WiFi for vacation rentals: speed, router, and LTE failover

The first WiFi-related 1-star review I got was for a router that worked perfectly for 47 weeks of the year. The week it didn't, a remote-working guest watched a video call die three times in twelve minutes, gave up trying to file an emergency support ticket from her phone hotspot, and left me a review titled "WiFi unusable" with no body text. The router rebooted itself about an hour after she checked out and ran fine for the next nine months. I would never have known it failed if she hadn't told me — in public, at the worst possible moment, in the highest-traffic place on the listing.

This is the post about the WiFi setup that prevents that review. The Mbps number guests actually need (it isn't the one on the speedtest sticker), the router that lasts past month four in a real apartment, and the $10-a-month LTE failover device that turns the worst kind of guest complaint into a non-event before they notice.

The Mbps number guests actually need

Hosts overspend on speed because the ISP page makes the 100 Mbps plan look ten times better than the 25 Mbps plan for €15 more a month. Then a guest writes "WiFi was slow" in a review of a property with a 500 Mbps fibre line that did 480 Mbps the morning the cleaner left.

What "slow" almost always means in those reviews:

• The signal dropped to one bar in the bedroom. The router was in the entry hall behind a metal radiator.
• The router crashed for 40 minutes during the day and rebooted itself before they messaged you.
• DNS was misconfigured and Netflix took 14 seconds to load a thumbnail.
• The guest connected to the 2.4 GHz band because the 5 GHz one was named the same and their phone picked the louder, slower signal.

Almost none of those problems are fixed by buying more Mbps. They're fixed by router placement, a working second band, and DNS that isn't the ISP's default. The actual bandwidth budget per common use case:

Two adults streaming 4K Netflix on separate TVs while a third person does a Zoom call needs about 80 Mbps down, 8 Mbps up. That is the worst plausible case for a 2-bedroom listing, and a 100 Mbps symmetric fibre line covers it twice over.

The number that actually matters in remote-work-heavy markets — Lisbon, Mexico City, Bali, Tbilisi — is the upload speed. A guest doing four Zoom calls a day on a 100/10 Mbps cable plan with three other guests using the network will hit the upload ceiling once a week and call it "the WiFi". Get 50/50 or 100/100 fibre if you have the choice; pay €10 more, never see that review.

The router that lasts past month four

I have replaced enough consumer routers in enough apartments to have a rule: the failure rate of a £40 plastic router in a real STR is high enough that by the fourth listing, the breakeven against a mesh system is already crossed.

A breakdown of what I've actually seen die:

• Cheap dual-band routers (TP-Link Archer C6, AC1750-class): one in three fails within 18 months of 24/7 operation. Capacitors, mostly, in humid kitchens.
• Single-unit mid-range (ASUS RT-AX55, Netgear Nighthawk AX1800): one in eight fails in the same window. Usually firmware-related — the device works but the 5 GHz band silently drops and only a reboot fixes it.
• Mesh systems (TP-Link Deco, Google Nest, eero): one in twenty fails in 24 months. The redundancy is built into the system. If a node dies, the others still serve WiFi while you ship a replacement.

The cost math at four listings, over a 24-month window:

The mesh number looks worst on hardware. It looks best on guest experience — the dead zone in the back bedroom is gone, the dropped-signal review never happens, and the one full refund you avoid is worth six months of the difference. By listing five or six, the mesh setup is the cheapest line in the table.

For one listing in a small apartment, a single solid mid-range router is fine. The break-even point at which mesh starts paying for itself is around 80 m² of floor area or three rooms with closed doors between the router and the bed. Below that, a single AX1800-class router in a central spot wins.

The LTE failover almost nobody installs

Here is the number that changes the conversation: an LTE failover router (or USB modem plugged into your main router's USB port) with a £8–12/month data-only SIM costs around £140 over a year. One refunded outage night on a £90/night listing is £90 plus the lost review. The math on LTE failover stops mattering after the first outage you avoid.

The setup most hosts who run one is one of:

• A consumer router with a USB port that accepts a 4G USB modem (TP-Link Archer C7, Netgear Nighthawk, most AsusWRT firmware boxes). Plug in modem, configure failover in the router admin, done. ~£40 modem, £8/month SIM.
• A dedicated LTE failover device (TP-Link MR600, Teltonika RUT240, Cradlepoint for the corporate end). Sits between modem and switch; when the WAN drops, traffic seamlessly migrates to LTE. ~£100 device, £10/month SIM.
• A mesh system that natively supports an LTE backhaul (TP-Link Deco X20-4G, Nest WiFi Pro with a tethered phone). Higher hardware cost but the whole network — including the dead spots — keeps working.

The eSIM market makes this cheap in 2026. Airalo and Holafly sell €5–10/month data-only eSIMs in most countries, no contract, no SIM swap; many failover devices now support eSIM directly. Three years ago this option did not exist for hobby hosts; now it is the single highest-ROI £150 a host can spend on a property.

There is one caveat: failover speeds are 4G, not fibre. A guest mid-Zoom-call when the fibre drops will see the call quality dip for one or two seconds while the failover kicks in, and if the property gets only 2 bars of 4G the experience for the rest of the day is "WiFi is slow" instead of "WiFi is dead". The first one is a 4-star review. The second one is a refund and a 1-star.

The monitoring that catches the silent failure

The router can be working and the internet can be dead. The cable to the building was cut, the ISP had a regional outage, the modem locked up but the router didn't notice. None of those show as a router fault in your admin panel. The only way to know is to test from outside the network.

The setup I run on every property, for free:

1. A small heartbeat script running on a £35 Raspberry Pi or, more commonly now, on a UK-based VPS, pings the property's public IP every 5 minutes.
2. After three consecutive failures (15 minutes of no response), it pushes a Telegram or Slack message to me.
3. After 30 minutes of failures, it also messages the cleaner with the boilerplate "please power-cycle the router box marked with the green sticker".

Total setup time: 20 minutes per property the first time, 5 minutes per property after the first one. You will get the first alert about an hour before the guest notices, and 80% of the time you can reboot the router remotely (or have the cleaner walk in) before the guest sees anything.

If you do not want to run a script, the cheapest hosted equivalents:

• UptimeRobot free tier: 50 monitors, 5-minute interval, email + push alerts. £0/month.
• BetterStack (formerly Better Uptime) starter: 30-second intervals, on-call rotations. £20/month for serious operators.
• A consumer router with built-in ping-watchdog (most AsusWRT and OpenWRT builds): no external service needed.

This is the highest-leverage hour of WiFi work a host will ever do. You catch the failure that would otherwise be a 1-star, and you find out at 11:30 in the morning when you have time to deal with it, not at 23:45 when the guest is already typing.

Where guests actually look for the password

Three places. Guests look at one of them and never the other two:

1. A physical card on the kitchen counter or the table by the door. Plastic, laminated, A6. SSID and password in 24 pt bold so a phone with cracked glass can read it from a metre away. Cost: €2 to print, €1 to laminate, takes 4 minutes.
2. The digital guidebook. The same SSID and password as the first line of the WiFi section. If you use a digital guidebook tool, this is already a field. If you don't, send the WiFi info in a single message 30 minutes after check-in instead of in the welcome paragraph.
3. A small sticker on the smart-lock keypad or the lockbox. Three lines: SSID, password, "if WiFi is down please message me first". This is the one that catches the guests who arrive at midnight and can't find the kitchen card because the lights are off.

I have stopped writing the WiFi info in the welcome message. Guests scroll past it and ask for it later in 1 of 3 stays. The card on the counter and the lock sticker are read once, by every guest, on minute one.

What to do when the WiFi actually fails mid-stay

The playbook is similar to the self-check-in failure procedure — minute one is a message, hour one is a fix, after-action is the review.

Minute one: acknowledge inside 5 minutes. "I see the alert, the LTE backup should be active in 30 seconds — please wait 60 seconds and reconnect; if it's not working, message me back." If your monitoring alerted you before the guest did, this message can go out before they message you, which turns the entire incident from a complaint into a customer-service win.

Hour one: if the LTE failover is up, the network is back. If it isn't, two options:

• Remote reboot (smart plug on the router/modem combo, £15 on Amazon, controllable from the host phone): 90% of router-side problems resolve.
• Cleaner or co-host walkthrough: 30-minute response time in most cities, 5 minutes onsite, problem solved.

The compensation: if the outage was less than 90 minutes, the right offer is "I'm extending your check-out by an hour and sending you a coffee voucher for the place downstairs." If it was 6+ hours, the right offer is one night's refund and an explicit "we will not let this happen again, here is what we changed". Do not offer a partial refund before the guest asks — it primes them to ask for more next time. After the guest mentions it once, refund quickly and visibly, before they post a review.

One opinionated take

Most hosts overspend on speed and underspend on uptime. A guest will forget about 100 Mbps on day three of their stay. They will never forgive zero Mbps on day one. The host who reads this post and only takes one thing should set up a monitoring alert before they shop for a router — because the router you already have is probably fine, and the silent failure you don't know about is the one writing the review.