Home › Blog › Build, buy or outsource

Build, buy or outsource — what to decide on

Start this decision from cost and you will usually get it wrong, because the three options incur cost at different times. Look first at differentiation, rate of change, and whether you can operate it.

3 criteria Decision table Common misjudgements

The short answer

Three criteria: is this capability part of your competitive position, how often does it change, and can you run it? Those three answers largely determine the option.

Cost is not the decision criterion; it is the verification step after. Choose the approach, then compute twelve-month total cost and check it against the budget.

The three criteria

Differentiation — is this a competitive capability? — If it connects directly to why customers choose you, build it or outsource it under your control. Work every company does the same way — accounting, mail, scheduling — should be bought.

Rate of change — how often do the rules move? — Frequently changing rules run into the configuration limits of off-the-shelf tools. Unchanged for years, and building it leaves you only with maintenance.

Operating capacity — can you actually run it? — Maintaining is harder than building. If one person owns it and it stops when they leave, building is a liability, not an asset.

Decision table

Situation	Recommended	Why
Standard work, changes rarely	Buy	Cheapest and fastest. Most cases land here
Standard work, changes often	Buy plus configuration/automation	Buy the tool and wire it with a workflow layer
Differentiating work, changes rarely	Outsource the build	Built well once, it lasts
Differentiating work, changes often	Build in-house (or embedded contractors)	Speed of change is the competitive advantage
Differentiating work, no in-house capability	Outsource with a handover plan	Put source, documentation and account transfer in the contract
Genuinely unclear	Buy and trial for six months	Decide again once the real requirements surface

Four common misjudgements

"Our process is too specific for off-the-shelf tools" — Usually what is specific is one or two steps, not the whole process. Handling the rest with a bought tool and attaching just those steps is cheaper and more stable.

"Building it means no subscription fee" — The subscription goes away and maintenance, infrastructure and staff time arrive. On a twelve-month total, building is usually more expensive.

"Outsourcing means we own it" — You own it only if the contract names source code ownership, documentation, the deployment environment and account transfer. Without those clauses, switching later is very expensive.

"We only need to build the AI part" — The AI part is usually a fraction; integration, permissions, logging and review screens are the rest. Scope the build narrowly around "the AI" and the remainder grows outside the quote.

The verification after deciding

Once the approach is chosen, compute twelve-month total cost. Buying is subscription plus usage plus review time; outsourcing is build plus maintenance plus infrastructure; building is staff time plus infrastructure plus opportunity cost.

Then look at the cost of reversing. If this decision is wrong in six months, what gets thrown away? For buying, usually little. For building, a lot. That asymmetry is why high uncertainty argues for starting by buying.

Finally, write the stop condition in advance: "if this metric is not at this level in three months, we revisit the approach." Without that sentence, a wrong approach gets pushed indefinitely.

Frequently asked questions

The off-the-shelf tool is missing one feature we want.

Establish whether that one feature is competitive or merely convenient. If convenient, adapting the process to the tool is cheaper; if competitive, look at attaching just that piece. Building everything because of one gap is the most expensive option available.

Can we mix outsourcing and building?

It is a common arrangement: outsource the initial build, bring operations and improvement in-house. The success condition is the handover — settle documentation, code review participation and account ownership at contract stage.

We have developers. Is outsourcing still better?

Having developers and having capacity are different things. Ask whether pausing their current work for this is the best use of them. Opportunity cost frequently exceeds the outsourcing fee.

Which side do AI capabilities usually fall on?

The capability itself is usually bought — models and tools improve fast enough that a home-built version falls behind quickly. What remains is the part that connects to your data and process, so "buy the model, build the connection" is the realistic shape.

Compare the costs

Compare total cost across all three

The outsourcing cost calculator estimates a range by feature scope; the ROI calculator gives you the twelve-month total for the buy option to compare against.

Estimate cost