This playbook is distilled from 300 VoIP migrations our team has managed over the past decade. It is not theoretical — every recommendation comes from watching what works and what fails in production environments.
Phase 1: Discovery (Week 1)
What Most Companies Skip
Everyone audits their phone bill. Almost nobody audits their call patterns. Before choosing a provider or architecture, answer these:
- What is your peak concurrent call count? (Not average — peak.)
- What percentage of calls are internal vs. external?
- Do you have seasonal spikes? (Tax firms in April, retailers in December)
- How many remote workers make business calls daily?
These answers determine your architecture, bandwidth requirements, and provider selection more than any feature comparison.
The Network Assessment Nobody Does Correctly
Run these tests at 10 AM on a Tuesday, not at 6 PM on a Sunday:
Required measurements:
Bandwidth: Minimum 100 Kbps per concurrent call
Jitter: Must be < 30ms (test with iperf3 -u)
Packet loss: Must be < 0.5% (test with 1000+ ping)
Latency: Must be < 150ms one-way to provider DC
Test to your actual VoIP provider's data center, not to 8.8.8.8. The path to Google is irrelevant — the path to your SIP registrar is what matters.
Phase 2: Preparation (Week 2)
Network Configuration
QoS Policy (apply to all network egress points):
Voice RTP: DSCP EF (46) — priority queue, 30% bandwidth reservation
SIP Signaling: DSCP CS3 (24) — guaranteed bandwidth, 5% reservation
Video: DSCP AF41 (34) — weighted fair queue
Data: DSCP BE (0) — best effort, remaining bandwidth
VLAN Configuration:
VLAN 100: Data (all workstations, servers)
VLAN 200: Voice (all IP phones, softphone traffic)
VLAN 300: Management (switches, APs, SBC)
The SIP ALG Problem
Disable SIP ALG on every router and firewall in the path. This single configuration change resolves approximately 40% of the "VoIP doesn't work" tickets we see post-migration. SIP ALG modifies SIP headers in ways that break NAT traversal, cause one-way audio, and drop calls after 30 seconds.
Phase 3: Migration (Week 3-4)
The Parallel Run
Never cut over everything at once. Run old and new systems simultaneously for at least one week:
- Day 1-2: Migrate IT department (they can troubleshoot their own issues)
- Day 3-4: Migrate one business department
- Day 5-7: Monitor, fix issues, document solutions
- Day 8-10: Migrate remaining departments in waves
- Day 11-14: Number porting completes, old system decommissioned
Training That Actually Works
Two sessions, 45 minutes each:
Session 1 (all staff): Make a call, transfer a call, check voicemail, use the mobile app. Nothing else. These four tasks cover 95% of daily phone use.
Session 2 (admins only): Add a user, create a ring group, change the auto-attendant, pull a call report. These four tasks cover 95% of admin needs.
Skip the feature tour. Nobody remembers a 2-hour training session covering 40 features they will never use.
What We Have Learned
After 300 migrations, three truths:
Network problems cause 80% of post-migration complaints. Fix the network before migration, not after.
The provider matters less than the implementation. A mediocre provider with excellent implementation beats an excellent provider with rushed deployment.
Users adapt in two weeks. The first week feels chaotic. By week three, nobody wants to go back. Do not panic during week one.
For organizations evaluating providers, VestaCall assigns a dedicated migration team to every deployment — which, based on our experience, is the single most important factor in migration success.
Compiled by the DialPhone Limited deployment team. Data from 300 enterprise VoIP migrations, 2016-2026.
Top comments (0)