What a cyberattack really costs, and how to bring the bill down
Where the cost of a cyber incident comes from, from response to regulatory fines, and how testing your defenses before an attacker does reduces it.

When an incident happens, the bill does not arrive all at once. It builds up over weeks: hours of response, systems down, notices to customers and to the regulator, fraud, contract penalties and customers who do not come back. Knowing where each part of the bill comes from helps you decide where to invest before the incident, not after it.
How an incident turns into cost
Most serious incidents combine tactics that MITRE ATT&CK describes, not always in this order: initial access (phishing, a leaked credential, a flaw in an exposed application), privilege escalation, lateral movement and, finally, impact. The damage usually comes from data exfiltration, from servers encrypted for extortion (technique T1486, Data Encrypted for Impact) or from direct fraud in payment systems.
Every step an attacker takes unseen widens the blast radius. An intrusion contained on the first server costs an investigation. The same intrusion found after it reached the domain controller costs a rebuild of the environment. That is why the cost depends less on the kind of attack than on how long it went unnoticed and how many systems it reached.
The parts of the bill
- Incident response. Forensic investigation, containment, eradication and recovery. NIST’s guide on the subject, SP 800-61 Rev. 3, ties this response to the company’s risk management. It includes internal staff hours, outside specialists and legal counsel.
- Business interruption. Sales that do not happen, production stopped, queues at customer service. In retail and financial services, every hour offline has a value your business teams can estimate.
- Notification. Under Brazil’s data protection law (LGPD), the controller must notify the national authority (ANPD) and the people affected of any incident that may cause them relevant risk or harm. The ANPD states that its Security Incident Reporting Regulation (Resolution CD/ANPD No. 15 of 2024) sets, as a rule, a deadline of three business days from the moment the incident becomes known, doubled for small processing agents. Without an up-to-date data inventory, just finding out who must be told uses up that time.
- Sanctions. Article 52 of Law 13,709/2018, in the updated text published by Brazil’s Chamber of Deputies, provides for a fine of up to 2% of the revenue of the company, group or conglomerate in Brazil in its last fiscal year, net of taxes, capped at R$ 50 million per infraction. The same law provides for other sanctions, such as making the infraction public and blocking the data involved.
- Fraud and extortion. Diverted transfers, altered payments, a ransom demanded. Paying the ransom does not guarantee the data comes back, nor that it will not leak later.
- Contracts and reputation. Contractual penalties, corporate customers who demand an audit or switch suppliers, insurance that costs more at renewal. It is the hardest part to measure and the one that lasts longest.
What the numbers say
The Cost of a Data Breach report, published by IBM with the Ponemon Institute, is a well-known reference on the subject. In its 2026 edition, the global average cost of a data breach was US$ 4.99 million, 12% more than the year before. The report itself attributes the rise to the costs of detection, escalation and lost business.
For Brazil, IBM’s announcement of the 2026 edition gives an average cost of R$ 7.44 million per breach, based on 42 Brazilian organizations breached between March 2025 and February 2026. The same announcement reports that the use of offensive security testing was associated with average savings of R$ 579,422 per incident.
Read these figures as an order of magnitude, not as a forecast for your company. They are averages over a sample, and an association between a factor and a lower cost does not prove that the factor alone caused the difference. What they do show is the composition: detection, escalation, response and lost business weigh on the bill, and those are costs a prepared company can reduce.
Where testing early reduces the cost
An offensive test works on the variables that weigh most on the bill.
- Fewer paths to impact. A pentest chains flaws the way an attacker would and shows which sequence leads from an exposed application to customer data. Fixing one link in that chain can cut the whole path.
- Earlier detection. A Red Team exercise measures whether the SOC notices the attack and how quickly. Every step that passes without an alert becomes a detection use case to build.
- A rehearsed response. Knowing before the incident who decides to isolate a server, who talks to the regulator and where the personal data is cuts the time lost in meetings during the crisis.
- Priority by business value. A critical flaw in an isolated test system can matter less than a medium one in the payment system. Translating findings into financial risk helps the business approve the right fixes first.
An illustrative example
This scenario is a composite of flaws common in tests, not a real customer’s case. A retailer has a partner portal with an access control flaw (A01:2025, Broken Access Control, in the OWASP Top 10): by changing an identifier in the URL, one partner sees another’s orders. On the same portal, a service account uses a password that also works on the integration server.
In a real incident, the sequence would be this: the attacker enumerates orders carrying customers’ personal data, reuses the service account’s password, reaches the integration server and from there the orders database. The bill includes the forensic response, the portal offline during the fix, notices to the regulator and to the affected customers, since a leak of customers’ personal data can cause relevant risk, and renegotiation with partners who demand an audit.
In a pentest, the same chain appears in the report with proof of every step and three fixes: check on the server who owns each order, change the password and give each system its own credentials, and restrict the integration server’s network access to the database. The cost is the test and the fixes, in a window you choose.
What to ask a supplier
- Do findings come with the full exploitation chain and evidence for each step, or only a list of vulnerabilities?
- How does the report translate findings into business impact, and who on your team takes part in that assessment?
- How much of the test is manual, and how much is done by tools?
- Does the scope include the systems that hold personal data and payments, or only the easiest ones to test?
- Is the retest of the fixes part of the project?
- How does the supplier protect the data it can reach during the test?
The Leonnes pentest explores your applications, APIs and infrastructure by hand, chains flaws the way an attacker would and delivers, beyond the technical and executive reports, a financial risk report. After the fixes, the retest confirms what was closed.