I build software for a living, and a fair share of the advice I give is not to. People are sometimes surprised to hear a builder say it, but I pay for products of my own as well as building for clients, so I know what it costs to build the wrong thing, and I would sooner tell someone before they spend the money than explain it afterwards. These are the situations in which I say no, and what I usually suggest in its place.

When the Process Is the Problem

The most common reason is that the trouble sits in how the work is done, and software would only make a poor process run faster. If an order needs four approvals because nobody ever removed the three that stopped mattering, the answer is to remove them, which costs nothing. I ask every client to walk me through the work before we talk about systems, and it is normal for the first few improvements to need no code at all. We have set out how to map an operation so that you can make the same check yourselves.

When Something on the Shelf Already Does It

Accounting, payroll, email and a great many other jobs are done the same way in every company, and the products made for them are better than anything I could justify building for one client. If the work is standard and nobody is retyping data to get round the tool, keep the subscription. I have written separately about where the line falls between one platform and ten subscriptions.

When the Business Is Still Changing Shape

A bespoke system fixes a way of working in place, which is its strength when that way of working is settled and a liability when it is not. A company that is still changing its offer or its structure needs tools it can pick up and drop cheaply. I tell those founders to come back once the same process has held steady for a good while, because that is the point at which there is something worth building around.

When Nobody Will Own It

Every system needs somebody inside the business who cares whether it is used properly, decides what changes and answers the questions of the people using it. Where I cannot find that person I advise against building, however strong the case looks on paper, because software without an owner is first ignored and then blamed.

When the Numbers Do Not Add Up

Some problems are real and still too small to justify a build. If a task costs one person an hour each week, a system to remove it will take a long time to pay for itself, and the money would do more good elsewhere. I work the figures through with the client using their own numbers, and I am content for the answer to be that a spreadsheet and a better habit will do.

What I Recommend in Its Place

What I findWhat I usually recommend
A process with steps nobody needsChange the process first, then look again
A standard job done badly by the current toolBuy a better tool, or set up the existing one properly
A team that is not using what it already hasTrain the people before replacing the software
Routine work that follows a patternHand that one job to AI inside the systems you already run
Work that is yours alone, held back by every tool you have triedBuild it

Why I Can Afford to Say It

Advice from somebody who only earns when you build will lean towards building, and clients know it. Our consultancy is paid for as advice, the report belongs to the client, and nobody is obliged to build anything with us afterwards. I also take a longer view than one project. My longest client relationship has run for more than twelve years, and relationships like that rest on being told the truth, including when the truth is that you should keep your money.

When I Do Say Build

I say it when the work is what sets the business apart, when the same information is being typed into several systems, when the tools have become something the team works around, and when somebody inside is ready to own the result. By then the client has heard me say no to other things, which is why the yes carries some weight.

BY BANKS consultancy tells you what to build, what to buy and what to leave alone before you spend on any of it, from engineers who have built these systems themselves. If you are about to commit to a project and want a second opinion first, get in touch and I will tell you what I would do.