Most of your stack should be bought. Some of it is already paid for and just needs fixing. A small part should be built. Knowing which is which is the whole game.
We build custom solutions for general contractors, so you might expect us to say build. We don't. Accounting, email, payroll, document storage: buy them, and buy the boring, well-supported option. A custom build earns its place only in a narrow band of workflows. Here is how to find that band, and how to spot the problems that need neither.
Buy when the workflow is a commodity.
If a hundred other contractors do it the same way you do, someone has already built a good product for it. Your edge is not in how you run payroll. Take the product, take the support, move on.
Fix when you already own the tool.
A lot of the pain we see is not missing software. It is software you already pay for that was never set up for how your team works: a PM tool that does not talk to accounting, approvals nobody configured, fields nobody fills in. Before you buy or build anything, check whether a few days of setup and a connection between two systems would solve it. Often it does.
Build when the workflow is your edge.
Some processes are the reason you win work or protect margin. The way you buy out a job. The way you level bids. The way a super files a daily. Off-the-shelf tools flatten those into a generic version, and your team ends up working around the software instead of through it. That is the signal.
Three questions before you decide
- Does the current tool force workarounds? Spreadsheets next to the system are a workaround.
- Does the same data live in more than two places? Every copy is a place for the truth to drift.
- Would you do this workflow differently from your competitors if the software let you? If yes, the software should let you.
What owning it means
When we build, the contractor owns the code, the data and the roadmap. No per-seat surprises, no feature held behind a tier, no vendor deciding your workflow changed. That ownership is the real value of a build.