Microsoft 365 network performance tuning

When users complain about Microsoft 365 feeling slow, the network is usually the cause. Understanding the network performance principles and the diagnostic tools turns "the cloud is slow" into a specific fixable problem.

The cardinal principles

Recap of Microsoft's published network connectivity principles:

Identify Microsoft 365 traffic — know what to optimise.
Egress locally — local internet breakout, not backhaul through HQ.
Avoid hairpins — don't bounce Microsoft traffic through remote inspection.
Bypass inspection for Optimize endpoints — Microsoft already inspects.

Tenants that follow these have dramatically better Microsoft 365 experience than tenants that don't. Any performance investigation should start here.

Diagnostic tools

Microsoft 365 Network Connectivity portal

The Network Connectivity dashboard in the Microsoft 365 admin center collects telemetry from real users to show what they actually experience:

Latency to Microsoft 365 endpoints per location.
Comparison to a Microsoft baseline for similar geographies.
Bandwidth estimates.
Connection quality scores.

Surface for "is Microsoft 365 actually slow for our users in Singapore, or is it just one user's complaint?"

Microsoft 365 Network Connectivity Test

The connectivity test at connectivity.office.com runs a point-in-time test from the current user's location:

DNS resolution timing.
Network latency to specific Microsoft 365 endpoints.
Local egress vs proxy detection.
TLS handshake timing.
HTTP performance.

Useful for diagnosing specific user complaints — have them run the test, share results.

Teams Network Planner

For Teams-specific scenarios, the Teams Network Planner helps size network capacity for expected user count and meeting volume — useful for new offices or capacity-planning exercises.

Endpoint Analytics

For per-device performance signals, Intune Endpoint Analytics surfaces:

Microsoft 365 Apps performance per device.
Boot and sign-in time trending.
Anomaly detection for sudden performance changes.

Useful for ruling out endpoint-side issues vs network issues.

Common patterns

"It works at home but not at the office"

Usually means office network is the bottleneck:

Backhaul to HQ before reaching Microsoft.
Proxy / firewall inspection of Microsoft 365 traffic.
Saturated office internet.
DNS lookups going through slow resolver.

Fix: implement local egress, bypass inspection for Optimize endpoints, scale office internet.

"It works at the office but not at home"

Usually means home network or remote-user policies are the issue:

VPN routing all traffic through corporate.
ISP performance problems.
Wi-Fi congestion.

Fix: split-tunnel VPN to bypass for Microsoft 365 Optimize endpoints, or use Entra Global Secure Access for identity-aware modern access.

"Teams meetings are choppy"

Teams media is especially sensitive to latency, jitter, packet loss:

End-to-end latency under 100ms ideal, under 200ms acceptable.
Jitter under 30ms.
Packet loss under 1%.

Fix: prioritise Teams traffic with QoS markings, ensure local egress, avoid inspection, optimise Wi-Fi if endpoints are wireless.

"OneDrive sync is slow"

OneDrive sync is bandwidth-sensitive rather than latency-sensitive:

Initial bulk sync of many files takes time on slow links.
Sustained sync can saturate available bandwidth.

Fix: configure delivery optimisation in OneDrive sync client, schedule large operations off-peak, ensure office bandwidth is sufficient.

Specific configurations

For most enterprise networks, these specific items matter:

DNS — use fast resolvers (the Microsoft 365 NCSI service recommends Microsoft's, but local resolvers work fine if fast).
Proxy bypass — Optimize endpoints bypass any proxy.
TLS inspection bypass — Optimize endpoints bypass any TLS-inspecting middlebox.
QoS — DSCP marking for Teams media (real-time / EF), other Microsoft 365 (assured forwarding).
Capacity — appropriate uplink bandwidth for office size.
Wi-Fi — current-gen Wi-Fi with proper RF design.

When the network is fine and it's still slow

Sometimes the issue isn't network:

Endpoint performance — old laptops, full disks, antivirus conflicts.
Microsoft service incident — check Service Health.
Specific Microsoft 365 features — certain capabilities (Microsoft 365 Copilot, especially) involve more latency than basic features.
User behaviour — running 50 browser tabs and three video apps simultaneously.

For Microsoft 365 customers serious about user experience, network performance is one of the highest-leverage investments. The diagnostic tools have matured; the principles are well-documented. Most performance complaints have specific, fixable network causes.