Internet keeps dropping at your company: how to find the real cause
Repeated outages almost always have an identifiable cause. Here is the method to find it, and why a second router usually fixes nothing.
An internet outage has three possible origins: the operator link, the network hardware, or internal saturation. Before switching providers, you need to know which one. Log the exact date and time of every incident for two weeks: the rhythm of the outages points to the cause almost every time, and saves you paying for a second line that will fail alongside the first.
A company that suffers three outages a month switches operator, then suffers three outages a month with the next one. The pattern is common enough to take seriously: in most cases, the operator was never the problem.
The word outage covers three completely different situations, fixed in three completely different ways. Telling them apart is the first thing to do.
The three possible causes, and how to tell them apart
| What you observe | Most likely cause | Who can act |
|---|---|---|
| Everything drops at once, phones included, then comes back on its own | The operator link | The operator |
| Wi-Fi drops but a wired desk still works | The internal network | You or your provider |
| Slow at fixed times, never fully down | Saturation | You or your provider |
| One service stops responding, everything else works | Neither: the service itself | The vendor |
The two-week log
No serious diagnosis is made from memory. For two weeks, record every incident.
- The exact date and time, to the minute.
- How long before things returned to normal.
- What was affected: everything, Wi-Fi only, one application.
- What was happening at that moment: backup, video call, staff arriving.
The rhythm speaks. Outages around lunchtime point to saturation, not a fault. A weekly outage at a fixed hour almost always points to a scheduled task. Short, random outages spread across the day point to hardware or to the link.
Why a second router is not enough
It is the most common reflex, and usually a useless one. A second connection only protects you if it is genuinely independent of the first.
- Same technology and same infrastructure: both links run through the same connection point and fail together.
- Same physical path: one digger cuts both cables at once.
- Manual failover: if someone has to unplug a cable to activate the backup, there is no backup on a Saturday.
Real redundancy requires two distinct infrastructures, typically fibre plus a 4G/5G radio link, and a router that switches over on its own within seconds, with nobody involved.
What to check before calling
- Note the router indicator lights during the incident, not afterwards: that information is the only one that disappears.
- Test from a wired desk. If the cable works, this is not an internet outage, it is a Wi-Fi problem.
- Check the age of the network hardware. A switch or access point over seven years old fails intermittently long before it fails for good.
- Look at what is consuming bandwidth. A cloud backup running during the day saturates a link on its own.
- Record the operator ticket number at every call: with no history, every call starts from zero.
What an outage actually costs
The calculation needs no software. Take the hourly payroll of the people who are blocked, add the work that cannot be billed during the stoppage, then multiply by the total outage time over a year.
As an illustration: a twenty-person company where half the staff is blocked for two hours, six times a year, loses the equivalent of more than a working week. The figure almost always exceeds the cost of a backup link.
Frequently asked questions
- How do I know whether the outage comes from my operator or my network?
- Test from a desk connected by cable during the incident. If that desk has no internet access and IP telephony is also down, the problem is on the operator link. If the wired desk works, the fault is internal, usually Wi-Fi or a switch.
- Is a second router enough as a backup connection?
- Rarely. If both connections use the same technology and the same connection point, they fail together. Effective redundancy combines two distinct infrastructures, for example fibre and a 4G/5G link, with a router that switches over automatically.
- How long should I observe before drawing conclusions?
- Two weeks of logging is enough in most cases. It is long enough for a pattern to appear (fixed hour, fixed day, correlation with a scheduled task) and short enough that the company does not tire of recording incidents.
Next step
Sources
Published 8 September 2026 · Équipe Skill Group, Télécom & réseaux
Going further
FTTH or FTTO: which fibre should your business choose?
Both are called fibre, yet they offer very different services. The real deciding factor is not speed, it is what happens when the line goes down.
When the internet goes down, who is actually responsible?
Three providers, three diagnoses, and a company at a standstill while they argue. This is not bad faith: it is a predictable consequence of splitting the work.
Ambulances and approved medical taxis: stop losing jobs over the phone
In medical transport, a missed call is a lost job, and often a client who will call a competitor next time. Here is what can actually be fixed.