What a Single Internet Connection Actually Risks for a Growing Company

A 110-person company has moved its phones, its accounting system, its file storage, and its customer database into the cloud over the last five years. Every one of those moves was approved on its own merits. None of them came with a question about the single cable that now carries all of it.

That cable is one circuit from one provider, installed when the office had 20 people and the internet meant email and web browsing. When it goes down on a Tuesday morning, the company does not lose a tool. It loses the building's ability to work, and the leadership team finds out what the connection was actually worth.

Why the Single Circuit Became a Single Point of Failure

Ten years ago, an internet outage was an inconvenience. Staff worked from local files, the phones were on a separate line, and the applications that mattered ran on a server in a closet. Losing the connection meant no email and no web.

That is not the dependency structure of a growing company today. Voice calls run over the internet. The line-of-business application is a web service. Files live in a cloud drive. Payroll, billing, and the customer relationship system all sit behind a login that requires a working connection. A single circuit now sits underneath nearly every function in the business, and nobody designed it to carry that weight. It simply stayed in place while the company grew onto it.

The FCC's 2026 Section 706 broadband report shows how the market looks from the other side. Roughly 96.9 percent of Americans can get fixed broadband at 100/20 Mbps when fixed wireless is counted, and 77 percent have access to three or more fixed providers at that speed. When terrestrial providers alone are counted, only 43.4 percent have three or more. Those figures measure availability at the home and address level, not the reliability of any one provider. But they carry a practical point: for most office locations, a second path to the internet is available. The single connection is usually a choice nobody revisited, not a constraint.

What Actually Takes a Circuit Down

Leaders tend to picture an outage as a provider failing in some dramatic way. The common causes are more mundane, and several of them are outside the provider's control.

• Physical damage to buried cable. Construction crews cut lines constantly. The Common Ground Alliance's 2024 DIRT Report analyzed 196,977 unique reports of damage to buried utilities in a single year, and telecommunications lines are among the most frequently struck. A backhoe on the next block can take out a circuit for a day.
• Equipment failure at the provider. The provider's local node, the modem at your site, or the router behind it can fail independently of the cable itself.
• Provider maintenance and billing errors. A scheduled maintenance window that runs long, a provisioning change that goes wrong, or an account issue that suspends service all produce the same result as a cut line.
• Last-mile problems. A weather event, a power outage in the provider's local facility, or a congestion problem in the neighborhood can degrade the connection without fully removing it.
• Your own hardware. A single firewall or router with no spare is a second single point of failure sitting right behind the first. A company can buy two circuits and still go down because everything funnels through one device.

The pattern across these causes matters more than any one of them. Some are rare, some are routine, and almost none can be predicted. A company with one circuit is relying on every one of them not happening at a bad time.

How to Price the Risk Before Anyone Quotes a Fix

The most common mistake is to treat redundancy as a technology purchase and evaluate it on the cost of the equipment. The better framing is to price the outage first, then decide what protection is worth.

Start with three numbers from your own business.

1. Payroll cost of idle staff. Take the loaded annual cost of the people who cannot work without the connection, divide by roughly 2,000 working hours, and multiply by the share of the workday they are blocked. A 110-person company where 80 people are blocked and the average loaded cost is $85,000 a year is paying about $42.50 per person-hour, or roughly $3,400 an hour for the group. A four-hour outage costs about $13,600 in labor alone. These inputs are illustrative. Substitute your own.
2. Revenue that cannot be captured or invoiced. If the business takes orders online, books appointments through a portal, or bills by the hour from a cloud system, calculate the revenue that is delayed or lost per hour. Delayed revenue often recovers. Lost revenue does not.
3. Customer and contract exposure. If client commitments include response times or availability, an outage can trigger penalties or erode trust in ways that do not appear in the first two numbers.

Multiply the result by a realistic number of outage hours per year. If the connection has gone down twice for three hours over the last 18 months, that is a defensible starting estimate. Compare the annual figure to the annual cost of a second connection and the hardware to manage it. For most companies in the 25 to 200 employee range, the outage math is lopsided enough that the decision gets made quickly once it is written down.

The point of doing this first is that it changes the conversation with the CEO, COO, or CFO. The question stops being whether to buy a second circuit and becomes how much downtime the business is willing to absorb.

What Redundancy Actually Involves

A second circuit is the visible part of the fix. The design decisions around it determine whether it works when needed.

Path diversity

Two circuits from the same provider often share the same buried conduit, the same local node, or the same upstream facility. When that shared piece fails, both circuits fail together. Real redundancy means a different provider, ideally on a different physical medium. A cable or fiber primary paired with a fixed wireless or cellular backup is more resilient than two wired circuits that enter the building through the same trench.

Backup type and speed

Backup options differ in how much of the business they can carry.

• A cellular failover connection is the simplest to deploy and is independent of local wiring. It is often slower and may carry data caps or throttling, so it works best as a way to keep critical functions running, not to replicate normal performance.
• A second wired or fixed wireless circuit can carry close to full load, at higher recurring cost.
• A satellite link is a last resort where nothing else is available, with higher latency that affects voice and video.

Decide which functions must keep working during an outage. Phones and the customer-facing application are usually on the list. Large file transfers and software updates usually are not.

Automatic failover

A backup connection that someone has to plug in or configure by hand is not a backup. The firewall or router should detect the primary failure and switch traffic to the second path without anyone intervening, and switch back when the primary returns. Staff should notice, at most, a brief interruption to active calls.

Spare hardware and power

Redundant circuits behind a single firewall leave the original problem in place. Evaluate whether the firewall, the core switch, and the modem have spares or a high-availability pair. Confirm that the network equipment runs on battery backup long enough to ride out a short power event, and that the backup path does not depend on equipment that loses power at the same time as the primary.

Phones and cloud applications

Check how voice service behaves when the primary circuit fails. Some phone systems register to a single public address and need a configuration change to survive a failover. Cloud applications that restrict access by source IP address can lock out staff when traffic leaves through the backup connection with a different address. These are the failures that surface during the first real outage unless someone tests for them in advance.

A Checklist for Reviewing Your Connection

Use this list to find out where the company stands before any purchase.

1. Document every internet circuit at every location: provider, speed, contract end date, and the physical path into the building.
2. Identify which business functions fail if each circuit goes down.
3. Confirm whether a second circuit, if one exists, uses a different provider and a different physical path.
4. Check whether failover is automatic, and when it was last tested by deliberately disconnecting the primary.
5. Verify that the firewall and core network equipment have no single point of failure the second circuit does not cover.
6. Test voice calls, remote access, and IP-restricted cloud applications over the backup path.
7. Review the provider contract for the repair commitment, not just the advertised uptime. A service-level agreement that credits the bill does not restore the connection.
8. Assign a named owner for this review and put it on the annual calendar.

Where Redundancy Falls Short

A second connection protects against one failure mode, and it helps to be clear about the others.

It does not protect against a failure in the cloud service itself. If the application provider has an outage, both of your circuits will connect to a service that is down. It does not protect against a compromised account, ransomware, or a bad configuration change pushed to the firewall. And it does not substitute for a plan that tells staff what to do when the network is degraded, who decides to send people home, and who contacts customers.

Redundancy also adds something to manage. Two providers mean two contracts, two renewal dates, and two sets of support contacts. Without a technology owner tracking them, the backup circuit can lapse, change hands, or get cancelled in a cost-cutting pass without anyone connecting the dots to resilience. The configuration drifts, the failover stops working, and the company discovers it only when the primary fails.

How This Fits With Your Existing IT Support

Elevaire Systems works alongside the managed service provider or internal IT staff a company already has. Your MSP keeps the network running, the tickets answered, and the devices patched. Fractional IT Leadership is the layer that asks whether the network is still the right one for the business: whether the connection design matches what the company now depends on, whether the contracts reflect it, and whether the cost of an outage has been written down and compared against the cost of preventing it.

That review is a strategic question, not an operational one, which is why it tends to fall between the two. The MSP is accountable for uptime of what exists. Nobody is accountable for deciding what should exist. Infrastructure Modernization work starts there.

Frequently Asked Questions

How much does it cost to add a backup internet connection?

Costs vary by location, provider, and the type of backup. A cellular failover is generally the lowest-cost option, a second wired or fixed wireless circuit costs more, and the firewall that manages automatic failover is an added line item if the current one does not support it. Quotes differ widely by market, so collect at least two for your address and compare them to the annual outage cost you calculated.

Is a cellular backup enough for a 100-person office?

For keeping critical functions running, often yes. For carrying the whole office at normal speed, usually not. Cellular connections tend to have lower bandwidth, higher latency, and possible data limits, so plan around which applications and phones must stay up, and use quality-of-service rules to prioritize them.

How does this work with our existing MSP?

Your MSP typically operates the firewall, monitors the circuit, and handles the provider when something breaks. A redundancy project adds a second provider, a failover configuration, and a testing schedule that your MSP can run. The decision about how much resilience the business needs is a leadership call that someone has to own, and that is the gap Fractional IT Leadership fills.

How do we know if our second circuit really protects us?

Test it. Disconnect the primary during a planned window and confirm that voice, remote access, and your key cloud applications keep working on the backup path. Also confirm that the two circuits use different providers and different physical routes into the building. A failover that has never been tested is an assumption.

How do we get started?

Start with the inventory in the checklist above: every circuit, every location, and which business functions depend on each. Then calculate what an hour of outage costs your company using your own payroll and revenue numbers. Those two documents tell you whether the risk justifies a fix and how much resilience to buy.