Custom software vs SaaS: when should you build your own?

Should you build custom software or buy SaaS?

Buy the commodity and build the difference.

If thousands of businesses have your problem in roughly the same shape, a product already exists, and more engineering has gone into it than your budget will buy. Build where the way you work is what makes you money, and no product models it without you distorting the business to fit the software.

Most companies end up with both. The interesting question is usually not which one, but where the line sits.

Buy, almost always, for

Accounting. Payroll. Email and calendar. Helpdesk. Document storage. Payment processing. General purpose CRM. Marketing email. These are solved, commodity, and competitively priced, and nothing about doing them your own way makes you money.

Rebuilding any of these is the most expensive mistake available in this category, and it is usually made because one feature was missing.

Build when

• The workflow is the product. If how you quote, schedule or deliver is what customers choose you for, software that forces the average version of it costs you the thing you compete on.
• Nothing fits without distortion. Not nothing is close. Nothing fits without you changing how the business runs in ways that cost real money.
• The integration tax has passed the build price. Five tools and a person moving data between them has a running cost. When that cost, annualised, approaches the build, the build is cheaper within two years.
• You are the product. If you sell software, you build software.

The hybrid most companies should consider first

Keep the products. Build the connective layer.

A custom system that owns your records and your workflow, integrating the accounting package and the calendar and the payment processor rather than replacing them, is smaller, cheaper and far less risky than a system that tries to be everything. It also leaves you able to swap a vendor later.

This is what most of our business systems work actually is, and it is usually a focused build at $25,000 to $60,000 rather than a platform.

The costs people forget on each side

Buying looks cheaper and often is, but the running total includes per seat pricing that grows with headcount, tier jumps that arrive at the worst moment, integration tooling to make products talk, and the staff time spent on the gaps between them. Plus the risk that the vendor changes direction, raises prices or is acquired.

Building has a visible number at the start and then needs hosting, maintenance and someone to change it when the business changes. Software that nobody is allowed to modify becomes a constraint within about two years, which is why a service plan is part of an honest quote rather than an upsell.

A test worth applying before you call anyone

Write down the three things your current tools stop you doing. If all three are features, buy a better product. If all three are about how the work moves between people and systems, that is where custom software earns its price.

What to do next

If you already know the shape, the configurator prices a build without a conversation. If you want the question answered honestly rather than sold to, describe the problem: the plan will say which side of the line you are on, including when the answer is buy something.