Key Takeaways
Trace every breach to six recurring weaknesses, then fund the fixes
Start with the causes. The book groups most breaches into six technical root causes: unencrypted data, phishing, malware, third-party compromise or abuse, software vulnerabilities, and employee mistakes other than phishing. Behind them sit three organizational failures: security is not prioritized, funded, or executed well enough. The authors argue this practical map is more useful than trying to treat every compliance requirement as an equally urgent problem.
The cases make the pattern visible. Target involved phishing, malware, and a compromised supplier. Equifax was breached through a known software vulnerability that went unpatched. Capital One’s cloud incident involved a software flaw and overly broad access. Different headlines, recurring weaknesses. The map helps teams ask which exposures matter most in their own systems and assign people and money to reduce them.
The framework’s strength is its usefulness as a triage tool: it turns a huge, shifting threat landscape into a short set of questions that business leaders can act on. It also connects technical failures to management choices, rather than treating a breach as a freak event or a lone employee’s mistake. The categories overlap, however. Phishing can deliver malware, and weak third-party controls can expose unencrypted data. That overlap is not a defect if the framework is used to find multiple points of intervention, rather than to force every incident into one exclusive box.
Assume intruders will spread from one compromised account to the next
A break-in is a beginning. The attacker lifecycle describes how an intruder gains an initial foothold, explores internal systems, takes over more accounts or machines, and works toward a specific goal while trying to stay unnoticed. This is why defenders should not stop at changing one stolen password or removing one infected computer. They need to find the attacker’s broader access and contain it.
OPM shows the cost of missing the second intruder. One group spent about 18 months gathering technical information about the agency. Another used stolen contractor credentials and malware to reach sensitive systems, ultimately taking background investigations on 21.5 million people. A zero trust approach treats each user and device as untrusted until verified, rather than assuming anything already inside the network is safe. Segmentation can also limit how far a compromised machine can reach.
This view resembles containment planning in other high-consequence systems: the question is not only how to prevent failure, but how to keep one failure from propagating. A single perimeter is a poor fit for organizations whose employees, vendors, cloud services, and devices connect from many places. Repeated verification and segmentation make the intruder’s path harder and can reduce the damage if prevention fails. These controls do not guarantee detection, though. They work best when paired with reliable logs and an incident process capable of noticing suspicious movement.
Treat suppliers, platforms, and acquisitions as part of your security perimeter
Trust travels through relationships. Target’s attackers used credentials stolen from Fazio Mechanical, an HVAC supplier. JPMorgan Chase was entered through Simmco, which ran a charity-race website where employees reused corporate credentials. Facebook’s developer platform enabled Cambridge Analytica to obtain and use profile information in ways users did not expect. In each case, the organization’s exposure extended beyond systems it directly operated.
An acquisition inherits risk. Marriott bought Starwood without uncovering an intrusion that had begun years earlier. Marriott then faced the breach, regulatory penalties, and the burden of untangling old systems and security gaps. Vet suppliers according to the data and access they receive, monitor their posture, and examine acquisition targets before a deal closes. The same scrutiny applies to customers who receive data and developers who can access a platform’s users.
The idea of a perimeter becomes less about a company’s network boundary and more about its web of dependence. Procurement, product design, marketing, and mergers all make security decisions, even when they do not use that name. This also explains why a partner’s incident can become a reputational crisis for the organization that supplied the data. Due diligence has limits, particularly when an acquisition target is secret or time is short. Still, treating security review as part of the transaction cost is more realistic than assuming integration will erase inherited weaknesses.
Encrypt and minimize sensitive data before a break-in makes it public
Possession creates exposure. Encryption makes data unreadable without the necessary key, so a stolen laptop or disk need not automatically become a data breach. But encryption is only useful if the keys are protected separately and access is limited. The book distinguishes data at rest, data moving across networks, and data being processed. Each state needs appropriate protection.
Keeping less can matter as much as locking more. The OPM breach exposed fingerprints that victims cannot replace, alongside intimate background information about employees and their families. Marriott retained millions of passport numbers and extensive travel records. Data minimization means collecting and keeping only information the organization genuinely needs. Less retained data means fewer valuable targets and less harm if controls fail. A security program should inventory sensitive data, find unencrypted copies, and set retention rules rather than treating every old record as permanent.
Encryption and minimization reduce different parts of risk. Encryption can make stolen storage less useful, while minimization reduces the amount available to steal in the first place. Together they resemble a basic principle of harm reduction: avoid creating unnecessary hazards, then add safeguards around what remains necessary. There is a practical tension, since data can support services, fraud detection, and analysis. The right question is not whether data has any possible future use, but whether that use justifies the durable risk of retaining it.
Give security leaders authority, budget, and access to executive decisions
Leadership sets the operating conditions. The book calls failure to prioritize, invest in, and execute on security the meta-level root causes of breaches. OPM had years of audit warnings but spent about $7 million annually on cybersecurity and lacked basic protections, including multi-factor authentication. Its leaders did not respond forcefully enough to known weaknesses. The agency held unusually sensitive records, yet security was not treated as commensurate with its mission.
A title alone is not power. A CISO needs a capable team, a budget, and authority to influence work across technology, legal, and business units. The authors recommend a direct relationship with the CEO and regular board access. Security should be discussed alongside business risks and plans, including product launches and acquisitions. Funding should follow an assessment of the organization’s most valuable assets and its actual gaps, not simply a peer average or a vendor’s pitch.
Security is a governance problem because the costs of prevention and the benefits of growth often fall on different teams. A product group may gain from launching quickly, while customers and the organization bear much of the risk. Executive authority can make that trade-off explicit and resolve conflicts before they become incidents. Yet reporting lines and spending totals are proxies, not outcomes. An empowered security leader can still choose poor priorities, and a larger budget can be wasted. Leadership must pair authority with evidence about whether controls are working.
Design for human fallibility, then automate routine security work
Do not build plans around perfect people. Equifax sent warnings about a critical Apache Struts flaw, but the relevant developer did not receive the message. A vulnerability scanner then failed to find the flaw, and no closed-loop process verified that patching had succeeded. Capital One’s cloud breach likewise involved an access policy that gave a web firewall role more access than it needed. These were system and process failures, not problems that a reminder to be more careful could reliably solve.
Make safe behavior the default. The principle of least privilege gives each user or program only the access needed for its task. Secure defaults, automated configuration checks, and patch tracking reduce reliance on memory and manual follow-up. Automation still needs monitoring: a failed process can create a new blind spot. The goal is not to eliminate human judgment, but to reserve it for decisions that genuinely require it.
This approach aligns with human-factors engineering, which studies how systems can accommodate predictable limits in attention and memory. A workflow that depends on hundreds of people remembering every setting will fail at scale, even when those people are skilled. Defaults and verification move protection into the infrastructure instead of asking individuals to supply it continuously. Automation can also amplify mistakes, however. A badly designed rule can spread quickly across many systems, so automated changes need testing, monitoring, and a clear way to reverse them.
Make security a daily operating discipline, not a reaction to breaches
The book’s habits turn intent into practice. Its seven habits of effective security are:
1. Prepare proactively and stay alert to threats.
2. Tie security work to the organization’s mission.
3. Build security and privacy into systems and products.
4. Treat security as the goal, with compliance as a byproduct.
5. Measure whether protections work.
6. Automate repeatable safeguards.
7. Improve continuously.
Practice matters more than slogans. The authors recommend ongoing workforce training, rehearsed incident response, and regular improvement rather than an annual exercise that checks a box. They also urge organizations to maintain outside support networks and prepare communications before a crisis. Security is never complete, so teams should use measurements and incidents to improve the system instead of treating a successful audit as proof that risk has disappeared.
The habits combine culture, engineering, and operations rather than pretending that cybersecurity is only a technical specialty. That integration is valuable because a secure design can be undermined by a rushed deployment, and good technology can sit unused if people do not know how to respond. Continuous improvement is especially useful in a field where threats and systems change together. The challenge is avoiding ritual: training completion, automated tools, and frequent metrics can become activity measures that look impressive while leaving the underlying exposure untouched.
Ask boards to test whether safeguards are consistent, adequate, reasonable, effective
CARE gives directors a practical test. The framework asks whether security is consistent across an organization, adequate for the data and mission, reasonable given the organization’s circumstances, and effective at producing protection. These questions are broader than asking whether a compliance checklist was completed. Target had PCI compliance but still suffered a major breach, showing that certification alone does not establish that defenses work.
Tell the risk story before showing the dashboard. Directors need to understand what the organization protects, who might target it, and how a breach could affect customers or the business. Then leaders can support that explanation with measures, such as how quickly critical vulnerabilities are fixed or how often employees report phishing. Reports should be calm and evidence-based, with bad news reaching leadership promptly rather than being softened or delayed.
CARE helps boards discuss security without requiring every director to become a specialist. It focuses attention on whether protections fit the context and produce results, a more useful governance question than whether a control exists on paper. Metrics add discipline, but they can also distort behavior when a target becomes more important than the risk it was meant to represent. A low phishing-test failure rate, for example, says little about malware defenses. Boards need a small set of measures tied to specific threats, alongside narrative and candid discussion.
Put digital seatbelts on every account and device you control
Start with identity and access. Enable two-factor authentication on important accounts, use a password manager to create unique passwords, and freeze credit when not applying for new credit. Identity protection is broader than credit monitoring and can help detect misuse beyond credit files.
Harden the devices around you. Change the home router’s default password and keep its software current. Use anti-malware, enable device encryption, back up important files to the cloud, test that backups can be restored, and install system updates. For unexpected calls or messages, do not provide sensitive information on the caller’s say-so. Contact the institution through a number already known to be genuine. These steps cannot prevent a company holding your data from being breached, but they can make stolen information harder to use against your own accounts and devices.
The consumer checklist translates enterprise security controls into ordinary actions, a useful shift because personal devices and accounts are now part of work, finances, and family life. The metaphor of a seatbelt captures an important distinction: a precaution reduces harm without promising that accidents will never happen. Some protections, such as credit freezes and authentication options, vary in convenience and availability. The best setup is one a person can maintain. A complex routine that is abandoned is less protective than a simpler set of safeguards used consistently.
Direct cybersecurity talent and capital toward risks the market still neglects
The field needs more than hackers. The book cites about 500,000 unfilled cybersecurity jobs in the United States in 2021 and describes paths for analysts, security architects, and future CISOs. People can enter from software development, IT operations, auditing, project management, or product roles. Security operations can be a starting point, especially for those who learn to automate routine investigation as well as respond to incidents.
Investment should follow exposure, not fashion. More than $45 billion went into cybersecurity companies from 2003 to 2020, but the authors argue that some areas remained underfunded, including Internet of Things security, privacy, fraud detection, analytics, and protecting AI systems from attack. Their investment analysis uses imperfect company data, so its rankings are directional rather than definitive. The larger lesson is to connect funding to recurring breach causes and emerging dependence on insecure devices and services.
The workforce and investment argument broadens cybersecurity from a company’s internal program to shared capacity across an economy. Automation can reduce repetitive analyst work, but it also raises the value of people who can interpret uncertain evidence, design controls, and coordinate across departments. Investment categories are less reliable than investment outcomes: a fashionable label does not prove that a product reduces risk. The book’s own caution about imperfect data is important. Capital allocation should be tested against demonstrated need, adoption, and measurable reduction in exposure.
Analysis
The book’s main achievement is to connect breach histories to decisions that organizations can change. Its case studies are not merely cautionary tales: the same recurring weaknesses appear in very different settings, from a retailer’s supplier access to a credit bureau’s unpatched application and a government agency’s neglected systems. The practical value lies in moving from incident detail to a shared vocabulary for executives, security teams, consumers, and people entering the field.
The six-cause model is useful precisely because it is compact, but it is not a strict scientific taxonomy. Its categories mix attack methods, missing controls, and human actions. A phishing email may install malware, and a third party may expose unencrypted data. That means the model works best as a diagnostic checklist, not as a set of mutually exclusive explanations. The book also relies on reported breaches, which cannot capture every incident, especially those that remain undetected or undisclosed. Its claims about prevalence therefore deserve more caution than its practical recommendations.
The most important underlying tension is economic. Organizations benefit from speed, data collection, and interconnected services, while many costs of weak security are paid later by customers, employees, and the public. Rules and fines can shift incentives, but compliance alone does not establish that defenses are effective. The book’s strongest response is a combination of leadership accountability, system design that anticipates human error, and measurable operational practice. Its broader argument is that cybersecurity is neither a one-time purchase nor a specialty function operating outside the business. It is ongoing maintenance of the trust on which digital services depend.
Review Summary
Big Breaches receives mostly positive reviews, with readers praising its comprehensive overview of cybersecurity failures and mitigation strategies. Reviewers appreciate the book's accessible language, practical advice, and analysis of major security breaches. Some highlight its relevance to current events and potential for generating new product ideas. The first half of the book, focusing on case studies, is particularly well-received. A few criticisms include a tedious second half and overgeneralizations about nation-state actors. Overall, readers find it informative for both cybersecurity professionals and laypeople.
People Also Read
Glossary
meta-level root causes
organizational failures behind breachesThe book’s term for three broad organizational failures that make technical breaches more likely: not prioritizing security, not investing enough in it, and not executing security initiatives effectively. These causes explain why known risks and available protections may remain unaddressed, even when a company has security staff or formal policies.
technical root causes
six recurring breach weaknessesThe book’s practical grouping of common technical causes behind breaches: unencrypted data, phishing, malware, third-party compromise or abuse, software vulnerabilities, and employee mistakes other than phishing. The categories are meant to help organizations focus defense and investment, not to serve as mutually exclusive scientific classifications.
attacker lifecycle
an intruder’s path to a goalA model of how an attacker can progress through an organization after gaining an initial foothold. The intruder may expand access to accounts and machines, explore internal systems, maintain persistence, and move toward sensitive data or another objective while trying to avoid detection. Understanding this progression supports containment beyond the first compromised device.
zero trust
verify each access requestA security approach that does not automatically trust a user or machine merely because it is already inside an organization’s network. Access is verified when requested, and the user’s identity and device security state can be checked continuously. The approach reduces reliance on a single network perimeter and can limit an attacker’s movement.
CARE
a board-level security testA framework for evaluating whether organizational security is Consistent, Adequate, Reasonable, and Effective. It asks whether protections are applied reliably, match the sensitivity of the information and mission, fit the organization’s circumstances, and produce the desired protective result. The book presents CARE as a useful basis for board and regulator discussions, not a guarantee against breaches.
technical debt
deferred system maintenance costsThe accumulated cost and risk created when systems are not maintained, upgraded, or replaced as needed. A business may continue operating on old software, but later security fixes and integration can become difficult and expensive. In the Marriott case, Starwood’s aging systems and costly maintenance needs complicated the security consequences of an acquisition.
server-side request forgery (SSRF)
tricking a server into making requestsA vulnerability that allows an attacker to cause a server to send requests on the attacker’s behalf. In the Capital One breach, the attacker used this flaw to reach a cloud metadata service that should have been available only to a local server process. The resulting requests exposed credentials that could then be used to access data stores.
security keys
hardware-based login verificationPhysical authentication devices that provide a cryptographic proof during login, in addition to or instead of relying only on a password. The book describes keys using standards such as WebAuthn and FIDO. Because the verification is tied to the legitimate website’s domain, a stolen password alone is not enough for a phisher to authenticate as the user.
Seven Habits of Highly Effective Security
a recurring security practice frameworkThe authors’ framework for building organizational security into routine work. Its habits emphasize preparation, mission alignment, secure design, prioritizing security over compliance as an end in itself, measurement, automation, and continuous improvement. The framework is intended as an enduring mindset adapted to an organization’s needs, rather than a one-time checklist or a prescribed technology stack.
Download PDF
Download EPUB
.epub digital book format is ideal for reading ebooks on phones, tablets, and e-readers.