The Only Place Silos Belong

By Ryan Kiechle, CIO, Annuity Health

Most organizations spend years trying to break down silos. In cybersecurity and business continuity, silos are often the only thing that will save you.

The uncomfortable truth is that most continuity failures do not start with a major outage. They start small. One click. One compromised credential. One vendor connection that never got shut off. From there, the issue is not whether you have backups. The issue is whether your environment is built to contain the problem, or whether it lets it spread.
We used to call this disaster recovery. That term does not really fit anymore. What we are dealing with now is survivability engineering.

Systems are more connected than ever. Workforces are more remote. Vendors are tied into everything. Attackers are more patient, and in many cases more automated. In that kind of environment, redundancy alone is not enough. Segmentation is what determines whether a mistake stays small or turns into a crisis.

Collaboration between people is good. Lateral movement between systems is not.

Mapping the Entry Points Before They Map You

Every connection to your network is a potential entry point. If you have not mapped them, someone else will.

In most environments, exposure grows faster than leadership realizes. When something goes wrong, it usually comes from one of three places.

The first is the human element.

Phishing, social engineering, MFA fatigue, impersonation. The human is still the weakest link, and that is not going to change. You can train people, you can put controls in place, but you cannot remove the risk completely.

The second is the vendor web.

Remote admin tools, integrations, temporary access, third-party platforms, misconfigured permissions. A lot of the worst breaches start with a trusted connection that was never locked back down. Someone needed access for a project, the door stayed open, and eventually someone found it.

The third is the environment itself.

Remote work, unmanaged endpoints, identity sprawl, legacy systems. The more complex the environment gets, the harder it is to see every path into it. Attackers only need one.

Security teams, including groups like InBalance IT, have been talking for a while about how AI-assisted attacks are shrinking the time between finding a weakness and exploiting it. Automated tools can scan faster than most teams can respond. In that kind of landscape, a perimeter firewall is not enough. Critical systems have to be isolated by design.

Containment Beats Confidence

One of the biggest mistakes organizations make is assuming prevention will always work.
It will not.
You cannot eliminate human error. You cannot eliminate vendor risk. You cannot eliminate new attack methods.
What you can do is build your environment so that one mistake does not become a company-wide failure.

This is where segmentation stops being a technical detail and becomes a leadership decision.
People should be able to collaborate.
Their machines should not automatically trust each other.
Their permissions should not open doors they do not need.
Good segmentation limits blast radius.
It keeps a breach in one area from reaching another.
It stops a stolen credential from unlocking everything.
It keeps your backups from living on the same plane as the threat.
That is what makes failure survivable.

Lessons From Recent Breaches

We have seen this play out more than once across the industry.

The Change Healthcare disruption showed how dangerous concentration risk can be. When attackers stay inside an environment long enough to affect both production systems and recovery layers, restoration becomes slow and painful. Backups exist, but they are not usable when you need them most.

More recently, the Conduent breach highlighted another problem. Exposure does not stop at your perimeter. Third-party platforms often hold sensitive data for millions of people who never even knew the company existed. In that case, attackers were reportedly inside the environment for weeks before detection, and notifications came long after the damage was done.

Both situations point to the same issue.
Continuity planning cannot start with restoration.
It has to start with architecture.
The real question is not how fast you can rebuild.
It is how much damage can spread before you contain it.

A Simple Model: Map, Segment, Prove

For organizations that want to move past the binder on the shelf and toward real resilience, I usually come back to three things.
Map. Segment. Prove.

Map every way someone connects to your environment.
Not just employees, but vendors, devices, applications, and anything else touching your data.

Segment critical systems from routine activity.
Segment sensitive data from general access.
Segment backups from production.
If everything can talk to everything else, you do not have containment.

Prove that it works.
Do not assume backups will restore.
Test them.
Run drills.
Review access.

Assume failure and see what actually happens.
By the time you need your continuity plan, every decision that mattered has already been made.

Questions Leadership Should Be Asking

Boards and executive teams do not need every technical detail, but they should be asking better questions.

Where is our blast radius unbounded?
How many vendor connections do we have that no one has reviewed recently?
Are our backups reachable with normal administrative credentials?
What parts of the business could keep running if our primary systems were offline for a day?
Are we measuring time-to-restore, or time-to-operability?

Those are not IT questions.
Those are continuity questions.

The Bottom Line

Continuity is not a document.
It is not a checklist.
It is not a once-a-year exercise.
It is an architectural discipline.

Organizations spend years trying to remove silos from their culture, and that makes sense. But in cybersecurity and business continuity, segmentation is not a weakness. It is what keeps one mistake from becoming a systemic failure.

You may not be able to stop every intrusion.
You can decide how far it is allowed to go.
That decision gets made long before the day you need it.