How We Killed 5 Hours of Monthly Downtime for an Accounting Firm

"Every month-end our server dies." Cost: R12,000 in lost billable hours. Cause: Nobody monitored it. Here's how we fixed it in 90 days.

The Before: Break-Fix Chaos

The R950 Audit Findings [23-Point Check]

• Critical: Switch CRC errors — failing hardware causing network drops
• Critical: No UPS on server — load-shedding = hard crashes, database corruption
• High: 0% disk monitoring — C: drive at 98%, Pastel crashing from no space
• High: Backups failing silently for 6 months — external HDD full
• Medium: 15 unused M365 licenses = R1,800/mo wasted

The Fix: 90-Day Plan

• Month 1: Replace switch [R4,000], add UPS [R3,000], setup PRTG monitoring
• Month 2: Migrate Pastel to Azure VM, decommission old server
• Month 3: Retire 15 licenses, enable disk alerts, document all passwords in client portal
• Total cost: R15,000 once-off + R2,000/mo managed [10 users]

The After: 6 Months Zero Outages

• Downtime: 5 hours/month → 0 hours
• Billable hours recovered: R12,000/month
• IT cost: R20,000/mo → R2,000/mo
• ROI: 6x in first year
• Client quote: "We didn't realize IT could be boring. That's the goal."

The 3 Root Causes We See Everywhere

• 1. No monitoring: "If it ain't broke..." until month-end and it is
• 2. Ancient hardware: "It still works" until it doesn't during SARS deadline
• 3. No documentation: IT guy leaves, passwords die with him, nobody can login