I spent, in the spring of 2024, eleven hours of video meetings lost to a Wi-Fi signal that could not reach my desk. The meetings were for work. The desk is on the second floor of a Brooklyn brownstone I share with my partner and a small, opinionated dog. The router lived, where the ISP installer had put it in 2021, behind the television on the first floor. I lost the signal every time the dog went downstairs and stood between the router and the stairs.

The ISP, when I called, offered to send a technician on a Thursday between 8 AM and 1 PM. The technician, when he arrived, told me the router was fine, the signal was fine, and that I should "think about ethernet." I thought about ethernet. I have, in my house, no ethernet wiring. I have, in my house, plaster walls from 1908. I have, in my house, a partner who does not want holes drilled in the plaster. The ISP was not, in any practical sense, going to fix it.

Below is what I bought, what I spent, and what changed. I have changed my mind about several things I used to believe about home Wi-Fi. I have opinions on what to skip. There is a brief history of how the home router went from a beige Linksys box in 2002 to a $399 mesh system in 2026. And there is, at the end, an honest summary — because anyone who tells you the ISP router is fine has not, in fact, tried to take a video call from a brownstone bedroom.

The list

1. Eero Pro 6E three-pack, $399 (2024). The single most useful piece of technology I have bought for the house. The Eero Pro 6E is, in 2026, the mesh system that Amazon owns (they bought Eero in 2019), with three identical nodes that talk to each other and to the modem. One node sits next to the modem in the living room. One node sits on the second-floor landing. One node sits on my desk. The result is, in every room of the house, the same network name and the same signal strength. The set-up took me, by my standard, twenty-five minutes. Cost per year over two years: $200.

2. Netgear Orbi 860 Series three-pack, $699 (alternative). The alternative I almost bought. The Orbi is, by Netgear's accounting, the gold standard of mesh — faster raw throughput, a dedicated backhaul radio, an aesthetic that, by 2026 standards, looks like a piece of audio equipment. The Orbi is also, by my standard, $300 more than the Eero for, in most home use cases, no measurable difference. Skip the Orbi unless you have a 2 Gbps fiber connection. Buy the Eero.

3. Google Nest Wifi Pro three-pack, $399 (alternative). The Google equivalent. The Nest Wifi Pro is, by Google's accounting, the mesh system that integrates best with Android phones and Google Home devices. I do not have a Google Home. I do not have an Android phone. The Nest was, for me, the wrong ecosystem. For a household deep in Google, the Nest is, by every review I have read, equivalent to the Eero at the same price.

4. TP-Link Deco XE75 three-pack, $300 (alternative). The cheapest three-pack from a major brand. The Deco is, by every review I have read, fine. The Deco is, in 2026, missing some of the polish of the Eero app. The Deco is also, by my standard, $100 less than the Eero for performance that, on a Speedtest, is within 10%. If $100 matters, buy the Deco.

5. ISP-provided router (the one I had), $0-$15/month rental. The router that came in the cardboard sleeve. The router the technician told me was fine. The router that, on most afternoons, gave me 12 Mbps to my desk. ISP routers are, in 2026, the single most common reason home Wi-Fi is bad. They are not, in most cases, bad hardware. They are, in most cases, 2021 hardware running 2026 demand. They are, in most cases, configured to optimize for the ISP's network, not for your house.

6. Single Wi-Fi 7 router (no mesh), $250-$500. The right answer for a small apartment. A single, well-placed Wi-Fi 7 router — central, off the floor, not behind the television — will outperform a three-pack of mesh nodes in a studio. The mesh industry has, by my standard, done an excellent job of convincing you otherwise. The mesh industry sells a lot of three-packs at $400 to people who needed one router at $150. If you live in under 900 square feet, buy a single router. If you live in over 900 square feet with plaster walls, buy the mesh.

7. Modem/router combo from the ISP, $0-$15/month. The all-in-one box. The combo is, in most cases, a worse deal than separate modem and router — the modem half is fine, the router half is the same underpowered ISP hardware. If you have the combo, you have, by my standard, two problems for the price of one.

8. Surge protector with battery backup, $80 (2024). Not a router accessory. A brownstone accessory. The power in my neighborhood goes out, on average, four times a year, usually for under an hour. The router, when the power blips, takes three minutes to re-acquire the IP address from the ISP. The surge protector with battery backup keeps the router up through the blip. The video calls do not drop. The dog does not notice.

9. Ethernet cable, 50-foot Cat 6, $12 (2024). For the desk. Even with mesh, the desk runs faster over a wire. The wire runs, in my case, along the baseboard, around the door frame, and into the wall plate behind the desk. The cable cost $12. The installation cost me twenty minutes and one command strip. The result is, on a Speedtest, 940 Mbps to the desk. The wireless result was 180 Mbps.

10. Wi-Fi analyzer app (free, 2024). The single best fifteen minutes you will spend on your network. The app shows you which channels are crowded in your neighborhood and which channel your router is on. I moved my Eero from channel 6 to channel 149. The 2.4 GHz performance improved, by the next Speedtest, by 30%. The app is free. The 30% is, by my standard, the highest-leverage fifteen minutes I have spent on home networking.

Total spend: $491. Cost per year over two years: $246. Hours of meetings recovered, by my estimate: eleven in 2024, probably forty by now. The math is, by any honest accounting, settled.

What I have changed my mind about

A few beliefs I held in 2023, that the last two years have killed.

I used to think the ISP router was fine. I was wrong. The ISP router is, in 2026, the single most common reason home Wi-Fi is bad. The ISP router is, in most cases, 2021 hardware running 2026 demand. The ISP router is, in most cases, configured to optimize for the ISP's network, not for your house. Replace it.

I used to think routers didn't matter. I was wrong. Routers matter enormously. The router is the foundation of every connected device in your house. The router, when it is bad, makes every device bad. The router, when it is good, makes every device invisible. Replace it.

I used to think Wi-Fi was just Wi-Fi. I was wrong. Wi-Fi 7 (certified 2024) is, by every measurement, three to four times faster than Wi-Fi 5 (2013) at the same coverage. Wi-Fi 6E (2021) added the 6 GHz band, which is, by my standard, the only band that handles a video call while someone else is streaming 4K in the next room. Wi-Fi standards matter. Buy recent.

I used to think I needed a separate access point. I was wrong. A mesh system is, in technical terms, three access points that cooperate. A mesh system is, in practical terms, one network name everywhere. The "separate access point" advice was, in 2018, the right advice. The advice is, in 2026, obsolete.

I used to think I had to log into the router. I was wrong. Modern mesh systems are configured by an app. The Eero app took me, on the first try, through the entire set-up. I have not logged into a router admin panel since 2018. I have not, by my standard, missed it.

I used to think Wi-Fi was bad in old buildings. I was wrong about this in a specific way. Wi-Fi is bad in old buildings when the router is in the wrong place. Wi-Fi is, in old buildings, just as fast as in new buildings, when the router is well-placed and the mesh is right-sized. The brownstone is, by my standard, the easiest building to fix. Plaster walls block signal the same way drywall does, when you have three nodes instead of one.

What to skip when buying a router

If I were advising a friend who was about to replace their home router in 2026, I would tell them to skip these categories entirely at the start.

Gaming routers. The "gaming" label is, in most cases, a marketing line on top of standard Wi-Fi 6 or Wi-Fi 7 hardware. The "gaming" router is, by my standard, the same router with a different logo and a $100 markup. Skip the gaming label.

Wi-Fi 6 (without 6E). Wi-Fi 6E added the 6 GHz band in 2021. The 6 GHz band is, by every measurement, the only band that handles a modern household. Wi-Fi 6 without 6E is, in 2026, the previous generation. Buy Wi-Fi 7 or, at minimum, Wi-Fi 6E.

Routers with external antennas. The antenna-prominent router is, by 2026 standards, a 2018 aesthetic. The antenna-prominent router is, in most cases, the same Wi-Fi 7 chip as the antenna-less mesh node. Buy the antenna-less version. It looks better on the bookshelf.

"Advanced security" subscriptions. The router's "advanced security" subscription ($30-$100/year) is, in most cases, a software layer on top of features your devices already have. The features are, in most cases, the same features Apple, Google, and Microsoft already ship. Skip the subscription.

Routers with built-in modems. The all-in-one box saves you one outlet and combines two failure modes into one. When the modem half fails, you replace the whole thing. When the router half is obsolete, you replace the whole thing. Keep them separate.

"Smart home hub" routers. The router with a Zigbee radio and a Thread radio and a Matter controller is, in most cases, the router that does three things badly. The mesh system does Wi-Fi well. The smart home hub does smart home well. Skip the combined version.

A brief history

The home router is younger than most people think. The first consumer models appeared in the late 1990s, after Linksys (founded 1988 in Irvine, California) and a handful of other companies realized that the modem-then-Ethernet-then-everything-else chain could be collapsed into one box. Netgear shipped its first router in 1996. Apple did not enter the category until 2001, with the original AirPort.

By 2002, the Linksys WRT54G had been released. The WRT54G was, by every measure, the most successful home router of all time. It ran Linux. It was, for several years, the de facto standard. It sold, by industry estimates, more than 70 million units. The WRT54G is, in 2026, the box your parents still have in a closet somewhere, running WPA2 on a 2.4 GHz radio, doing 54 Mbps on a good day.

The Wi-Fi standard itself has, since 1999, gone through eight iterations. 802.11b in 1999 at 11 Mbps. 802.11a in 1999 at 54 Mbps. 802.11g in 2003 at 54 Mbps. 802.11n (Wi-Fi 4) in 2009 at 600 Mbps. 802.11ac (Wi-Fi 5) in 2013 at 1.3 Gbps. 802.11ax (Wi-Fi 6) in 2019 at 9.6 Gbps. Wi-Fi 6E in 2021, adding the 6 GHz band. Wi-Fi 7 in 2024, with 320 MHz channels and Multi-Link Operation. Each one roughly doubled real-world throughput. Each one is, by my standard, the only one worth buying new.

The mesh concept was, in compressed form, an answer to the problem that the single router could not cover a whole house. Eero, founded in 2015 in San Francisco, launched the first consumer mesh system in 2016. By 2019, Amazon had bought Eero. By 2021, every major router company had a mesh line. The mesh industry is, by 2026, the dominant form of home Wi-Fi. The single router is, for most households, the previous generation.

The 2020s home network is, by every measure, more demanding than any previous generation. The average US household now has, by industry estimates, more than 25 connected devices. The video calls have replaced meetings. The streaming has replaced broadcast. The smart home has replaced the light switch. The router, in 2026, is not the box you ignore. The router is, in 2026, the foundation.

The honest summary

A router is, in the end, three things: a modem connection, a Wi-Fi signal, and a place to plug in ethernet. Everything else is, by my standard, "an excuse." The $399 mesh system I bought is, by every measurement that matters, the single best piece of technology in my house.

What works for me: a three-pack mesh system with Wi-Fi 6E or Wi-Fi 7, one node per floor, ethernet to the desk, and a battery backup on the modem-side node. That is it. The set-up took me twenty-five minutes. The meetings have not dropped since.

What does not work for me, and may not work for you: the ISP router that came in the cardboard sleeve. The ISP router is, in 2026, the single most common reason home Wi-Fi is bad. The ISP router is, in most cases, 2021 hardware. The ISP is, in most cases, not going to fix it. The fix is, in most cases, $300-$400 for a mesh system you install in an afternoon.

Two years in, my network is, by every measurement, more reliable than it was when I started. I have spent, in those two years, $491 on hardware. I have recovered, by my estimate, at least fifty hours of meetings. The math is, by any honest accounting, the easiest I have ever done. The math is also, in the end, the only thing that matters.

The router, on most afternoons, is invisible. The router, on most afternoons, is exactly what it should be.