Penetration testing cost in Australia: what sets the price and how to compare quotes
What a penetration test costs depends on its scope.
This guide shows what two Australian providers publish, as checked on 13 November 2026, explains each factor that moves the effort, and gives you a worksheet to scope your own test and line quotes up side by side.
In this guide
Australian penetration test budget worksheet
Copy this worksheet into a document and fill in Part A for your own application. Mark each factor light, moderate or heavy with Part B, then use Part D to line up the quotes you receive.
How to use itHide the instructions
Part C shows three invented example scopes filled in, and Part E lists the published market figures with the date they were checked.
penetration-testing-cost.txt
Introduction2 lines
Part A. Count your scope16 lines
Part B. Read the effort, factor by factor14 lines
Part C. Three example scopes30 lines
Part D. Compare the quotes11 lines
Part E. Published Australian market figures, checked 13 November 20264 lines
Whole template87 lines
AUSTRALIAN PENETRATION TEST BUDGET WORKSHEET
Fill in one copy for each application you want tested. Bring it to any provider's scoping call, and use Part D when the quotes arrive.
PART A. COUNT YOUR SCOPE
A1. Application name and what it does: ____
A2. Who asked for the test, and their exact wording (a customer, insurer, auditor, tender, or your own decision): ____
A3. Date the report must be ready by: ____
A4. Domains, subdomains and API addresses in scope: ____
A5. Anything that must be left alone, such as payment provider pages, third-party widgets or the hosting control panel: ____
A6. Routes, meaning pages and API endpoints that accept input or return data that is not public. Total: ____
Of these: file uploads ____ payments or refunds ____ password reset or invitations ____ exports ____ actions that send email or SMS ____
A7. Roles that can sign in, such as customer, staff member, administrator. List them: ____ Number of roles: ____
A8. Customer organisations sharing one database (write 1 if the app serves a single organisation): ____
A9. Who calls the API: only your own front end / mobile apps as well / partners with their own keys or tokens. Current OpenAPI or GraphQL schema available: yes / no
A10. Testing depth wanted: outside only / signed in as every role listed in A7
A11. Source code access you can offer: none / read-only for named areas / the full repository
A12. Environment to test: a staging copy with seeded data / production only / both
A13. Production conditions: testing windows ____ records that must not change ____ actions that email or charge real people ____ firewall or rate limits to lift for the tester ____
A14. Who will read the report: your developers / a customer / an insurer / an auditor / a tender panel. Attestation letter or framework mapping requested: ____
A15. Retest of the fixes needed before the date in A3: yes / no
PART B. READ THE EFFORT, FACTOR BY FACTOR
Mark each factor Light, Moderate or Heavy. The marks compare testing effort between factors. They are not prices.
B1. Routes (A6). Light: a few public forms and no signed-in area. Moderate: tens of routes behind a login, some with uploads or payments. Heavy: around a hundred endpoints or more, or several front ends sharing one API.
B2. Roles (A7). Light: visitors plus one administrator account type. Moderate: two or three roles, each tried against the others. Heavy: four or more roles, or permissions that differ from one customer to the next.
B3. Customer organisations (A8). Light: one organisation. Heavy: two or more, because every access check is repeated across organisations.
B4. APIs (A9). Light: only your front end calls the API and a current schema exists. Moderate: no schema, so endpoints have to be discovered. Heavy: mobile or partner clients with their own keys or tokens, tested as a separate surface.
B5. Testing depth (A10). Light: outside only, which cannot show what one signed-in user can do to another. Heavy: signed in as every role, through the screens and against the API directly.
B6. Source code access (A11). Moves the effort rather than only adding to it: suspected flaws are confirmed faster, and reading the code takes time of its own. A full code review is a separate scope.
B7. Production conditions (A12, A13). Light: a staging copy with seeded data. Moderate: production with agreed windows and labelled test records. Heavy: production only, with records that must not change, real emails or charges to avoid, and blocking that cannot be lifted.
B8. Reporting and retest (A14, A15). Light: a report for your own developers and a retest of the reported findings. Moderate: a report a customer or insurer will read, with a scope statement and evidence they can follow. Heavy: an attestation letter or framework mapping, a retest before a fixed date, or both.
Reading the marks:
Mostly Light: an external check or audit may answer the question, and a fixed-scope offer can fit, provided the person in A2 will accept it.
Any Heavy on roles, organisations or APIs: expect a scoping call before any quote, and read every fixed-scope limit against A6 to A9.
Several Heavy marks: ask each provider how the work could be staged, starting with the routes that would do the most harm if they failed.
PART C. THREE EXAMPLE SCOPES
These are invented examples that show the worksheet in use. They do not describe real clients or real quotes.
Example 1: a local business website with online booking
A6: six routes (contact form, quote request, newsletter sign-up, the booking page, admin login, password reset). No uploads, and no payments on the site itself.
A7: site administrators only. A8: one organisation.
A9: no API of its own. The booking widget belongs to a third party and is listed in A5.
A10: outside, plus the admin login and reset flows. A11: none.
A12 and A13: production only, with test submissions labelled and deleted afterwards.
A14: the owner and the web developer. A15: yes.
Marks: Light on every factor. A website security check or audit is likely the right depth rather than a full penetration test.
Example 2: a customer portal for a trades business
A6: about 40 routes, including the API the portal's own front end calls. Job photo uploads, invoice downloads, and card payments through a hosted checkout that reports back to the portal by webhook.
A7: customer, field staff and office administrator, so three roles. A8: one organisation.
A9: own front end only, with a current OpenAPI schema.
A10: signed in as all three roles. A11: read-only access to the authorisation middleware and the payment webhook handler.
A12: a staging copy with seeded customers.
A14: a commercial customer's security questionnaire asks for a recent test and evidence that serious findings were fixed. A15: yes, before the questionnaire is due.
Marks: Moderate on routes, roles and reporting. Light on APIs and environment. Each role is tried against the other two, and the uploads, downloads and webhook each need their own checks. Code access shortens the work of confirming how each check is enforced.
Example 3: a multi-tenant software product
A6: about 150 API endpoints behind a single-page app, plus background jobs that send email.
A7: owner, administrator, member and read-only roles in each customer organisation, plus an internal support role.
A8: many organisations in production. Two test organisations requested.
A9: own front end plus a partner API with its own keys. Schema exists for the partner API only.
A10: signed in as every role in both test organisations, and with partner keys.
A11: read-only access offered for the tenancy and permission layer.
A12 and A13: staging for the application, but two integrations exist only in production. Emails to real users must be suppressed.
A14: larger customers send security questionnaires and ask for an attestation letter. A15: yes.
Marks: Heavy on routes, roles, organisations, APIs and reporting. Every role check runs in both organisations and again with partner keys, and the production-only integrations need agreed windows. Expect a longer scoping call, a staged plan, or a referral to a firm that specialises in large platforms.
PART D. COMPARE THE QUOTES
Write each quote's answer beside the line.
D1. Targets, environment, routes and roles named in writing: Quote 1 ____ Quote 2 ____ Quote 3 ____
D2. Tester-days or effort stated: Quote 1 ____ Quote 2 ____ Quote 3 ____
D3. A tester working through the application while signed in, or mainly an unattended automated scan: Quote 1 ____ Quote 2 ____ Quote 3 ____
D4. API tested directly, and code access used for what: Quote 1 ____ Quote 2 ____ Quote 3 ____
D5. Exclusions listed: Quote 1 ____ Quote 2 ____ Quote 3 ____
D6. Assumptions the total depends on, and what happens if the scope grows: Quote 1 ____ Quote 2 ____ Quote 3 ____
D7. Report contents, and the reader it is written for: Quote 1 ____ Quote 2 ____ Quote 3 ____
D8. Retest included or extra, which findings it covers, and the window to request it: Quote 1 ____ Quote 2 ____ Quote 3 ____
D9. Total, and whether it includes GST: Quote 1 ____ Quote 2 ____ Quote 3 ____
D10. Matches the exact wording in A2: Quote 1 yes / no Quote 2 yes / no Quote 3 yes / no
PART E. PUBLISHED AUSTRALIAN MARKET FIGURES, CHECKED 13 NOVEMBER 2026
Each figure is the provider's own, for its own scope, as published. They do not combine into an average. Check the page again before relying on any of them.
E1. Siege Cyber, Penetration Testing Cost in Australia (siegecyber.com.au/pentest-pricing). Prices stated as current as of 18 October 2026, in Australian dollars and excluding GST, each including manual testing, the report, a debrief call and the retest. Basic website: $3,400 to $6,500. Web application and API: $5,900 to $16,500.
E2. Siege Cyber, Web Application Penetration Test: Essentials (siegecyber.com.au/pentest-essentials). One web application, one environment and up to three authenticated roles: $4,400 excluding GST, fixed. Retest: an optional extra at $975 excluding GST. No debrief call, standalone API testing, infrastructure testing or code review.
E3. Compliance Council, What Does a Penetration Test Cost in Australia? (compliancecouncil.com.au/penetration-testing-cost-australia). Indicative bands, ex GST. Small external footprint, up to 20 external hosts, one web application and one user role: $8k to $14k, with 4 to 6 tester-days. Application with real privilege, one or two applications tested as an anonymous visitor, a standard user and an administrator with authenticated API testing: $16k to $30k, with 8 to 14 tester-days. A retest of Critical and High findings within 90 days is included.How penetration tests are priced in Australia
There is no standard price list for penetration testing in Australia, and a figure quoted without a scope tells you very little.
What a provider sells is testing effort: a tester working through your application for a set amount of time, then writing up what was found. Two Australian providers that publish prices say this plainly. Siege Cyber's pricing page says penetration testing is priced on tester days, and Compliance Council's cost guide says most Australian tests are priced in tester-days underneath.
So the useful question is less what a penetration test costs and more how much testing your application needs. That depends on things you can count before you speak to anyone: the routes that accept input or return private data, the roles and customer organisations that must be kept apart, the APIs, how far behind the login the testing goes, whether the tester can read the code, what production allows, and who will read the report.
The sections below take each in turn, and the worksheet turns your answers into a scope you can hand to any provider.
Clever Ops does not publish prices. We quote each test after a free scoping call, once those factors are known, and the written scope that comes with the quote states what is tested, what is excluded and what the retest covers.
What Australian providers publish, checked on 13 November 2026
The figures below come from two Australian providers' own pages, opened and read in full on 13 November 2026.
Each is the provider's published figure for its own scope, excluding GST, with ranges written using "to". Prices change, so open the provider's page again before relying on any of them.
Siege Cyber states that the prices on its pricing page are current as of 18 October 2026, in Australian dollars and excluding GST, and that every price includes manual testing, the report, a debrief call and the retest.
Its basic website test, for a brochure or marketing site or a WordPress site, is listed at $3,400 to $6,500. Its web application and API test, for SaaS platforms, customer portals and anything behind a login, is listed at $5,900 to $16,500.
For a web application, the page names two things that move a price through that range: the number of user roles that need testing and the number of distinct user journeys.
Siege Cyber also sells a fixed-scope test it calls Essentials, priced at $4,400 excluding GST. It covers one web application in one environment, up to three authenticated roles, the OWASP Top 10:2025 categories and the APIs the application itself calls, and it includes a one-page attestation letter.
A debrief call, standalone APIs, infrastructure, source code review, mobile apps and social engineering are excluded, and a retest is an optional extra at $975 excluding GST.
The page lists cases where its full test is the better fit, including a multi-tenant platform, more than three roles, authorisation rules that differ per customer, and a standalone API a customer has named. A fixed price holds because the scope is capped, so read the exclusions and the role limit against your own application before you set it beside an open-scope quote.
Compliance Council's 2026 Australian pricing guide publishes indicative bands, ex GST, for comparable scopes, with the tester-days behind each. A small external footprint of up to 20 external hosts, one web application and one user role is shown at $8k to $14k, with 4 to 6 tester-days.
An application with real privilege, meaning one or two applications tested as an anonymous visitor, a standard user and an administrator with authenticated API testing included, is shown at $16k to $30k, with 8 to 14 tester-days.
The guide describes the three roles in that second band as each a separate pass, which shows plainly why roles move effort. It also says no one can quote your environment from a web page, and that a retest of Critical and High findings within 90 days is included.
Each figure belongs to its provider's own definition of a scope, so the figures do not combine into a market average or a typical price. Three differences between them matter more than the numbers.
- What the price includes. Siege Cyber's ranges include the debrief call and the retest. Its fixed-scope test leaves out the debrief call and sells the retest as an extra. Compliance Council says a retest of Critical and High findings is included.
- What the scope counts. Siege Cyber frames web application scope in user roles and user journeys, and Compliance Council in hosts, applications and roles. Translate both into your own routes and roles before you compare.
- Whether effort is stated. Compliance Council publishes tester-days beside each band, and Siege Cyber suggests asking every provider how many tester days are in the price. A quote that states its days can be compared with another; one that does not leaves you comparing totals.
Seven factors that change the testing effort
Each factor below either adds things to test or adds time to each test.
None of them carries a fixed price, and providers weigh them differently, but every one of them shows up in how a scope is written. Count them for your own application before you ask anyone for a quote.
A route is any page or API endpoint that accepts input or returns data that is not public: a login form, a search box, an invoice download, the endpoint the browser calls to save a booking. Testers work through each one, so the route count sets the floor for the effort.
It matters far more than page count. A marketing site can have hundreds of pages and only a few routes worth testing, while a small product can hide a hundred endpoints behind six screens.
Some routes weigh more than others. File uploads, payment and refund steps, password reset, invitations, exports and anything that sends email or SMS each need checks of their own. Ask your developer for the route list from the routing files or the API documentation. A provider who scopes from that list is estimating your application rather than guessing at it.
Access control is tested between people, so effort grows with the number of combinations, not only the number of roles. With a customer, a staff member and an administrator, each role is tried against the routes and records of the other two.
If the application serves several customer organisations from one database, the checks are repeated between organisations too, because one customer reaching another customer's records is among the most serious findings an application can have.
OWASP's Web Security Testing Guide says the best way to test direct object references is with at least two users, and often more, who own different objects, plus users with different privileges where that is relevant. It also has the tester first map every place where user input points at a record or file.
That is why the worksheet asks for two accounts in each role, and why a test limited to one or two roles needs less effort than one that covers every role your application has.
Most modern web applications are a front end over an API, and the API is where the rules have to hold. Testing it directly, rather than only through the screens, finds the requests the interface never sends: another user's ID in a path, a field the form hides, a method no button calls.
Broken object level authorisation, the first entry in OWASP's 2023 API security list, is exactly this: changing the ID of an object in a request to reach another user's data.
Effort depends on how well the API is described and who calls it. A current OpenAPI or GraphQL schema lets a tester cover endpoints in order instead of discovering them one at a time. An API that mobile apps or partners call directly, with their own keys or tokens, is a separate surface with its own identities, and the fixed-scope offer above leaves standalone APIs to its full test.
An unauthenticated test looks at what anyone on the internet can reach: exposed files, the login and reset flows, response headers, visible software versions and the public forms. It takes less time because there is less to reach, and it cannot tell you what a signed-in user can do to someone else's data.
An authenticated test signs in as each role and works through the application from inside, which is where access control, business rules and most of the API sit.
Siege Cyber's table of price drivers draws the same contrast: testing as an anonymous visitor only keeps the price lower, while testing as an admin, a standard user and a customer, then checking whether one can reach another's data, pushes it higher. If your customers log in, an outside-only test answers a different question from the one you are asking.
A test without code access works from the outside, which is how an attacker starts. Read-only access to the repository changes how the time is spent rather than simply adding to it. The tester can read the authorisation layer instead of inferring it, confirm a suspected flaw quickly and find routes that no screen links to.
Reading code takes time as well, and a full secure code review is a separate piece of work with its own scope.
When a quote mentions code access, ask what it will be used for: targeted reading of the riskiest areas during the test, or a line-by-line review. The two differ a great deal in effort and in what the report can say about the code.
A staging copy with seeded data is the lighter path, because the tester can create, change and delete records freely. Production brings conditions that take time: agreed testing windows, rules about which records may be written, care around anything that emails or charges real customers, payment providers and other third parties that must be left alone, and a list of test data to clean up afterwards.
A web application firewall or rate limit that blocks the tester partway through adds time too, unless the tester's traffic is allowed through for the window.
None of this is a reason to avoid testing production. It is a reason to say up front which environment is in scope and what it permits, so the quote reflects the conditions you actually have.
The report is the part you keep, and its audience changes the work. A report for your own developers can be lean. One that goes to a customer, an insurer or an auditor needs a scope statement, evidence they can follow and severity ratings they can read, and some buyers ask for an attestation letter or a mapping to a framework they work under. Each of these adds writing and review time.
Retest terms differ widely between offers. Among the published offers above, one includes a retest in every price, one sells it as an optional extra, and one covers Critical and High findings within a set window. Check what the retest covers, how long you have to request it, and whether findings that were not retested are marked that way rather than as passed.
How to read a penetration testing quote
Two quotes for the same application can differ widely without either provider being unreasonable, because they describe different work.
Before comparing totals, put each quote through the same questions. Part D of the worksheet sets them out as a grid.
- Scope named in full: the domains and API addresses, the environment, the routes or features, and the roles and customer organisations to be tested. If any of these is missing, the total rests on an assumption you have not seen.
- Effort stated: the tester-days, or a description of the work clear enough to judge it by. A low total with no stated effort tells you little about what you are buying.
- Method: whether a tester works through the application while signed in and checks each finding, or whether the work is mainly an unattended automated scan. Both have a place, and they are different purchases.
- Depth: authenticated or outside only, the API tested directly or only through the screens, and code access used or not.
- Exclusions, written down. A quote that lists none leaves you to discover them when the report arrives.
- Assumptions: what the total depends on, and what happens if the application turns out larger than described. Settle that before testing starts.
- Report: who it is written for, what each finding contains, and whether it will satisfy the customer or auditor who asked for the test.
- Retest: included or extra, which findings it covers, and the window for requesting it.
- GST: whether the total includes it. Both providers above publish their figures excluding GST.
Ways to reduce the effort without narrowing what matters
Part of any test goes on work that is not testing: waiting for accounts, rediscovering routes, working around a build that changes halfway through.
Taking that time out lowers the effort a provider has to allow for, and none of it reduces what gets tested.
- Send the route list and any API schema with your scoping request, so discovery starts from your map rather than from nothing.
- Have test accounts ready for every role, in two customer organisations if your application has more than one, with sample records in each.
- Offer a staging copy that matches production in configuration, with seeded data, so the tester can write freely.
- Clear what you already know about. Fix the findings from your platform's own security checks and any recent scan, so the test spends its time on what a scan cannot find.
- Freeze the build for the testing window, or tell the tester which areas are changing and when.
- Name a technical contact who can answer questions and reset data while testing runs.
- If the budget is tight, rank routes by the harm a failure would do and test the top of the list first: sign-in and password reset, payments, anything that shows one customer's data, and admin functions. Schedule the rest rather than dropping it.
When the lower quote is the wrong one
A lower quote is the right one when it covers what you need.
It is the wrong one when the scope leaves out the thing that prompted the test, or when the report will not satisfy the person who asked for it. Watch for these mismatches.
- Your customers log in, but the quote covers only what is visible without signing in.
- Your application serves several customer organisations, but the quote tests one role in one organisation.
- A customer, insurer or tender asked for a penetration test, and the quote describes an automated scan.
- The request requires an accreditation the provider does not hold. If a tender requires CREST, Clever Ops is not CREST accredited, so we refer that work to a provider that is.
- The request asks for an independent tester, and the quote comes from the people who built or host the system. When Clever Ops tests a site it built or hosts, the work is a security review rather than an independent test, and the report says who built and who hosts it.
- The person who asked for the test wants evidence that serious findings were fixed, and the quote has no retest.
How Clever Ops scopes and quotes a test
Clever Ops tests what customers reach through a browser or an API: websites, WordPress sites and WooCommerce stores, custom web applications and their APIs, apps made with AI builders, and AI chatbots and agents.
We do not publish prices or packages, because the factors on this page decide the work, and a quote follows a free scoping call.
Bring the worksheet on this page to that call. It covers the questions the call needs answered, so the conversation can go straight to the parts of your application that carry the risk. You then receive a written scope and quote naming the targets, environment, routes, roles and exclusions, and what the one included retest covers. Testing begins once the owner of each target has signed an authorisation.
Some work belongs with another kind of provider, and saying so early saves you a scoping round: internal networks and staff devices, phishing, load or denial-of-service testing, and incident response or malware cleanup. Very logic-heavy platforms, such as marketplaces or lending products, are scoped individually or referred to a larger firm.
Frequently asked questions
It depends on the scope, and both Australian providers quoted on this page tie their published figures to it. On 13 November 2026, Siege Cyber listed its web application and API test at $5,900 to $16,500 excluding GST, and Compliance Council's indicative band for one or two applications with three roles and authenticated API testing was $16k to $30k ex GST. Where an application lands depends on its routes, roles, APIs, testing depth, code access, production constraints, reporting and retest. Clever Ops quotes after a free scoping call and does not publish prices.
Mostly because the quotes describe different work. The number of tester-days, whether a tester works through the application or a scan does most of the work, whether the test signs in as every role, and whether a retest is included can each move a quote a long way. Ask every provider to state those four things, then compare scopes before you compare totals.
It depends on the provider and the offer. Of the published Australian offers on this page, one includes a retest in every price, one sells it as an optional extra at $975 excluding GST, and one includes a retest of Critical and High findings within 90 days. Every Clever Ops test includes one consolidated retest of the reported findings, requested within the window set in your statement of work.
Not from Clever Ops. A figure given before the routes, roles and constraints are known would be a guess, and the gap would surface later as a change of scope or a shallower test. The scoping call is free, and the worksheet on this page covers what it asks, so the call can be short.
It is a different purchase. A scan compares what it can reach against known issues and is useful between tests. It does not sign in as each role and try to break the rules your application is meant to enforce, which is where access control and business logic flaws sit. If someone else asked for a penetration test, check that what you buy matches their wording before you sign.
It can, when your application fits inside its limits. Read the role cap, the API limits and the exclusions against what you run. A multi-tenant application, more than three roles or a separately named API can fall outside a fixed scope, and the published fixed-scope offer on this page lists those cases itself.
Scope the right test for your application
Tell us what you have built, who uses it and when it needs to launch. You will get a plain-English recommendation: a website audit, a web application test or a referral to a specialist.