How Much Internet Bandwidth Does Your Business Actually Need?

Most bandwidth conversations start in the wrong place. Someone looks at a bill, sees a number in megabits, and asks whether a bigger number would fix the problem people keep complaining about. Sometimes it would. Often the complaint has nothing to do with the size of the circuit.

Sizing business internet correctly means counting what your applications actually consume, understanding which direction that traffic moves, and being honest about what happens when the circuit goes down. Here is how to work through it.

Start with what your applications ask for

Bandwidth planning is additive. Every simultaneous session takes a slice, and the total has to fit inside the circuit during your busiest hour, not your average hour.

A few reference points for planning purposes:

• A hosted voice call consumes roughly 100 Kbps in each direction once overhead is counted. Twenty concurrent calls is a couple of megabits, which is almost nothing on a fiber circuit and a serious problem on an oversubscribed cable connection.
• HD video conferencing runs in the low single-digit megabits per participant, and it runs in both directions at once. A conference room, three people on calls at their desks, and a weekly all-hands can stack up quickly.
• Cloud file sync, offsite backup, and large design or video files are the heaviest consumers, and they are almost entirely upstream. A single 10 GB backup pushed during business hours will saturate a small upload pipe and take the phones down with it.
• Everyday browsing, email, and line-of-business SaaS applications are modest per user but numerous.

Add those together for peak conditions, then add headroom. Planning to run a circuit at 90 percent of capacity is planning for congestion, because traffic arrives in bursts rather than in a smooth stream. A common rule of thumb is to size so that peak usage lands near 60 to 70 percent of the circuit.

Upload matters more than it used to

For most of the last two decades, business internet was sold on download speed because that is what consumer marketing conditioned everyone to compare. That framing no longer matches how offices work.

Hosted voice, video conferencing, cloud backup, remote desktop sessions, surveillance camera feeds sent offsite, and file sharing are all upstream-heavy. On an asymmetrical connection, where download might be ten or twenty times the upload, the upload side is what fails first. The symptom is confusing: web pages still load fine, so it does not feel like an internet problem, but calls break up and video freezes.

Symmetrical fiber solves this by giving you the same capacity in both directions. When someone tells you their internet is fast but their phone system sounds terrible, an asymmetrical circuit is one of the first things worth checking.

Shared versus dedicated is a different question than speed

Two circuits can advertise the same speed and behave nothing alike.

Shared broadband delivers a best-effort service across infrastructure that many subscribers use at once. The advertised speed is a ceiling, not a floor. Performance moves with neighborhood demand, which is why a connection that tested beautifully on a Tuesday morning struggles at three in the afternoon.

Dedicated internet access gives you the bandwidth you purchased, in both directions, with committed performance targets. That commitment is what a service level agreement documents. A real SLA specifies uptime, latency, packet loss, jitter, and response and restoration time, with the remedy spelled out when the provider misses. Ask for those numbers in writing before signing anything. If a provider will not put restoration time in the contract, you have learned something useful.

The gap between shared and dedicated shows up in exactly the applications businesses care most about. Voice and video are sensitive to jitter and packet loss, not just raw throughput. A circuit can show plenty of available bandwidth on a speed test and still deliver unusable calls because the path is unstable.

Count the sites, not just the seats

Bandwidth planning for a multi-site business has an extra layer. Traffic between your locations and traffic to the internet are different flows, and treating them as one pool leads to trouble.

If your branch offices reach applications, file shares, or storage at a headquarters or a data center, that traffic can ride a private Layer 2 connection instead of the public internet. Metro Ethernet and point-to-point circuits give you predictable latency and throughput between sites, they keep that traffic off your internet circuit entirely, and they avoid the overhead and variability of routing everything through a VPN tunnel.

The practical effect is that your internet circuit only has to carry internet traffic. Sizing gets simpler and more predictable, and the site-to-site applications stop competing with someone streaming a webinar.

Plan for the circuit being down

Sizing exercises tend to assume the connection works. Build the plan for the day it does not.

Ask what your business actually loses during an outage. For a firm that runs on cloud accounting software and a hosted phone system, an internet outage is a full stop. Card processing fails, calls do not arrive, files are unreachable. Once you can put a number on an hour of downtime, the cost of redundancy usually answers itself.

Redundancy has several forms, and they are not equivalent:

• A second circuit from the same provider over the same physical path protects against equipment failure but not against a contractor cutting the conduit.
• A second circuit over diverse media, such as fiber plus a wireless or 5G connection, protects against physical path failure.
• SD-WAN sits above both and makes failover automatic, steering traffic across available connections and moving sessions off a degraded circuit before users notice. Without it, failover usually means someone unplugging something and rebooting a router.

For voice specifically, ask whether your provider supports disaster routing, which redirects inbound calls to mobile numbers or another location when a site is unreachable. Calls that reach a person somewhere are better than calls that reach nothing.

A practical sizing checklist

Before you talk to any provider, gather the following:

• Peak concurrent users at each location, including hybrid staff who connect back to the office.
• Number of concurrent voice calls at peak, plus any contact center volume.
• Video conferencing usage at peak, counted by simultaneous participants rather than by license count.
• Backup and file sync volumes, along with the window those jobs run in. Moving backups outside business hours is often cheaper than buying more bandwidth.
• Applications that live in a data center or at another one of your sites, which may belong on a private circuit rather than the internet.
• Growth over the next 24 to 36 months, including any planned locations.
• Your tolerance for downtime, expressed in dollars per hour if you can manage it.

That list turns a vague request for faster internet into a specific engineering conversation. It also makes competing quotes comparable, because you will be able to tell whether two proposals are actually offering the same thing.

Getting a real answer for your building

Bandwidth on paper is one thing. What can be delivered to your address is another, and it depends on the fiber already in the ground, the building entrance, and the construction required to reach you.

Silver Star Telecom has spent more than 25 years engineering fiber, Ethernet, and voice networks across Oregon and Washington, and we design what we sell rather than reselling someone else's circuit and hoping for the best. If you want a sizing review for your locations, along with written service level commitments you can hold us to, call (360) 524-7498 or email sales@silverstartelecom.com.