Skip to content
Start a Project
Software Development

Build vs Buy: A Practical Framework for Business Software

Every build-versus-buy decision is really a question about where your business is actually differentiated — and most businesses overestimate how many places that is.

Illustration generated in code — replace with a project screenshot or photograph where one exists.

A software company advising you to build custom software has an obvious conflict of interest. So here is the honest version: for most processes in most businesses, buying is correct. Custom software earns its cost in a narrow set of circumstances, and it is worth being able to recognise them.

The core test

The useful question is not "does a product exist for this?" — one almost always does. It is: would adopting that product's way of working make us worse at something customers notice?

If the answer is no, buy. Adapt your process to the tool, accept the mild friction, and spend the saved budget elsewhere. If the answer is yes — if the process itself is part of why customers choose you — then the tool is asking you to give up an advantage in exchange for convenience, and that trade is worth examining.

Four signals that favour building

The process is the product

Some businesses are essentially a process with a company wrapped around it: a specialist logistics operation, an unusual underwriting model, a manufacturing workflow refined over two decades. Forcing that into generic software flattens exactly the thing that makes it work.

The integration cost exceeds the build cost

A product that needs six connectors, a middleware layer and a nightly reconciliation job to fit your environment is not cheaper than building. It is more expensive, with the cost moved into a category that is harder to see.

Per-seat pricing scales against you

Per-seat and per-transaction pricing is excellent at small scale and punishing at large scale. If a five-year projection shows licence costs crossing the build cost, the decision has a numerical answer rather than an architectural one.

The data is the asset

If the data a system accumulates is strategically valuable — pricing history, operational telemetry, customer behaviour — then control over its structure, portability and access matters more than the convenience of a hosted tool.

Four signals that favour buying

  • The process is a commodity. Payroll, accounting, email, calendars, helpdesks. Thousands of companies have solved these. None of them are why your customers chose you.
  • Compliance is involved. Certifications, audit requirements and regulatory reporting are expensive to build and expensive to maintain. Buying transfers that burden to a vendor whose entire business is carrying it.
  • You have no one to own it. Custom software needs a maintainer. If there is no plan for who handles it in year three, you are building a future problem.
  • The requirement is still moving. If the process changes monthly because nobody has settled it yet, building now means building the wrong thing precisely.
The short version

Buy the things every business does. Build the things only your business does. The mistake is rarely choosing wrong on a single system — it is failing to notice that you have started building commodity software by accident.

The third option people forget

Build-versus-buy is a false binary. The most common correct answer in practice is buy the platform and build the layer that makes it yours.

That means using established products for storage, authentication, payments, messaging and infrastructure, and building only the workflow, interface and integration logic that encodes how your business works. The custom code stays small, focused and genuinely differentiating, and the undifferentiated heavy lifting stays someone else's problem.

Practically, it looks like this: a managed database, an off-the-shelf identity provider, a payments platform, and a custom application that orchestrates them around your process. You are building perhaps 15% of the system, and it is the 15% that matters.

Counting the real cost

Build estimates are routinely wrong because they count construction and ignore ownership. A fair comparison includes:

  1. Initial build
  2. Hosting and infrastructure, annually
  3. Maintenance — dependency updates, security patches, platform changes
  4. Support, including out-of-hours if the system is operationally critical
  5. Onboarding and documentation for new staff
  6. The cost of change: what a significant new requirement costs in year two

Against a bought product, the equivalent list is licence cost, configuration, integration, and the cost of working around the gaps. Both lists are longer than the number that usually gets quoted in the meeting.

Making the decision stick

Whichever way it goes, write down why. Build-versus-buy decisions get revisited constantly, usually by people who were not in the original conversation, and usually at the moment of maximum frustration with whatever was chosen. A short document recording the constraints and the reasoning at the time saves that argument from being had from scratch every eighteen months.

Keep reading