The symptom is distinctive and baffling. You walk from the front room to the back, stand directly beneath the access point you just installed, and your phone stays connected to the one at the front, one bar of signal, and stuttering internet.

The first instinct is to blame the new access point. It's working fine; the only thing that didn't happen was the handover.

First, who actually decides

This one fact reshapes the entire approach to fixing it, and it nearly always surprises people: the decision to move belongs to the device, not the network.

Routers and access points cannot pull a phone across. All they can do is suggest, and only if the phone supports the suggestion mechanism. Otherwise the phone alone decides when to look, where to go, and whether to move at all.

Most devices decide with a simple, clumsy rule: stay on the current connection while it's still judged adequate, and only look elsewhere once the signal has become genuinely bad. Not when something better appears, but when the current one is close to failing.

This is the sticky client: a device that clings to its old access point until the connection is nearly gone. The behaviour is reasonable for a battery-powered device, scanning costs power, but the result is poor for the person carrying it around.

The basics, before anything else

Before touching advanced settings, three things must be right. Without all three, handover will never happen on its own.

  • Exactly the same network name on every access point. Including capitalisation.
  • The same password and encryption type. One access point still on WPA2 while others use WPA3 will cause handover to it to fail, see the guide to WPA2 versus WPA3.
  • One network behind them. The second access point must extend the same network rather than creating its own behind NAT. The arrangement is covered in the guide to two routers on one network.

Different names per floor, "WiFi-Floor1", "WiFi-Floor2", do remove the technical confusion, but move the work onto users. Almost nobody stops walking to switch networks.

What most often fixes it: lower the transmit power

This feels backwards, which is precisely why it's rarely tried.

If every access point transmits as strongly as it can, their coverage overlaps far beyond what's needed. A device in the back room still receives an "adequate" signal from the front access point, so it sees no reason to move.

Lowering the power narrows each area of coverage, so the old access point's signal genuinely drops as you change rooms. That drop is what prompts the device to look, and when it looks, it finds the nearer one.

Reduce it gradually, one level at a time, checking that nowhere is left without coverage. What you're after is enough overlap for smooth handover, not maximum overlap. Physical placement matters just as much, see the guide to router placement.

802.11k, v, and r: what each one does

The three are usually named together as though they were one package, though each handles a different stage of the same handover.

  • 802.11k, the neighbour list. The network tells the device which access points are nearby and on which channels. Without it, the device must scan every channel itself, which is slow and drains battery, so it postpones doing so for as long as possible.
  • 802.11v, the move suggestion. The network can suggest a device move to a particular access point. A suggestion only; the device may decline. But most modern devices comply, and this is where the improvement is most noticeable.
  • 802.11r, fast reconnection. Speeds up rejoining by caching part of the security handshake. It matters greatly for voice and video calls, where even a one-second gap is audible; it's barely noticeable when loading a web page.

All three work only if the device supports them too. Enabling them on the router doesn't improve a device that doesn't understand them. And 802.11r occasionally stops older devices connecting at all, if something breaks after enabling it, that's the first suspect.

Band steering: useful, but not for this

Band steering nudges devices from 2.4 GHz onto 5 GHz where both are available. It's often raised in discussions of roaming, though it solves a different problem: choosing a band, not choosing an access point.

The difference between the bands is covered in the guide to 2.4, 5, and 6 GHz. What matters here: 5 GHz has shorter reach, and on a multi-access-point network that's an advantage, its signal falls off more decisively with distance, so handover is triggered sooner.

If a particular device always struggles with band steering, some older ones do, giving it a separate 2.4 GHz network name is better than disabling band steering for everyone.

Mesh doesn't automatically solve it

Mesh systems are sold on the promise of seamless handover, and most do enable 802.11k, v, and r by default. But they can't change the underlying fact: the decision still belongs to the device.

A mesh installed with nodes too far apart, or at full transmit power in a small space, produces exactly the same symptom as ordinary access points. The comparison is in the guide to mesh Wi-Fi versus repeaters.

Repeaters are worth distinguishing too: most create a standalone network, so devices reconnect rather than roam, see the guide to boosting Wi-Fi signal properly.

How to test it

A useful test is simple, and better done walking than standing still:

  1. Open a signal analysis app that shows the MAC address of the access point in use, see the guide to Wi-Fi signal analysis apps.
  2. Note each access point's MAC address first, so you know where you are.
  3. Walk slowly from one end to the other, watching for the address to change.
  4. Note the signal strength at which handover happens. If it only moves below -80 dBm, the device is clinging too long, and lowering transmit power is the next step.

If the address never changes until you're standing directly beneath the second access point, re-check the basics above, one of them is most likely still wrong.

Handover can't be forced from the network side, because the decision isn't the network's. What you can do is make staying put an obviously bad option.

The order that usually works: match name, password, and encryption across every access point; lower transmit power until the overlap is merely sufficient; then enable 802.11k and v. Leave 802.11r for later, and only if voice calls are suffering.

Raising transmit power, the thing people try first, prolongs the problem instead.

Frequently asked questions

Why does my phone stay on a distant access point when a closer one exists?

Because the decision belongs to the device, not the router. Most devices only look for another access point once the signal has become genuinely poor, not when a better one appears. As long as the existing connection is judged adequate, they stay.

Must every access point use the same network name?

Yes, if you want handover to happen by itself. The name, password, and encryption type must match exactly. Different names push the choice onto users, and almost nobody makes it while walking between rooms.

What do 802.11k, v, and r actually do?

Each assists a different stage: k tells the device which access points are nearby, v lets the network suggest a move, and r speeds up reconnection. All three only take effect if the device supports them too.

Does raising transmit power help?

It usually makes things worse. More power widens every access point's reach, so devices hold on to distant ones for longer. On a multi-access-point network, lowering power often resolves more.