Picking Wi-Fi channels without guesswork
Automatic channel selection works most of the time. When it doesn't, a quick survey and a few simple rules beat random switching.
Most routers choose a Wi-Fi channel automatically, and most of the time that is fine. But in apartment buildings, where a single phone can see twenty or thirty networks, "auto" sometimes picks badly and then sticks with it. Here is the short process I use before touching any settings.
Step 1: look before you change anything
Use any Wi-Fi analyser app on a laptop or Android phone and walk around the rooms where you actually use the network. Write down which channels the strongest neighbouring networks occupy. Signal strength matters far more than the total number of networks: one neighbour at −55 dBm hurts more than ten at −85 dBm.
Step 2: 2.4 GHz — only 1, 6 or 11
In the 2.4 GHz band, channels overlap heavily. Only 1, 6 and 11 (in most regions) are fully separate from each other. Choosing channel 3 or 9 means you overlap with two groups of neighbours at once instead of sharing cleanly with one. Keep the channel width at 20 MHz; 40 MHz in this band mostly creates interference for everyone, including you.
Step 3: 5 GHz — width is a trade-off
The 5 GHz band has far more room. An 80 MHz channel gives the highest speeds, but in a crowded building a 40 MHz channel that nobody else uses can be faster in practice than an 80 MHz one shared with three neighbours. DFS channels (roughly 52–144) are often empty because many routers avoid them; they're worth trying if your devices support them, with the caveat that the router must briefly go quiet if it detects radar.
Step 4: change one thing, then measure
Change the channel, wait a minute for clients to reconnect, and run the same test from the same spot. I use the speed test here from a fixed chair in the furthest room. If you change channel, width and transmit power all at once, you'll never know which change helped.
When channels aren't the problem
If the analyser shows a quiet channel and performance is still poor, the issue is probably distance or walls rather than interference. No channel choice will push a signal through two reinforced-concrete walls. That's where a second access point (see mesh versus wired access points) earns its keep.