Nobody had ever added the bills up in one place.
Eleven active subscriptions across four departments, each one signed by whoever needed it, most renewing silently. The first week of work was not deployment. It was arithmetic.
The two “keep renting” rows are the important ones. Accounting and payroll carry regulatory change the vendor absorbs on your behalf every year — that is continuing value, and continuing value is exactly what rent is for.
One capability at a time, each one live before the next one started.
Video, and only video
One system, deployed on their own cloud account, with recording, private sharing, transcription and team permissions. Ninety-six seats moved across in nine working days. Nothing else was touched.
- Deployed to their cloud, their domain, their branding
- Storage lifecycle rules set before the first upload
- Two admins on their side, trained and documented
What it proved: the thing they were afraid of — “we have no engineers, who fixes it at 2am” — turned out to have a boring answer. It didn’t break.
Knowledge and files follow
The wiki and the client file store moved next, onto the same infrastructure, with the same identity provider and the same backup rules. Two systems, one login, one place to look when something is wrong.
- Single sign-on across all three owned systems
- Nine years of documents migrated with history intact
- Restore from backup tested in front of their ops lead
What went wrong: the first migration pass mangled folder permissions for one department. We found it, said so in the weekly note before they noticed, and re-ran it. Two days lost.
Workflows, then the keys
The last month was the part that only makes sense once you own the systems: wiring them to how the firm actually works. Then a handover week, and we stopped.
- Client walkthrough recording → transcript → project log entry
- Intake form → ticket → owner assigned, no inbox in the middle
- Runbook, credentials and cost model handed to their ops lead
The test: in week twelve we answered nothing. Two support requests came in that month; their own ops lead closed both from the runbook.
The bill stopped growing with headcount.
That is the structural change, and it matters more than the first year’s saving. Four owned systems now cost what the infrastructure costs. Hiring twenty people next year adds twenty salaries and no software line.
Ownership is not free, it is fixed.
They pay for infrastructure every month, and they paid us once per system to set it up. The win is that neither number moves when they hire. Anyone quoting you a saving without naming the running cost is selling, not advising.
Two systems we told them to keep renting.
Accounting and payroll. Both carry statutory change their vendor tracks continuously, both have real consequences for getting a filing wrong, and neither has a mature self-hosted equivalent worth the operational risk. Owning those would have been ideology, not economics.
The whole method is that question, asked once per capability, honestly: is the vendor still creating value here, or are we just paying rent on something that stopped changing?
Start with one system. The audit is the part that tells you whether the rest is worth it.
Client details are generalised at their request; the sequence, the failures and the numbers are as they happened. We will walk you through the unredacted version on a call if you are seriously considering this.