Your firewall is not a checkbox: what a business-grade edge actually does
The difference between the ISP router and a real firewall, what the subscriptions buy and the 5 configuration basics that matter more than brand.
· Jake Schaaf, Founder of Atticus Rowan
Walk into the network closet of a typical 15-person office and follow the internet cable. In a surprising number of businesses it lands on the same combination modem-router the ISP shipped 6 years ago, with the default admin password on a sticker and every device in the company sitting directly behind it. The business “has a firewall” in the sense that the box technically contains one. It has never been configured, updated or looked at.
On a cyber insurance application, that box and a properly run business firewall produce the same checked box. In an actual incident they produce very different months.
What the ISP box does and does not do
The carrier gateway does 1 job: move traffic. It performs basic network address translation, which incidentally blocks unsolicited inbound connections, and that is roughly where its security story ends. What it does not do:
- Inspect traffic for known attack patterns
- Block outbound connections to known-malicious infrastructure
- Separate guest Wi-Fi, office devices and servers from each other
- Log anything you could reconstruct an incident from
- Receive security patches on any schedule you control
A business-grade firewall from any of the mainstream vendors does all of the above. For a 10 to 30 user office the hardware runs roughly $500 to $2,500, plus an annual subscription. It is one of the least glamorous purchases in IT and one of the most defensible.
What the subscription actually buys
The recurring fee on a business firewall is not vendor rent. It funds the parts that do the active work:
- Intrusion prevention. The firewall compares traffic against a continuously updated library of attack signatures and drops matches. Without the subscription the library goes stale in weeks
- Web and content filtering. Blocks connections to known-malicious and newly registered domains, which quietly kills a large share of phishing payloads and malware callbacks
- Geo-blocking. Most 20-person Ohio businesses have no reason to accept inbound connections from the other side of the world. Turning that traffic off shrinks the attack surface for free
- Application control. Visibility into what is actually leaving your network, which pairs naturally with the SaaS discovery problem we covered in shadow IT and the SaaS inventory
The uncomfortable part: the firewall itself is a target
The last few years of security headlines have included a steady drumbeat of firewall and VPN appliance vulnerabilities being exploited for initial access. Attackers like edge devices for the same reason businesses do: they sit on the internet, they are always on and nobody watches them. A firewall that has not been patched in 2 years is not a control. It is a doorway with a maintenance contract nobody renewed.
This is the part the checkbox mindset misses. Owning the appliance is maybe a third of the value. Operating it is the rest.
5 configuration basics that matter more than brand
Vendor debates are mostly noise at this size. Configuration discipline is not. The 5 things we check first on any firewall we inherit:
- No management interface on the internet. Administration happens from inside the network or through a dedicated secure channel, never from a login page exposed to the world
- Patching on a schedule. Firmware updates applied within days of release for security fixes, not annually. This single habit would have prevented a large share of the edge-device breaches in recent memory
- Outbound rules that say no to something. Default configurations allow all outbound traffic. Tightening outbound, even modestly, breaks malware callbacks and data exfiltration paths
- Real network separation. Guest Wi-Fi, cameras, printers and servers in separate zones with rules between them. This is the firewall’s half of the work we described in network segmentation for small business
- Logs that go somewhere. When something happens, the first question is “what talked to what, when.” A firewall logging to itself with a 2-day buffer cannot answer it. Ship logs to storage that survives the incident
The lifecycle nobody budgets
Firewalls have end-of-life dates like everything else. Vendors typically support a model for 5 to 7 years, after which security updates stop even if the subscription renews. An end-of-life firewall fails insurance and customer security reviews for the same reason an end-of-life operating system does. Put the replacement on the same staggered hardware budget as your laptops, the way we laid out in the hardware lifecycle post, and the spend becomes a planned few hundred dollars a year instead of an emergency.
5 questions to ask whoever runs yours
You do not need to become a network engineer to supervise this well. Ask your IT provider or internal admin these 5 questions once a year and listen for specific answers:
- What model is our firewall and when does the vendor stop supporting it?
- When was the last firmware update applied, and who applied it?
- Is the management interface reachable from the internet? The only good answer is no
- If a workstation started sending data to an unfamiliar country at 2 a.m., would anything alert us?
- Where do the logs go, and how far back could we look after an incident?
Confident, dated, specific answers mean the edge is being operated. Vague answers mean you own an appliance, not a control. The same annual conversation is a natural place to review the rest of the perimeter, including remote access, since VPN endpoints share the firewall’s exposure profile and its patching discipline. If the answers worry you, that conversation is also a fair test of whether your current provider fits the model we described in break-fix to managed IT.
Where this fits in the bigger picture
A well-run firewall does not replace MFA, EDR or backups. It sits in front of them and cuts the volume of junk they have to handle. In a managed relationship the firewall is simply operated as part of the stack, patched, monitored and reviewed, which is part of what the per-user price actually covers in what managed IT costs.
If you are not sure which category your edge falls into, the test is 1 question: who applied the last firmware update, and when? If nobody knows, Atticus Rowan will happily take a look. Reach out and we will tell you plainly whether the box in your closet is a firewall or a formality.
Related insights
More on Security tools →August 22, 2026
Dark web monitoring: real signal or scare marketing?
What dark web monitoring actually tells you, what it cannot do and how to tell a useful alert from a recycled 2019 breach dump used as a sales prop.
July 20, 2026
Is Microsoft 365 Business Premium worth it for a 15-person office?
What the roughly $10 per user per month jump from Business Standard to Business Premium actually buys, and when Standard is genuinely enough.
July 15, 2026
Security awareness training a 20-person company will actually complete
The annual 45-minute security slideshow fails because it is designed for auditors, not employees. Short monthly training and blame-free phishing simulations work better and cost less than most owners expect.