Deployment automation is unusual among software projects: the scope is well understood, the work is finite, and the result is visible on the first day. It is also the project most often postponed, because nothing is currently on fire.
What the standard job contains
For one application with one production environment:
- A pipeline that builds, checks and deploys on every push.
- Staging and production, genuinely separate.
- A containerised build, so what was tested is what runs.
- A rollback, performed once with your team watching.
- Monitoring and alerts that reach a person.
- Backups, with one restore actually carried out.
- A runbook and a handover.
Two weeks of one experienced person, which lands in the low thousands. Compared with almost anything else in software, it is cheap — and it is the thing that makes everything after it faster.
What pushes the number up
Several services that must deploy together. Coordinating three deployments is more than three times one.
An application that cannot be containerised as it stands. Hardcoded paths, state on local disk, configuration compiled in. Fixable, but it is application work rather than infrastructure work and should be quoted as such.
Servers nobody can rebuild. If the current machine was configured by hand two years ago by someone who has left, the first job is reproducing it from a file. That is often the most valuable part of the engagement and it takes time.
Compliance requirements. Audit trails, restricted access, data residency. All doable, none of it free.
What should not push it up
A cluster you do not need. Monitoring with two hundred alerts. A dashboard product with a per-seat licence. Each of these appears in quotes regularly, and each is a monthly cost for a problem most small products do not have.
The comparison worth making
Work out what a deployment costs you today: the hour or two of the person who does it, the evening it happens, the changes that wait because nobody wants to deploy on a Friday, and the last incident where the fix was slow because nobody was sure what was running.
Set that against two weeks of work once. For most teams the arithmetic is not close, which is exactly why this is the project that gets postponed until an incident makes it urgent — and doing it under pressure costs the same money and considerably more sleep. Done calmly it is two weeks, from $2,400, on your own accounts, with the pipeline yours at the end of it. Keeping it current afterwards is a separate, optional arrangement, covered by our software maintenance services.