White Paper
Detection

Prevention keeps most attacks out.Detection decides what the rest will cost you.

When Detection Champions Over Prevention

A rack of servers with green network cabling
A rack of servers with green network cabling. Photo: Abigor, CC BY-SA 3.0, via Wikimedia Commons.

Summary. Prevention controls are essential but never complete. Organisations that assume a breach will happen, and invest in finding it quickly, limit the damage far more effectively than those that rely on the perimeter alone.

The Appeal of Prevention

In data security, prevention has long been treated as the highest level of protection. It is easy to see why. An intrusion prevention system that blocks an exploit, an email filter that removes a malicious attachment or a firewall rule that closes an exposed service stops the problem before it starts. Detection, by comparison, sounds like an admission of defeat: it tells you that something has already gone wrong.

Prevention also fits budgets and audits neatly. Controls can be listed, configured and checked. A detection programme, which depends on people, judgement and constant tuning, is harder to measure.

Why Prevention Always Leaks

  • Credentials are the new perimeter. An attacker with a valid password, obtained through phishing, reuse or a help desk call, does not need to break anything.
  • Unknown vulnerabilities exist in every large software estate, and some are exploited before a patch is available.
  • Trusted suppliers and software updates can carry an attacker straight past the gate, as several supply chain attacks have shown.
  • People make mistakes, and attackers only need one.

None of this means prevention is pointless. It means prevention reduces the number of intrusions, while detection determines how long the remaining ones last and how much they cost.

Assume Breach

"Assume breach" is the mindset that follows: plan and invest as if an attacker is already inside, because sooner or later one will be. The NIST Cybersecurity Framework reflects this balance. Alongside identifying assets and protecting them, it gives equal weight to detecting, responding to and recovering from incidents, and its 2024 revision added governance as an overarching function.

In practice, assuming breach means asking different questions. Not only "how do we stop phishing?" but "if a laptop is compromised tonight, how long until we know, and what can the attacker reach in the meantime?"

What Good Detection Needs

  • Visibility across identity systems, endpoints, network traffic and cloud services, with logs kept long enough to investigate.
  • High-fidelity signals, because analysts can only act on what they trust. Deception is valuable here: decoys and honeytokens produce very few false alarms.
  • Clear ownership of alerts and the authority to act on them, including outside office hours.
  • Practice, through exercises and red team tests that check whether alerts actually fire and responders know what to do.

Measuring Success

Two numbers tell most of the story: the time to detect an intrusion and the time to contain it. Industry reports show these have fallen sharply over the last decade, from many months to days in typical investigations. The organisations at the fast end of that range are not those with the most prevention tools, but those that watch their own networks closely and act on what they see.

An Example from Online Gaming

Online casinos and sportsbooks illustrate the balance well. They invest heavily in prevention, such as DDoS protection, web application firewalls and strong authentication, yet their fraud and security teams spend most of their time on detection: spotting account takeovers, bonus abuse rings and suspicious withdrawals that got past every gate. In that industry, detection is the daily work, and prevention is what keeps its volume manageable.

Back to Top