Skip to content
Tutorials Center
My learning

Topics

Lesson 33 of 38 ยท Reporting and Remediation

Risk Rating and Evidence Handling

Learn how to rate penetration-test findings by likelihood and impact, adjust CVSS for the client's context, and collect, store and redact evidence responsibly.

A penetration test is only as useful as what the client does with it. A report with 60 findings all marked "critical" gets ignored; a report whose ratings are consistent, justified and backed by solid evidence gets fixed. This lesson covers the two foundations of a good report: a rating method you can defend, and evidence that proves each finding without creating new risk.

Severity, likelihood and impact

Risk is usually expressed as a combination of two questions:

  • Likelihood: how probable is it that someone finds and exploits this? It depends on how easy the flaw is to discover and use, whether an exploit is public, what access an attacker needs (internet, internal network, authenticated user) and whether existing controls would detect or block it.
  • Impact: how bad is it if it happens? Consider confidentiality (data exposed), integrity (data or systems changed), availability (outages), financial loss, legal and regulatory consequences (privacy law, payment-card rules), safety and reputation.

A common approach is a risk matrix: rate each factor on a small scale and combine them.

Impact: Low Impact: Medium Impact: High
Likelihood: High Medium High Critical
Likelihood: Medium Low Medium High
Likelihood: Low Low Low Medium

CVSS (see the first lesson of this chapter) gives a standard score for the flaw itself. Many firms report both: the CVSS base score for comparability and a contextual rating for the client's environment. You can raise a rating when the vulnerable system holds crown-jewel data or is directly internet-facing, and lower it when strong compensating controls exist (but explain why).

Rating a finding

Ease How hard to find and use?
Exposure Internet, internal, authenticated?
Impact Data, systems, money, law
Controls Would anything stop or detect it?

Write down the reasons, not just the label.

1 / 2
Python A small risk matrix calculator

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

The adjustment list is the important idea: every change from the base rating carries a written reason, so a reader can challenge it, and two testers rating the same finding should reach nearly the same result.

Rating pitfalls

  • Inflation: calling everything critical teaches readers to ignore you. Reserve the top rating for issues that can plausibly cause severe harm soon.
  • Scanner copy-paste: a scanner's severity is a starting guess. Validate and re-rate.
  • Ignoring chains: two medium findings that combine into a path to domain admin may deserve a higher combined rating. Report the attack path as its own finding.
  • Forgetting the positives: note defenses that worked (detection that fired, a blocked exploit). They matter to the people who built them.

Evidence: what to collect

For each finding, keep enough to let someone reproduce and verify it, and no more:

  1. What: the affected asset (URL, host, IP, account, resource ID).
  2. How: exact steps or the request and response, in order.
  3. Proof: a timestamped screenshot or text capture showing the result.
  4. When and from where: date, time zone and your source IP, so the client can match it to their logs.
  5. Tool and version if it matters.

Prefer plain-text captures (copyable, searchable) alongside screenshots. Include enough context in each screenshot, such as the host name and prompt, to show where it came from.

Evidence: how to protect it

Evidence often contains sensitive material: tokens, personal data, internal addresses. Treat it as confidential:

  • Minimize: capture a single row, not the table; the first few characters of a token, not all of it.
  • Redact secrets and personal data in anything that goes in the report. Mask, do not just draw a box over text in a way that can be undone (flatten the image or use a black fill in the original capture).
  • Hash important files with SHA-256 so integrity can be shown.
  • Encrypt the evidence folder at rest, store it only on approved systems, share it through the agreed secure channel, and delete it by the date in the contract.
  • Never put raw credentials, private keys or full personal records in an email or ticket.
Python Build an evidence manifest

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

The manifest ties each capture to a hash and prepares a redacted excerpt for the report, while the original stays in the encrypted evidence store. If the client disputes a finding months later, you can show the file has not changed.

Writing the finding summary

Keep each finding to the same shape: title, rating, affected assets, description, evidence, impact, recommendation, references. The next lesson turns this into a full report.

Check your understanding

Quick check ยท 4 questions

of 4 answered You scored / 4 ยท perfect!

  1. 1. A risk rating combines which two ideas?

    Risk is how likely a problem is to be exploited and how bad the result would be.
  2. 2. Why should rating changes from a scanner's severity include written reasons?

    A documented reason makes the rating defensible and repeatable.
  3. 3. Which is the best way to include a captured API token in a report?

    Reports travel widely; secrets should be minimized and masked.
  4. 4. Two medium findings combine into a path to full domain control. What should you do?

    Chained issues can be more serious than each link alone.

Summary

  • Rate findings by likelihood and impact; use CVSS for the flaw and context for the client's real risk.
  • Write the reasons for every adjustment, avoid inflation, and report attack chains.
  • Evidence must be enough to reproduce a finding and no more: what, how, proof, when and from where.
  • Protect evidence: minimize, redact, hash, encrypt, share securely and delete on schedule.

Next: Writing the penetration test report