Custom software vs off the shelf

We build custom software, so read what follows with that in mind. It still argues that most companies should buy, because most of the time they should, and because a client who should have bought is a client who resents the invoice.

The odds, on both routes

The costs of buying that never appear in the licence quote

• Implementation. Frequently one to three times the first year of licence fees, and usually charged by a partner rather than the vendor.
• Per-seat growth. The price you agreed is for the headcount you have now. Check what it does at twice that, because that is the version you will be paying.
• The tier you actually need. Connecting to your other systems (listed as API access), a log of who changed what, and signing in with your company accounts are routinely held back for a higher plan, and you will discover which one in month two.
• Data migration. Getting what you already hold into the product, in the shape it expects.
• Training, and the productivity dip while people learn a system built around somebody else’s process.
• The annual increase. Ask what it has been for the last three years rather than what the contract caps it at.

The costs of building that never appear in the build quote

• Maintenance, at a benchmark of 15 to 25 per cent of the build cost per year. Across a full lifecycle it typically exceeds the original build, which is set out in what custom software costs.
• Hosting and third-party services. Modest, and never zero.
• Change, because the business will move and the software has to follow.
• Key person risk. A build depends on somebody understanding it, which is an argument for documentation and for owning the system outright rather than an argument against building.

Five questions that settle most cases

1. 1Is the awkward part of your process a reason clients choose you? If it is merely awkward, buy something and spend the difference on growth.
2. 2Have you actually looked at the products, or at one product two years ago? This market moves. The thing that did 70 per cent in 2024 may do 90 per cent now.
3. 3What percentage does the best product cover, and is the missing part the part that matters? A product covering 70 per cent where the missing 30 is your whole differentiator is worse than one covering 50 per cent of things you do not care about.
4. 4What would it cost to bend the product into shape? When configuration, consultancy and integration approach the cost of a build, the build is the better purchase and it fits properly.
5. 5Can you live with somebody else’s roadmap? Buying means the thing you need next year arrives when the vendor decides, or never.

The third option most people miss

Build or buy is a false pair. The third answer is buy, and build the part the product cannot do.

Keep the accounting package, the CRM and the e-commerce platform. Build the layer that connects them, or the one screen your operation needs that none of them provides. This is the cheapest route to a system that fits, it keeps the vendors carrying security and compliance for their own products, and it is what a large share of our work actually is.

It is also the option a supplier selling a full build has no reason to raise.

Where buying clearly wins

Accounting, payroll, email, file storage, e-commerce, and anything where the rules come from outside your business. HMRC decides what payroll does, not you, so there is nothing to differentiate and a product will always be cheaper and more correct.

The same goes for anything a regulator changes regularly. A vendor amortises that maintenance across thousands of customers. You would carry it alone.