What the specs actually mean

OFDMA arrived with Wi-Fi 6 and is part of the standard, not an extra a manufacturer chose to add. Every Wi-Fi 6, 6E and 7 router supports it. When a spec sheet lists it as a feature, it is listing a consequence of the Wi-Fi generation already printed on the box.

What it changes is airtime efficiency, not peak speed. Before OFDMA a router handed the whole channel to one device at a time, however little that device had to send — a thermostat's status ping got the same slot as a 4K stream. OFDMA subdivides the channel so a dozen small transfers ride together. The benefit is lower latency in a house full of chatty devices, and it does not show up in a one-laptop speed test.

Filtering on OFDMA narrows the catalog to Wi-Fi 6 and newer, which the Wi-Fi standard filter does more directly and more honestly.

Channel width is the most direct lever on Wi-Fi speed: double the width, roughly double the data rate. Wi-Fi 5 topped out at 80 MHz, Wi-Fi 6 made 160 MHz common, and most of the advertised speed jump between generations is this and not anything cleverer.

Width trades robustness for throughput. A wider channel is more exposed to interference and reaches less far at the same power, so a device at the edge of range often does better on a narrower one — routers fall back automatically, so the advertised number is a ceiling reached in the same room, not a promise.

On 5 GHz there is a second catch worth understanding before filtering on this. The band is not wide enough for many non-overlapping 160 MHz channels, and the ones that exist mostly sit in DFS spectrum. A router that supports 160 MHz but has DFS disabled, or is in a location where radar keeps evicting it, spends its life at 80 MHz. On 6 GHz the room is there and the caveat disappears.

Wi-Fi 7 gear sometimes advertises 240 MHz on 5 GHz — that is a 320 MHz channel with an unusable 80 MHz section punched out, a workaround for exactly this crowding.

MU-MIMO and OFDMA get described in nearly the same words, and they solve the same problem from opposite ends. OFDMA divides a channel in time and frequency; MU-MIMO divides it in space, aiming distinct streams at devices that happen to be in different directions. Both mean "serve more than one device at a time".

It has been in every mainstream router since Wi-Fi 5 Wave 2. The number that actually varies is how many streams the radio has to give away, written as 2×2 or 4×4 in the radio table on each product page. A 4×4 radio can split into two 2×2 conversations; a 2×2 radio has nothing to split.

The catch is on the other end. Phones and laptops are overwhelmingly 2×2, and plenty of cheap devices are 1×1, so the router's extra streams only pay off when several capable devices are transmitting simultaneously from different parts of the room. Uplink MU-MIMO — devices talking back at the same time — is rarer still in client hardware than the router marketing implies.

A router with several antennas can time the same transmission slightly differently on each one so the waves reinforce in the direction of the device and cancel elsewhere. Nothing moves; the shaping is arithmetic.

Explicit beamforming has been standard since Wi-Fi 5 and is in essentially every router sold today, which is why finding it on a spec sheet tells you nothing. It also needs the device to cooperate by reporting back what it received, and phones and laptops have done that for years.

Keep the effect in proportion. Beamforming buys a modest improvement at the edge of range. Moving the router out of the media cabinet, off the floor, and away from the kitchen buys considerably more, and costs nothing. If coverage is the problem you are shopping to solve, a second node is the answer, not a better-shaped beam from one.

Every Wi-Fi generation before 7 made a device choose a band and stay there. A laptop on 5 GHz that walked out of range had to notice the connection dying, disconnect and rejoin on 2.4 GHz — a gap of a second or two you feel as a video call freezing. MLO lets one connection span both bands at once, sending traffic down whichever is healthier moment to moment, or down both together.

This is the Wi-Fi 7 feature worth paying for. The headline 320 MHz channel widths are a lab number most homes will not see; MLO improves the thing people actually complain about, which is Wi-Fi that is fast on the speed test and unreliable in the back bedroom.

It requires Wi-Fi 7 on both ends. A Wi-Fi 7 router talking to a Wi-Fi 6 phone behaves exactly like a Wi-Fi 6 router, and client hardware is where the adoption gap is. Buying a Wi-Fi 7 router for MLO before you own any Wi-Fi 7 devices buys a feature that switches on later.

Mesh systems use it for a second purpose: an MLO link between nodes is a better backhaul than a single-band one, which is why some Wi-Fi 7 mesh kits dropped the dedicated backhaul radio their Wi-Fi 6E predecessors had.

This is where Wi-Fi 7's headline numbers come from. A 320 MHz channel carries twice what 160 MHz does, and the five-figure Mbps ratings on the front of the box are mostly this multiplied by a stream count.

It exists only on 6 GHz. The 5 GHz band has no room for a channel that wide, which is why Wi-Fi 7 routers quote 240 MHz there instead — a 320 MHz channel with a chunk punched out.

6 GHz is the constraint on all of it. The band is clean and empty, and it is also the shortest-range Wi-Fi band there is: high frequencies lose more energy passing through walls, and 6 GHz gear is power-limited indoors on top of that. A 320 MHz link is a same-room link. Two rooms away the router falls back to a narrower channel on 5 GHz and the headline number stops being relevant.

It is a genuine differentiator — few routers have it, and it is not arriving in older hardware — but it differentiates on a case most homes do not have: a 6 GHz-capable device, close to the router, moving enough data to notice. A desktop beside the router is better served by the Ethernet port.

Most of the 5 GHz band is not exclusively Wi-Fi's. A large middle section is licensed to radar — airport and weather systems, military installations — and Wi-Fi is a guest there. DFS is the rule of tenancy: before using one of those channels the router must listen for a full minute, and if radar appears while it is transmitting it must vacate within ten seconds and stay off that channel for half an hour.

The upside is real and often large. The non-DFS 5 GHz channels are a handful, and in an apartment building every router you can see is fighting over them. DFS roughly triples the usable spectrum, and it is frequently empty.

So is the downside. A false radar detection — and they happen, especially near airports, coastlines and some doorbell cameras — drops every device on that radio and takes up to a minute to come back. If your household notices a one-minute Wi-Fi outage, that is worse than a crowded channel. It is also why 160 MHz channels and DFS are entangled: on 5 GHz there is barely room for a 160 MHz channel that does not overlap DFS spectrum.

Worth knowing which way you want it, and worth checking that the router lets you choose. Not every one does.

A router with 2.4 GHz and 5 GHz radios usually presents them as one network name, and leaves the choice of band to the device. Devices choose badly. A phone that connected on 2.4 GHz in the garden will happily stay on 2.4 GHz after you walk back to the router, because the connection still works and nothing makes it reconsider. Band steering makes the router refuse or delay the 2.4 GHz association so the device tries 5 GHz instead.

When it works you never notice it, which is why it is hard to shop for.

The reason it is not simply good is the opposite case. A lot of cheap smart-home hardware — plugs, sensors, older cameras — is 2.4 GHz only, and some of it cannot complete setup on a merged network at all. The fix is to split the bands into separate names temporarily, and that is a capability, not a feature you can steer your way out of. If your house is full of that kind of device, the question to ask a router is whether it will let you split, not whether it steers.

Mesh has always been proprietary. A Netgear satellite does not join an eero network, and a two-node kit bought in 2022 may not accept the same maker's 2026 node. EasyMesh is the Wi-Fi Alliance's answer: a certified protocol for node discovery, steering and backhaul that any certified device can speak.

The value is optionality. You can extend an existing network with whatever is on sale, and you are not stranded when a vendor discontinues a line.

In practice the ceiling is lower than the promise. Cross-brand meshes work, but the vendor's own app, its per-device controls and its more aggressive steering usually only apply within its own family, so a mixed network converges on the lowest common denominator. And most people buy a matched kit, put it in, and never add a node from anyone.

Worth having if you already own a compatible router and want to extend it cheaply. Not worth choosing a worse router to get.

In a mesh system, a satellite node has to relay everything it receives back to the main unit. If it does that over the same radio your laptop is using, every byte crosses the air twice and the node roughly halves the throughput of anything connected to it. This is the single biggest reason a mesh system disappoints.

A dedicated backhaul radio is a third (or fourth) radio the mesh keeps for itself. Devices never see it, so relayed traffic stops competing with them.

Read the spec carefully, because "tri-band" alone does not mean this. Plenty of tri-band mesh kits use the third radio as ordinary client capacity and share it with backhaul when needed; only some hold it back. That distinction is what this facet records, and it is the one worth paying for in a wireless mesh.

None of it applies if you can run a cable. Wired backhaul removes the problem entirely rather than mitigating it, costs the price of Ethernet cable, and turns a mid-range mesh into a better system than a premium one relaying wirelessly. Check for wired-backhaul support first; treat dedicated wireless backhaul as the answer for the rooms you cannot reach with a cable.