Security Guide · For Owners & Managers

Practically Managing Risk with IP Allowlisting.

You don't have to be technical to make one of the highest-value security decisions available to you. Restricting your sensitive systems to an approved network address removes them from the open internet — so most attacks never reach you in the first place.

Shrink the Target Reduce Reachability Defence in Depth A Business Decision
The question worth asking

Start with reachability, not just passwords.

Most security conversations jump straight to "is our login strong enough?" That matters — but it assumes the whole internet already gets to try. A better first question for an owner or manager is simpler, and it changes everything.

Who can even reach it?

If a system faces the open internet, every scanner, bot and automated attacker on earth can knock on its door — continuously, day and night.

Who are they?

Sign-in and multi-factor prove identity — but only for people who already reached the door. They can't reduce how many people get to try.

How much could go wrong?

If something ever does slip through, the real question is how far it can travel. Managing risk means limiting reach and consequence, not just entry.

Don't make your systems easier to find than they need to be.

In plain terms

What IP allowlisting actually does.

Every device on the internet has an address — an IP. Allowlisting means your system agrees to accept connections from only the addresses you approve. Everyone else is turned away at the network, before they ever reach your application.

Your team or office
Approved IP addressthe one known front door
Your entityOS space
Sign-in & permissions

Everyone outside that approved path is stopped at the door.

Dedicated-IP VPNa fixed address for your team
Corporate VPN egressyour office network's IP
Cloud workstationa fixed cloud address
  • Automated scanners and bots simply can't see the system.
  • Opportunistic attacks that rely on reaching you don't land.
  • Your sign-in page stops being a target the whole world can poke at.
  • You shrink the problem to a handful of known, controllable paths.
One important caveat

An allowed IP is a gate, not a badge. It answers should this connection reach us — never who is this. Keep strong sign-in and least-privilege access working behind it. And prefer a dedicated IP that's yours alone: a "static" address shared across many VPN customers is weaker than it first looks.

A note on residential proxies

Attackers increasingly route traffic through ordinary home internet connections, so "this address looks normal" is no longer a reliable signal — a point reinforced by recent ASD guidance. Exact allowlisting sidesteps this entirely: it never asks whether an address looks trustworthy, only whether it is the specific one you approved. As telling good traffic from bad gets harder, reducing who can reach you becomes more valuable, not less.

The technical detail — IP reputation vs exact allowlisting →
Drawing the line

Which systems to gate — and which to leave open.

Allowlisting isn't for everything. It shines on systems the wider public never needs to reach. The test is simple: does the outside world actually need to get to this? If not, it doesn't need to be reachable.

Gate these

Only your people need them

  • Admin & operations dashboards
  • Internal apps, APIs and databases
  • Partner and supplier integrations
  • Management & infrastructure controls
  • Anything used only by staff or contractors
Reachable only from approved network paths.
Leave these open

The world genuinely needs them

  • Your public marketing website
  • Customer sign-up and login
  • Public APIs your customers call
  • Anything a stranger must be able to use
Protect with sign-in, rate limits and monitoring instead.

Gating your internal and administrative systems typically covers most of the risk with none of the disruption — customers never touch those systems, so they never notice the change.

A manager's hierarchy

From most exposed to most contained.

Not every option costs the same or protects the same. Here's roughly how the choices stack up — and where most teams land for a sensible balance of effort and protection.

Approach Who can reach it Attacker effort Practical fit
Open to the internet Everyone, everywhere Lowest Public services only
Shared VPN fixed IP You — and others on that IP Low Weaker than it looks
Dedicated VPN with fixed IP Only your organisation (via VPN) Higher Sweet spot for most
Dedicated cloud IP (Elastic IP) Only your organisation Higher Strong, no VPN dependency
Private network / zero-trust No normal public reach High Strong, more setup
Private + verified device No public reach, checked device Highest Strongest model

A dedicated VPN with a fixed IP is the practical starting point for most teams — much smaller exposure, without a major rebuild. A dedicated cloud IP does the same job with one fewer moving part. Either way, read the next section first: the VPN is a risk of its own.

Being honest about it

The VPN is a risk too — plan for it.

Whitelisting via a VPN doesn't remove risk; it moves it. You take away a large, well-understood risk and take on a smaller, different one. Worth doing — but only with eyes open, because the new risk shows up in a place people forget: keeping your own access working.

The risk it takes away
  • Exposure to every scanner and bot on the internet, around the clock.
  • Opportunistic and automated attacks that depend on reaching you first.
  • Your sign-in page being probed and fuzzed by the whole world.

This is the big one. For most organisations it is the dominant day-to-day threat — and it is exactly what a fixed-IP gate removes.

The risk it adds
A new link in the trust chain

The VPN account, provider and infrastructure now sit in front of your systems — so each becomes something that could be compromised. Real, but much smaller than internet-wide exposure. This is why the IP is a gate, never proof of identity.

A single point of failure for access

This is the one people miss. If the VPN drops, the IP changes, or the account lapses, legitimate users — including your own administrators — can be locked out of their own systems. That's a continuity risk, not a security one, and it points the opposite way: with no gate at all, a VPN outage could never lock you out.

Managing the continuity risk

The security case is strong, so the honest job isn't to avoid the VPN — it's to make sure the VPN can never become the reason no one can get in. Four low-cost habits do most of the work.

01

Keep a break-glass way in

A documented, tightly-controlled path back that doesn't depend on the VPN — used only in an emergency, and audited every time it is.

02

Have a backup path

A second dedicated IP or egress route, so one VPN outage can't cut off every administrator at the same moment.

03

Write the recovery down

So restoring access isn't improvised under pressure. Everyone who needs in knows exactly how, before they need to.

04

Monitor the gate itself

Watch the VPN and the fixed IP. If either changes or fails, you hear about it before your users and admins do.

So is it worth it? On balance, yes — a dedicated VPN with IP whitelisting removes far more risk than it adds, because internet-wide exposure is the bigger threat by a wide margin. But that judgement only holds if you treat access continuity as a first-class concern. Whitelisting without a break-glass plan can quietly trade a security gain for a continuity loss — and that trade is avoidable.

Defence in depth

One gate is never the whole answer.

Think of security as a series of independent checks. Allowlisting handles the first and does the heavy lifting on volume — the layers behind it make sure a rare, determined attacker still can't get far.

01
Reachability — who can even connect?
IP allowlisting removes most of the internet's noise at the door.
Network
02
Identity — who is this, really?
Multi-factor sign-in or passkeys confirm the person behind the connection.
Sign-in
03
Authorisation — what may they do?
Least-privilege access keeps each person and system to only what they need.
Permissions
04
Consequence — how far could it spread?
Contain and separate systems so a rare breach reaches very little.
Containment

IP allowlisting removes most of the pressure at the door; the layers behind it make sure the rare attacker who gets past one control still can't get far. That combination is what actually manages the risk.

Where to begin

A practical starting point.

You don't need to boil the ocean. A short, ordered set of steps gets most of the value — and each one is a decision an owner or manager can direct without writing a line of code.

List what's internal

Write down which systems only staff, partners or operations ever touch. Those are your candidates for gating.

Get a dedicated IP

A dedicated-IP VPN or a cloud workstation with a fixed address gives you one known front door to approve.

Allow it, block the rest

Restrict those internal systems to your approved IP. Everyone else is turned away before they reach the application.

Keep sign-in strong

The IP is the gate. Behind it, keep multi-factor or passkeys and short sessions — never rely on the address alone.

Right-size access

Give people and systems only the access they need. A smaller blast radius means a smaller problem if anything goes wrong.

Review it regularly

IPs and teams change. Revisit the allowlist so it keeps matching exactly who actually needs to get in.

Read next

Go a level deeper.

The technical companion to this page — for whoever will actually configure the allowlist, and for understanding why this control is getting more valuable, not less.

Technical guide

IP reputation is weakening. Exact allowlisting isn't.

Why residential proxies and autonomous AI erode reputation, geo and ASN controls — and why exact /32 allowlisting keeps its value. Includes the reference architectures for restricted and public services.

Read the technical guide →
The takeaway

Reduce who can knock on the door.

01

Shrink the target

Most attacks depend on reaching you. Take internal systems off the open internet and most of that pressure disappears.

Internet
—— blocked ——
Approved IP
Your systems
02

Don't over-trust the gate

A permitted IP decides who may connect — not who they are. Keep identity and permissions doing their own job behind it.

Reach the door
then still prove
Who you are
What you may do
03

Contain the rest

Assume something could still slip through one day, and make sure it can reach very little when it does.

Least access
+ separation
Small blast radius
Damage stays tiny

The goal isn't a perfect wall. It's to force any attacker down a much smaller, more specific path before they can even interact with your systems — and to keep the damage tiny if they ever do.