How to Test Your Mesh Wi-Fi Speed (Eero, Nest, UniFi)
How to test your mesh Wi-Fi speed the right way: use iPerf3 to measure real throughput at each node and the backhaul between them, not your ISP.
You added a second or third mesh node to cover the far bedroom. The app on your phone shows full bars everywhere and a green checkmark. So is the far room actually fast now, or just “connected”? Bars and a happy app dashboard tell you a device is associated with a node. They say nothing about how many megabits per second actually make it through. To know that, you have to measure throughput — and a Speedtest won’t do it.
The short answer
To test mesh Wi-Fi speed, run iPerf3 between a wired device at your main node (the server) and the phone in your hand (the client), then walk to each node and room and run the test. The number you read is the real throughput from that spot, back through the mesh, to your network core — the ISP is never involved. Comparing those numbers room by room shows you exactly which node or which backhaul hop is dropping your speed, instead of guessing from signal bars.
Why Speedtest can’t test a mesh
A mesh system’s whole job is the local path: your phone to a node, that node to another node (the backhaul), and finally to the gateway. A Speedtest measures the opposite thing — the path from your device all the way out to a server on the public internet. If your ISP plan is 500 Mbps, every room will read about 500 Mbps on Speedtest even if one node is internally crawling at 80 Mbps, because the ISP link is the cap, not your mesh.
The moment you want to compare nodes, validate a new satellite, or prove the back bedroom improved, you need a LAN speed test, not an internet speed test. iPerf3 measures throughput between two endpoints you control, so when both sit inside your house, the result is your mesh’s real performance.
What actually limits mesh speed
Three different hops can be the bottleneck, and they fail in different ways:
- Client to node (the radio link). Your phone’s Wi-Fi to the nearest node. Limited by distance, walls, band (2.4 vs 5 vs 6 GHz), and how many devices share that node.
- Node to node (the backhaul). How satellites talk back to the main unit. This is the one people forget. If the backhaul is wireless, every hop roughly halves throughput, because the node is talking to your phone and to the next node on the same radios. If it’s wired (Ethernet backhaul) or uses a dedicated band, the penalty mostly disappears.
- Node to gateway (the core). The main node to your router/modem. Usually wired and rarely the problem, but worth ruling out.
A single number from one spot can’t separate these. A set of numbers, taken room by room, can.
What you need
iPerf3 runs as a client/server pair. For mesh testing, put the server on a wired device plugged directly into your main node — a Mac, a NAS, a mini PC, anything on Ethernet. The client is the phone you carry from room to room.
- Server: a wired device at the main node. Wired matters — if the server is on Wi-Fi too, you double the wireless variance and can’t tell which link is slow.
- Client: your phone, running the iPerf3 client where you’re standing.
- Both on the same network, port 5201 open.
Running iPerf3 on a Mac as the server and an iPhone or iPad as the client is the most convenient setup, because one purchase covers both.
Step by step
1. Start the server at the main node
On the wired device next to your main mesh unit:
- Open iPerf3 Client & Server, switch to Server mode, leave the port at 5201, tap Start.
- Note its local IP (something like
192.168.1.20).
2. Walk the house with the client
On your phone, in Client mode, enter the server IP and port, pick TCP, and run a test in each location that matters:
- Next to the main node (your baseline — this is the best the mesh can do).
- Next to each satellite node.
- In the rooms where people actually use Wi-Fi: the far bedroom, the patio, the garage.
Run each spot two or three times and write down the average. Thirty seconds per spot; the whole house takes about five minutes.
3. Build a quick table
| Location | Throughput |
|---|---|
| At main node | (baseline) |
| At node 2 | |
| Far bedroom (via node 2) | |
| Patio (via node 3) |
The shape of the numbers tells the story, not any single value.
Reading the result
What’s “normal” depends on the backhaul:
| Mesh setup | Realistic phone throughput near a node |
|---|---|
| Wired (Ethernet) backhaul, Wi-Fi 6 | 500 Mbps to 1+ Gbps |
| Dedicated-band backhaul (tri-band), Wi-Fi 6/6E | 400 to 800 Mbps |
| Shared wireless backhaul, one hop | 200 to 400 Mbps |
| Shared wireless backhaul, two hops out | 80 to 200 Mbps |
If the number at the main node is high but each satellite is much lower, your backhaul is the bottleneck — that’s the classic shared-wireless-backhaul tax. If one specific room is low but its node tests fine up close, that room is a Wi-Fi dead zone and the node needs to move closer.
Pin down the mesh bottleneck
The mesh isolation trick
Test once standing right next to a satellite node, then again in the room beyond it. If both are low, the backhaul to that node is the limit — moving the node won’t help, you need wired backhaul or a closer placement to the main unit. If only the far room is low, the client radio link is the limit and the node needs to be nearer the room.
A few more moves that turn guesses into answers:
- Wire one satellite. If you can run Ethernet to a single node and its throughput jumps, you’ve proven the wireless backhaul was the cap — and you know exactly what to fix.
- Test before and after moving a node. Save the session, move the node a few meters, test again. Now “I think it’s better” becomes “the far bedroom went from 110 to 260 Mbps.”
- Try parallel streams. Single-stream TCP often understates modern Wi-Fi. Set 4 or 8 parallel streams in the app to see what the link can really do.
- Check 6 GHz separately (Wi-Fi 6E/7 mesh). Stand close and confirm the 6 GHz band actually delivers its headline speed before blaming the mesh.
Why a native app for this
iPerf3 on the command line works, but mesh testing is the one job where walking around with a phone is the entire point. A native iPerf3 app on iOS, iPadOS, macOS, and Android lets you:
- Run the test from the exact spot you care about, phone in hand.
- Save each room’s result and compare sessions over time — before vs after adding a node, or before vs after switching to wired backhaul.
- Export the whole survey as CSV or JSON for a coverage map or a note to your ISP.
The wire protocol is the standard iperf3, so a phone reading and a homelab dashboard reading mean the same thing.
Bottom line
A mesh dashboard tells you a device is connected. iPerf3 tells you how fast that connection actually is — node by node, room by room. The next time you add a satellite, switch to wired backhaul, or move a node six feet, run the test before and after. You’ll see the real delta in megabits, not bars, and you’ll know whether the upgrade did anything at all.