitly in the paper rather than leaving it as an unspoken preference.
A build costed for year one against a subscription costed for year one compares a project to a service.
ChatFinRunning the comparison properly
Four steps, and the first one is where most of the missing information is.
Price the status quo
Hours, contractor spend, overtime, error correction and the opportunity cost of experienced staff on assembly work. It is the baseline both alternatives should be measured against.
Cost the build over three years
Including ownership, integration maintenance and key person risk. Year one development cost alone is not a comparison, it is a project estimate.
Include opportunity cost
What the engineering capacity would otherwise deliver. It is usually the largest omitted item and it is a real trade.
State the reversibility position
Explicitly, especially where the requirement is uncertain. It belongs in the paper rather than in the unspoken preferences of whoever writes it.

Where ChatFin sits, stated plainly
ChatFin is the buy option, and it is worth saying that directly rather than presenting a neutral framework with a thumb on the scale.
What that means concretely: a configurable platform that connects to NetSuite, SAP Business One, JD Edwards, Acumatica, Dynamics 365 and Sage Intacct through their supported interfaces, with integration maintenance, model changes and platform upgrades handled rather than owned by your team.
It also means the trade offs that come with buying: configuration within a product's boundaries rather than unlimited freedom, a roadmap you influence rather than control, and a renewal decision every term.
For teams that want to see the range before any conversation, the published cost estimator produces a first year range and compares an internal build, a consulting build and a configurable product side by side, without recommending one.
AI Cost Estimator for the CFO Office
Four inputs, a first year cost range, a timeline, and three delivery models compared. No login required.
Frequently Asked Questions
Why do build cases understate cost?
Mainly through omission of what happens after the project. Ownership, integration maintenance, key person risk and opportunity cost are all recurring and are frequently absent from a comparison that costs development against a subscription.
Should the status quo be in the comparison?
Yes, and it usually is not. It is the real baseline, it has a measurable cost in hours, contractor spend, error correction and opportunity cost, and it is what happens by default when a decision stalls.
What horizon should we use?
Three years, because the cost profiles differ in shape rather than only in size. A build front loads visible cost and back loads recurring cost, and a single year comparison misrepresents whichever option it favours.
When does building make sense?
Where the requirement is genuinely specific to your organisation, where you have durable engineering capacity to own it, and where the process is stable enough that maintenance will not consume the benefit. Those conditions are real and are less common than build cases usually assume.
General information only. It is not legal, accounting or tax advice, and rules change. Check your own position with your advisers before acting on it.
Compare against the status quo, not against each other.
Cost the build over three years including ownership and opportunity cost, price the status quo, and state the reversibility position. The answer follows from an honest comparison rather than from the framework.
ChatFin is the buy option and says so, with the trade offs stated. The published estimator compares all three routes without recommending one.
Book a DemoSee the platform
Compare the routes against your own processes.
Bring your current run cost for one process. We will show what the configurable route covers, what it does not, and how it compares against the baseline you already carry.