Find and fix unreliable home internet

Use a four-way diagnosis

If one device fails, check that device. If Wi-Fi fails but Ethernet works, inspect placement, interference and access-point coverage. If both fail while the modem or ONT has power, check official provider status and ask support to test the line. If equipment loses power, plan electrical backup first; a faster service tier cannot solve that failure.

A connection can also stay online but feel unreliable because upload is saturated, latency varies or packets are lost. Test at the hour of trouble while noting simultaneous backups, cameras, calls and downloads. Repeat enough times to avoid making a decision from one lucky speed test.

• One device or all devices?
• Ethernet, Wi-Fi or both?
• Power available to ONT/modem and router?
• Provider incident or repeatable local pattern?

Choose the smallest effective intervention

For weak Wi-Fi, move the router or add a wired access point. For a line fault, gather test results and work with the provider. For upload saturation, schedule backups, use router quality-of-service settings if suitable, or compare a plan with stronger upstream delivery. For unavoidable outages, test a separate connection and its power supply.

A backup service is a business or household continuity purchase, not a substitute for repairing a broken primary line. Price it against the disruption it actually prevents and test the switchover periodically.

Worked example (illustrative): Fault log: every weekday 2 p.m., Ethernet is stable but the upstairs laptop drops; changing providers is unlikely to fix that room. If Ethernet and Wi-Fi both fail while the ONT reports no optical signal, take that record to provider support.

Separate four kinds of unreliability

An outage means the connection stops; congestion means it slows under load; a Wi-Fi dead zone affects particular places; and power loss turns off equipment even when a provider network is healthy. These require different evidence. A useful seven-day log records start/end time, device, Ethernet versus Wi-Fi, gateway lights and whether the provider reported an incident. Do not collapse every symptom into a single ‘slow internet’ note.

The first branch is one device versus all. The second is wired versus wireless. The third is powered equipment versus unpowered. Only after those should you compare service tiers or providers. Repeated wired failures with powered equipment belong in a support ticket. Repeated Wi-Fi-only failures belong in a placement or access-point review. A faster plan can help actual saturation, but it will not restore an unpowered ONT.

Match the remedy to the cost of failure

For a faulty provider line, pursue repair before adding a second bill. For local Wi-Fi, test router position or a wired access point before changing the outside connection. For upload saturation, schedule backups and assess an offer with useful upstream headroom. For rare external outages, a charged phone may be enough. For frequent work-disrupting events, a tested second path and backup power may be worth the recurring and maintenance cost.

Set a measurable acceptance criterion. ‘No work call drops in the office over two weeks’ is better than ‘get faster internet.’ After a change, repeat the same test and compare fault frequency, not just a single best speed result. If the remedy does not improve the original symptom, undo the purchase where terms allow and keep diagnosing; otherwise a sequence of speed upgrades can become a costly substitute for finding the failed link.

Worked case (hypothetical): Hypothetical fault log: four interruptions in seven days. Three affect only an upstairs Wi-Fi laptop while Ethernet remains online; one affects every device and appears in the provider's outage status. Treat these as two problems. A new provider might address one event, while a local access point addresses the other three.

Continue with Follow the outage decision tree when the line is down now; Size backup when lost connectivity has a measurable cost.