How to Speed Test Home Network

Speed testing a home network sounds simple until you realize most people are testing the wrong thing. Running a browser-based test to a public server tells you about your ISP connection — it says nothing about whether your router is bottlenecking clients at 300 Mbps, whether your switch is negotiating at 100M instead of 1G, or whether your Wi-Fi is handing off traffic through a congested 2.4 GHz backhaul. A meaningful speed test strategy requires testing at multiple points in the network stack: WAN throughput, LAN switching speed, wireless client throughput, and inter-device transfer rates — separately and deliberately.

The gear you use matters here. A modern Wi-Fi 7 router like the TP-Link Archer BE800 is rated for up to 19 Gbps aggregate across its 6 GHz, 5 GHz, and 2.4 GHz bands, but you’ll only see that potential on a client that supports Wi-Fi 7 and sits close to the access point. Meanwhile, a bottlenecked switch port or a misconfigured QoS policy can cap your real-world speed far below what your ISP delivers. The only way to know what’s actually limiting you is to test systematically — and to interpret the numbers correctly.

This guide covers the tools, methods, and gear decisions that affect network speed testing: what to run, where to run it, how to isolate variables, and which networking equipment gives you the visibility and throughput headroom to actually hit the speeds you’re paying for.

Quick Comparison

Understanding What You’re Actually Testing

Before running a single test, be precise about which segment of the network you’re measuring. There are four distinct layers where throughput can degrade: the WAN connection (modem to ISP), the router’s NAT/forwarding engine, the switching fabric (wired LAN), and the wireless radio subsystem.

WAN throughput is what browser-based tools like Speedtest.net, Fast.com, or the nperf CLI measure. These hit public servers over the internet, which means latency, server load, and peering agreements all influence the result. For multi-gigabit plans (1 Gbps+), run tests to at least three geographically distributed servers and discard outliers. On a 2.5G or 10G symmetrical fiber plan, a router with only a 1G WAN port will hard-cap your result — no software tweak fixes that.

LAN throughput is best measured with iPerf3, an open-source tool that runs a direct bandwidth test between two endpoints on your network. Install the server on a desktop or NAS (iperf3 -s) and the client on a laptop or second machine (iperf3 -c <server-IP>). A correctly wired gigabit segment should consistently show 940–950 Mbps. Below 850 Mbps suggests a duplex mismatch, a degraded cable, or a port auto-negotiating at 100M. Drop to 10–20 Mbps and you’ve found a FastEthernet device or a bad SFP.

Wireless throughput is the most variable. Signal strength, channel width, band steering, client capability, and interference all interact. For a clean wireless test, run iPerf3 over Wi-Fi with the server wired directly to the router or AP. This isolates the wireless hop cleanly. A Wi-Fi 6 client (802.11ax) with a 2×2 MIMO configuration on 80 MHz channels should hit 600–800 Mbps in good conditions. A Wi-Fi 6E or Wi-Fi 7 client on 160 MHz or 320 MHz channels can clear 2 Gbps — but only if the AP supports it and the RF environment cooperates.

TP-Link Archer BE800

The Archer BE800 is a tri-band Wi-Fi 7 router with a 6 GHz band rated at 11520 Mbps, a 5 GHz band at 4324 Mbps, and a 2.4 GHz band at 1376 Mbps, giving an aggregate spec of 19,220 Mbps. For speed testing purposes, what matters most is the port layout: it ships with one 10G WAN/LAN port, one 2.5G port, and four 1G LAN ports. If your ISP delivers multi-gig fiber, the 10G port lets you benchmark WAN throughput without the interface becoming the bottleneck.

The BE800 supports Multi-Link Operation (MLO), the Wi-Fi 7 feature that bonds multiple bands for a single client connection. In a speed test scenario, a Wi-Fi 7 client connecting via MLO across the 6 GHz and 5 GHz bands simultaneously can achieve lower latency and higher aggregate throughput than any single-band connection. The router also supports 4K-QAM, which increases spectral efficiency by roughly 20% over the 1K-QAM ceiling of Wi-Fi 6E.

For a power user who wants to benchmark the full capability of their internet plan and their wireless clients, the Archer BE800 provides the switching and radio headroom to avoid becoming the measurement artifact. It runs on a quad-core 2.6 GHz CPU, which matters for sustaining NAT at 10G speeds without CPU-bound packet loss. View on Amazon.

ASUS RT-AX88U Pro

The RT-AX88U Pro is a Wi-Fi 6 (802.11ax) router with a combined 6000 Mbps spec across its dual bands: 4804 Mbps on 5 GHz and 1148 Mbps on 2.4 GHz. The LAN side is unusually well-equipped for a consumer router — eight gigabit ports and a 2.5G WAN port. For networks where most devices connect over wired GbE, that 8-port LAN means you can skip a separate switch and still run iPerf3 between multiple wired endpoints simultaneously.

ASUS’s ASUSWRT firmware exposes real-time traffic statistics per client, adaptive QoS with per-application bandwidth allocation, and AiMesh support for multi-node Wi-Fi. For speed testing purposes, the built-in network diagnostic tools (ping, traceroute, nslookup) give you a starting point, but iPerf3 running directly to a NAS or desktop will give you the clean numbers ASUSWRT’s bandwidth monitor can’t. The router runs on a 1.8 GHz quad-core processor with 256 MB NAND and 1 GB RAM — enough to run OpenVPN at ~500 Mbps without crushing throughput.

At 2.5G WAN, the RT-AX88U Pro will saturate a gigabit fiber plan without any headroom loss. If your ISP eventually upgrades to a 2.5G tier, the port is already there. Wi-Fi 6 clients connecting on 160 MHz channels in the 5 GHz band can realistically hit 1.2–1.8 Gbps in close-proximity tests — numbers that approach what you’d see on an older Wi-Fi 6E device without the 6 GHz spectrum. View on Amazon.

TP-Link Deco BE85

The Deco BE85 is TP-Link’s flagship Wi-Fi 7 mesh system, sold as a two- or three-pack. Each node carries the same tri-band spec as the Archer BE800 (19 Gbps aggregate), but the mesh architecture changes the speed testing calculus significantly. In a mesh system, backhaul is the variable that most people ignore: if two nodes are linked over a shared radio band, any client test on that node is constrained by backhaul throughput, not just the access radio.

The BE85 addresses this with a dedicated 6 GHz backhaul radio and wired backhaul support via its 10G and 2.5G Ethernet ports. When nodes are wired together, the backhaul ceases to be a bottleneck and each node functions effectively as a standalone AP backed by a switched connection. For speed testing, this means you can run iPerf3 from a wireless client on a satellite node to a wired server on the primary node and get results that reflect wireless performance only — not wireless plus congested backhaul.

For larger homes where a single router can’t provide coverage, the BE85 is the most accurate way to deploy Wi-Fi 7 without degrading throughput at distance. Each node’s 10G port can connect to a dedicated switch or upstream router, making the system compatible with enterprise-style flat-network designs where your router and your APs are separate devices. View on Amazon.

Ubiquiti UniFi U7 Pro

The UniFi U7 Pro is a ceiling-mount access point supporting Wi-Fi 7 across three bands: 6 GHz (U-NII-5, U-NII-6, U-NII-7), 5 GHz, and 2.4 GHz, with an aggregate spec of 9.3 Gbps. It ships with a 2.5G PoE+ uplink (802.3at, 30W) and supports 320 MHz channel width on the 6 GHz band — a Wi-Fi 7 capability that doubles the spectral width available to Wi-Fi 6E devices. For users who want to isolate the wireless variable from the routing variable, deploying the U7 Pro behind a dedicated router and managed switch is the cleanest setup for repeatable speed tests.

What differentiates UniFi gear for power users is the controller software. UniFi Network (self-hosted on a UDMP, UDM, or a VM running UniFi Network Application) gives you per-client throughput history, RF environment charts, channel utilization data, and the ability to see exactly which radio a client is associated with and at what MCS rate. When your iPerf3 result comes in lower than expected, the controller tells you whether the client is on the wrong band, at a poor signal level, or competing with hidden-node interference.

The U7 Pro requires PoE+ (802.3at) from a compatible switch — the TP-Link TL-SG108E is a budget-friendly managed switch for VLAN segmentation but doesn’t supply PoE. For PoE deployment, you’d want the NETGEAR GS316EP or a TP-Link Omada PoE switch upstream. The U7 Pro is priced for prosumer deployments but delivers the RF analytics and hardware headroom to make your speed test data meaningful rather than anecdotal.

TP-Link TL-SG108E

The TL-SG108E is an 8-port gigabit smart switch with 802.1Q VLAN support, port-based QoS, port mirroring, and IGMP snooping. At its price point it has no meaningful competition for structured home lab switching. For speed testing, it provides two specific capabilities: the ability to segment traffic onto isolated VLANs (so your iPerf3 test isn’t competing with streaming traffic) and port mirroring, which lets you send a copy of any port’s traffic to a monitoring device running Wireshark or ntopng.

The switch operates at line rate on all 8 ports simultaneously — 1 Gbps full duplex per port, 16 Gbps switching capacity total. It’s fanless, draws under 6W at full load, and is managed via a Windows utility or web interface. There’s no CLI, which makes it unsuitable for scripted monitoring workflows, but for a home network where you need VLAN separation between a trusted LAN, an IoT segment, and a guest network, it handles the job without the complexity of an enterprise switch.

For users running iPerf3 between multiple endpoints, placing them all behind the TL-SG108E with VLAN isolation ensures that your test traffic stays on the switching fabric and doesn’t traverse the router’s NAT engine — which would introduce CPU overhead and skew your LAN throughput numbers. View on Amazon.

Who Should Buy Which Gear

Archer BE800 vs. RT-AX88U Pro: If your ISP delivers more than 1 Gbps or you have Wi-Fi 7 clients, the Archer BE800’s 10G port and MLO support make it the correct choice. If you’re on a 1G plan and have no Wi-Fi 7 clients yet, the RT-AX88U Pro’s 8-port LAN and mature ASUSWRT firmware give you more practical flexibility for less money. The RT-AX88U Pro is also the better choice if you run iPerf3 benchmarks frequently between wired endpoints — having 8 native GbE ports means fewer switches in the chain.

Deco BE85 vs. UniFi U7 Pro: The Deco BE85 is for users who need mesh coverage and want a single-vendor, self-contained system. Setup is simpler, backhaul is managed automatically, and 10G wired backhaul is available if you want it. The UniFi U7 Pro is for users who already have or plan to build a separated router/switch/AP architecture, want RF analytics, and are comfortable managing a UniFi controller. If you’re running speed tests to diagnose performance issues (rather than just benchmarking), the UniFi ecosystem’s per-client data makes root cause analysis dramatically faster.

TL-SG108E for everyone who has one router: If you have a single router with 4 LAN ports and you’re trying to run multi-endpoint speed tests, the TL-SG108E is the missing piece. It doubles your port count, adds VLAN segmentation for cleaner test isolation, and gives you port mirroring for traffic inspection — all without fan noise or complexity.

iPerf3 Quick Reference

For readers unfamiliar with iPerf3 syntax, here are the commands that cover 90% of home network speed test scenarios:

• Basic TCP test (1 stream): iperf3 -c <server-IP>
• Parallel streams (better for high-throughput links): iperf3 -c <server-IP> -P 8
• UDP test (useful for wireless jitter/loss): iperf3 -c <server-IP> -u -b 1G
• Reverse mode (test download from server to client): iperf3 -c <server-IP> -R
• 10-second test with JSON output: iperf3 -c <server-IP> -t 10 --json

Run the server with iperf3 -s on any machine with a static IP. On Windows, the executable runs identically. On a Synology NAS, iPerf3 can be installed via the SynoCommunity package repository — which makes a Synology DS423+ a convenient always-on iPerf3 server that doubles as your file storage.

Bottom Line

For most home networks, the speed bottleneck is either the wireless hop or a router CPU being asked to do too much at once — not the ISP connection. Run iPerf3 between a wired client and a wired server first to establish a LAN baseline, then compare wireless results to that number to quantify the Wi-Fi penalty. If you’re building or upgrading gear specifically to eliminate throughput ceilings, the Archer BE800 handles multi-gig WAN and Wi-Fi 7 without either becoming the constraint. If you need coverage across multiple rooms, the Deco BE85 with wired backhaul is the only mesh configuration that doesn’t reintroduce the bottleneck you’re trying to measure around.