SaaS is the right default until it is not. The tipping point is rarely the license fee. It is the hours spent copying between tools, the report nobody trusts, and the process that only one person understands. Another seat does not fix that. A smaller system that owns the job can.
Count the hops, not the features
If a weekly process touches four products and a spreadsheet, you are paying in attention. Map the hops. Export, reformat, re-enter, chase a status. Those steps are the real cost. Custom software is worth it when one screen can close the loop — intake, status, output — without a new login.
- The tool is used for one job, but that job is core. You should own the spine.
- The vendor roadmap will not include your exception. Your exception is the business.
- Integrations already exist, but nobody wants to maintain the glue. Own the glue.
Build the spine, not a clone of the suite
Do not rebuild the CRM. Rebuild the path that the CRM never quite covered. Keep the systems that are good enough. Replace the seam. That is how custom work stays small enough to ship and cheap enough to keep. We would rather a boring internal tool that operators finish in one sitting than a platform with a name and a dashboard nobody opens.
Keep an exit that is not a hostage note
Your data should leave. Exports, APIs, and a schema you can read. If a vendor change would freeze the operation, you already needed a system of record you control. Custom software is not a trophy. It is a decision that the job is yours — and the tools around it can stay rented.
