A new router is one of the most emotionally satisfying ways to respond to bad Wi-Fi. The box has a larger number on it, the antennas look serious, and there is a decent chance it will make at least one speed-test screenshot look better.

It is also often the wrong first move.

"Bad Wi-Fi" is a bucket that holds several different failures: a weak signal in one room, radio congestion at a particular time of day, a laptop that clings to the wrong access point, an overloaded internet connection, slow DNS, or a wired link that quietly fell back to 100 Mbps. A new router can accidentally mask one of these problems. It cannot reliably diagnose it.

The useful approach is to turn the complaint into a small observation: which device, in which location, doing which activity, at what time, and compared with what known-good path? Once you have that, the upgrade decision gets much less mystical.

Start by naming the failure

Before opening an app or shopping tab, capture a few details when the problem occurs:

• Is it one device or every device?
• Does it happen in one room, everywhere, or only while moving around the house?
• Is the failure slow downloads, high latency, a dropped video call, or a website that takes a long time to start loading?
• Is a wired computer affected at the same time?
• Is it repeatable at a particular time of day?

Those answers separate whole-network failures from local ones. A laptop with poor signal in a back bedroom does not establish that the router is bad. A wired desktop and every wireless device becoming laggy whenever a backup starts points toward the internet edge or bufferbloat, not radio coverage.

Keep the test modest. You are not trying to build a home NOC. You are trying to avoid changing five things and learning nothing.

Establish the wired baseline first

Connect a laptop or desktop directly to the router or switch with Ethernet if you can. Then run a speed test and a simple latency check while the network is quiet and while it is busy.

ping -c 30 1.1.1.1

This does not diagnose every internet problem, but it tells you whether latency and loss are already poor before Wi-Fi enters the picture. If the wired path is bad, work outward from the modem, router CPU, WAN service, or a saturated upload before buying access points.

For local throughput, iperf3 is more useful than an internet speed test because it removes your ISP from the experiment. Run a server on a wired machine:

iperf3 -s

Then run a client from the problem laptop while it is near the access point and again where it normally misbehaves:

iperf3 -c 192.0.2.10 -P 4

Use your server's real private address, not the documentation address above. You are looking for a meaningful difference between locations, not a heroic number. If the near-access-point test is solid and the distant test collapses, you have a coverage or placement problem. If both are mediocre while the wired baseline is fast, inspect the Wi-Fi configuration and local contention.

Separate coverage from congestion

Signal strength is not the same thing as usable Wi-Fi, but it is a good first discriminator. Walk to the problem area and inspect the network details on the device. A weak signal, frequent band changes, or a connection that falls back to 2.4 GHz when you expect 5 or 6 GHz are clues that placement matters.

Coverage fixes are usually physical:

• Move the access point out of a cabinet, away from a television, and above dense furniture.
• Put it closer to the middle of the area it serves rather than at one edge of the home.
• Avoid putting the only access point in a basement utility room because that is where the internet service enters.
• Add a wired access point where the failure actually occurs when one well-placed device cannot cover the space.

Congestion behaves differently. The signal might look strong, but performance degrades at busy hours or when many nearby networks overlap. Use your router's radio view or a Wi-Fi analyzer to see which bands and channels are busy. On 2.4 GHz, use the familiar non-overlapping channel plan where it applies; on 5 GHz and 6 GHz, favor a sensible automatic plan unless you have enough observations to justify manual tuning. Randomly locking a wide channel can make a neighborhood problem worse.

Test the device and the roaming behavior

The client is part of the network. Older laptops, phones, and IoT devices have different radio capabilities, drivers, and power-saving behavior. If only one client is unreliable, update its operating system and Wi-Fi driver, forget and rejoin the network if appropriate, and compare it with a newer device in the same spot.

Multi-access-point homes introduce a different failure: sticky roaming. A device can stay attached to a distant access point longer than you expect, especially if the network has uneven placement or inconsistent SSIDs and security settings. Do not solve this by adding another mesh node at random. First check whether every access point uses the same network name, authentication, and current firmware, and whether the new node would have a clean wired backhaul.

A wireless mesh hop can be useful when cabling is impossible, but it consumes radio airtime to carry traffic between nodes. Wired backhaul remains the boringly good answer because it lets each access point use its radio for clients. Wi-Fi 8 vs. Wi-Fi 7: Why Reliability Matters More Than Another Speed Claim explains why the next standards label does not change that physics.

Do not mistake DNS or bufferbloat for Wi-Fi

If a speed test is fine but new sites hesitate before loading, test DNS separately. Try resolving a known hostname from the affected device and compare it with another resolver only as a diagnosis, not as a permanent ritual:

dig example.com

Slow or unreliable DNS can make a healthy connection feel broken. So can bufferbloat: latency rises sharply when an upload or download fills the WAN queue. Start a large upload, cloud backup, or download, then run ping again. If latency turns ugly only under load, look for smart queue management or QoS features on a router that can actually sustain them at your service speed. The right fix may be traffic shaping, not a faster wireless radio.

Also inspect the mundane wired pieces. An Ethernet cable, switch port, or powerline adapter that negotiates at 100 Mbps can create a ceiling that looks suspiciously like weak Wi-Fi. Check link speed in the router or switch interface before replacing a perfectly capable access point.

Make the smallest justified change

After the tests, choose the change that fits the evidence.

There are good reasons to buy a new router: unsupported firmware, insufficient wired ports, a router CPU that cannot handle your internet service, a needed 6 GHz or Wi-Fi 7 client upgrade, or an architecture that cannot put access points where they need to be. A measurable failure is much better justification than an impressive aggregate throughput claim.

Keep one page of notes

Write down the location, client, test result, and change. That tiny record prevents you from repeatedly rediscovering that the guest-room problem happens only when the door is closed or that the real bottleneck is a backup job. It also makes a future upgrade rational: you can see whether the problem is coverage, contention, backhaul, or the edge.

Home networking gets better when you treat it like any other production system: define the symptom, establish a baseline, isolate a variable, and make the smallest change that removes the failure. The result may be a new router. More often, it is a better-placed access point, one Ethernet run, or a configuration adjustment that costs much less and works immediately.

For more practical systems advice, visit Slaptijack.