Duty analyst on shift

UK Cyber Defence

Penetration testing · Web applications · OWASP

The application,tested by hand.

Manual testing of a web application by someone signed in as each kind of user it has. It finds the flaws in access control and business logic that automated scanners cannot reason about, and shows your developers how to reproduce and fix them.

CREST-certified testers£950 + VAT a dayReport in five working daysRetest included

01What it answers

Scanners find missing patches. People find the invoice that belongs to another customer.

The most serious web application flaws are rarely exotic. They are a record that can be read by changing a number in the address, an approval step that can be skipped, a role that can do one thing more than it should. Finding them takes a tester who understands what the application is for, working through it as each role in turn. We test to the OWASP Testing Guide, treat the OWASP Top 10 as a floor, not a ceiling, and spend most of the time on authorisation and workflow, because that is where the findings are.

Scope
One application, every user role, and the calls it makes to its own back end
Typical duration
Two to seven days of testing, plus one for reporting
Typical price
£2,850 to £7,600 + VAT
You provide
A test environment where you have one, and accounts for each user role
Methods
OWASP Testing Guide · OWASP Top 10 · MITRE ATT&CK mapping

Coverage

What a web application test looks at

01

Authentication and sessions

Sign-in, password reset, multi-factor authentication, session lifetime and what survives signing out.

02

Access control

Whether one user can read or change another's data, and whether a lower role can reach a higher role's functions.

03

Injection and input handling

SQL injection, cross-site scripting, template and command injection, file upload and deserialisation.

04

Business logic

Steps that can be skipped, repeated or reordered, and prices, quantities and approvals that can be changed on the way to the server.

05

Multi-tenancy

In software sold to many customers, the walls between one customer's data and the next.

06

Configuration

Security headers, cookies, TLS, error handling and the administrative paths left over from development.

Findings

What these tests usually find

  1. _01Broken access controlA record, document or function available to a user who has no right to it, usually by changing an identifier.Most common
  2. _02Stored cross-site scriptingInput from one user that runs in the browser of another, often an administrator.Injection
  3. _03A password reset that can be abusedTokens that are predictable, long-lived or not tied to the account that asked for them.Authentication
  4. _04Leftovers from developmentVerbose errors, debug endpoints and old versions still answering.Configuration

Cost

What it costs

Penetration testing is charged at one published rate, £950 + VAT a day, with one day added to every engagement for reporting, quality assurance and the debrief. The days follow the scope.

ScopeDaysPrice
Small: brochure site or single-role application, up to about twenty pages2 testing + 1 reporting£2,850 + VAT
Standard: authenticated, two or three roles, forms and workflows4 testing + 1 reporting£4,750 + VAT
Complex: multi-tenant, many roles, payments or integrations7 testing + 1 reporting£7,600 + VAT

These figures come from the estimator on the penetration testing page, where you can change the numbers and combine this test with others; one reporting day covers the whole engagement. A scoping call confirms the days and turns the estimate into a fixed price.

Questions

Questions about this test

Should you test our live site or a test environment?

A test or staging environment that matches production is best, because we can try things that create or change data. Where there is only production, we agree limits and timing first.

Do you test the API behind the application?

The calls the application itself makes are tested as part of it. An API that is also offered to customers or partners in its own right is better scoped as a separate API test.

Will our developers be able to act on the report?

Yes. Every finding has reproduction steps, evidence and a recommended fix, and the debrief is a working session with your engineers. When the fixes are in, we retest them at no further cost.

We already run a scanner in our build pipeline. Is a manual test still needed?

The two do different jobs. A pipeline scanner catches known patterns on every build, which is valuable. It cannot tell that an approver should not be able to approve their own request. A manual test once a year, or before a major release, covers what the scanner cannot.

Start a conversation

Show us what the application is for.

A thirty-minute scoping call covers the roles, the workflows that matter and a fixed price.