Why Incidents Spike for Months After a Vendor Switch
It's common to hear that incidents actually went up for three to six months after a company switched maintenance (SM) vendors. The cause is rarely the new team's skill. It's usually gaps in the handover.
You have the source code, but it won't build: the repository doesn't match what's running in production, or internal libraries and build scripts are missing.Nobody has production access: cloud root accounts, domain registrar logins, and SSL certificate accounts often turn out to be tied to a former engineer's personal email.Knowledge that lived in one person's head: a month-end settlement fix-up query run by hand, or special-case logic for a single client. Nothing is written down, so it shows up as an outage in the first month after the switch.Assets to Secure Before the Transition
Code and Deployment
The complete, current source code and the branching strategy (which branch actually ships to production)Build and deployment runbooks, CI/CD configurationEnvironment variables, service accounts, API keys, and certificates, with the expiry date for eachData and Operations
Database schema and recent change history, backup schedule and restore procedureA list of batch jobs (run times and what to do when one fails)A list of external integrations (payment gateways, SMS providers, public APIs), plus contract owners and renewal datesIn practice, we recommend turning this into a checklist of 50–80 items and tracking each one in three stages: received, verified, and actually run by us. If you've only received something, you don't yet know whether it works.
Reverse-Engineering a System With No Documentation
When documentation is thin, start with the traces the system leaves behind.
Logs and traffic: Group the last 90 days of access logs by URL to see which features people really use and which have been abandoned. Most traffic often lands on just 20–30% of the screens.Batch history: Go through crontabs and scheduler logs to find the jobs nobody remembers creating.Code analysis: Map the calls between modules and every outbound call to build a dependency map. Flag files that change often but have no tests as high-risk zones.Designing the Parallel-Run Period
Overlap and Clear Ownership
Keep the outgoing and incoming vendors working side by side for at least 4–8 weeks. Write down, week by week, who handles incidents first and how issues get escalated. That's how you avoid "we thought they were handling it."
Validation Scenarios for the Incoming Team
Build the system and deploy it to staging and production on their ownRestore the database from a backupHandle simulated incidents (a failed batch job, an expired certificate) without outside helpAs a rule, don't end the outgoing vendor's contract until the new team has passed all three.
What Your Contract Should Cover
Handover obligations at termination: Put the list of deliverables, their formats, and deadlines in a contract appendix.Cooperation clause: Require the outgoing vendor to answer questions for a set period after termination (for example, 30 days), and agree on what that costs.Source code ownership: Make sure copyright in all deliverables belongs to your company.Access revocation: Using the account inventory, require written confirmation that all access has been revoked and all passwords and keys rotated.The 90-Day Roadmap After the Transition
Days 1–30, set up monitoring: Add alerts for availability, error rates, and batch job results first, so no outage goes unnoticed.Days 31–60, fix urgent technical debt: Start with the items most likely to cause incidents: certificates about to expire, end-of-life runtimes, and missing backups.Days 61–90, run an improvement backlog: Put requests and improvements into a prioritized backlog and share progress in a monthly report.Switching to SM With POLYGLOTSOFT
POLYGLOTSOFT's SM subscription also takes over systems from other vendors. Before we start operating a system, we review the asset checklist, reverse-engineer the system from its logs and code, and confirm we can deploy and restore it ourselves. Plans scale with your needs, from Basic SM at about $100/month to Pro SM with a dedicated PM. If you're thinking about changing vendors, tell us about your current system through our contact page or KakaoTalk channel. We'll start by pinpointing the biggest handover risks with you.
