- Market
- Saudi Arabia
- Written for
- CISO, CIO, Procurement
- Decision
- Budget for a penetration test and compare supplier quotations
- Languages
- EN · AR
Cybersecurity credentials
NCA registration
Registered cybersecurity service provider

ISO/IEC 27001
ISO/IEC 27001 certified
The unit
Everything is priced in tester-days
We do not have an office in Saudi Arabia
We serve organisations there from our Riyadh office and our engineering hubs in Lahore and Karachi, travelling for the parts of an engagement that need to happen in person.
A penetration test is a person with a defined skill set spending a defined number of days trying to break a defined thing, and then writing it up. Every quotation you receive is a day count multiplied by a rate, whether or not the document shows the arithmetic. So the two questions that make a price legible are: how many days, and at what seniority. Everything else on this page is an input to the first one.
This matters commercially because it is where quotations diverge without appearing to. Two suppliers quoting the same figure for "a web application penetration test" may be proposing eight days and three days. The three-day version is an automated scan with a manual sanity check and a report; the eight-day version includes authenticated testing across roles and time to chase business-logic flaws that no scanner will find. Both are legitimate products. Only one of them will find an authorisation bug.
We do not publish a day rate on this page. Rates in the Saudi market vary by supplier scale, tester seniority and whether the work is delivered locally or remotely, and a single figure from one firm would misrepresent the range. Ask each supplier for the day count and the seniority mix explicitly — the rate is then arithmetic you can do yourself, and the comparison becomes valid.
Scope to days
What drives the day count in each type of test
The unit of scope differs by test type, which is why "how many days" cannot be answered without knowing which kind of test is being bought. These are the units that matter and what typically stretches them.
| Option | Unit of scope | What stretches it most | Commonly under-scoped | Best for |
|---|---|---|---|---|
| Web application | Distinct functional areas plus the number of user roles, not page count. | Role count. Testing authorisation properly means exercising each function as each role and as no role — the work grows with roles, not screens. | Admin and back-office interfaces, which are frequently excluded and are where the highest-impact findings usually are. | Anything with a login and more than one type of user. |
| API | Endpoint count weighted by how many take an object identifier, plus role count. | Object-level authorisation. Every endpoint that accepts an ID has to be tested against every role, which is a multiplication rather than an addition. | Undocumented endpoints. Scoping from a specification misses what the specification does not know about. | Anything with a mobile client or a partner integration. |
| Mobile application | Platforms in scope, plus the backing API — which is a separate piece of work. | Whether both iOS and Android are in scope, and whether client-side controls need to be defeated to reach the API. | The API behind the app. A mobile test that stops at the binary has tested the least valuable half. | Consumer-facing apps and anything handling payment or identity. |
| Infrastructure | Live host count, and whether internal, external or both. | Internal testing, which involves lateral movement and privilege escalation and takes considerably longer per host than an external sweep. | Scoping from an asset inventory rather than from discovery. The hosts nobody remembers are the ones worth testing. | Estates with on-premise systems or a large public footprint. |
| Cloud configuration | Accounts or subscriptions in scope, and the number of distinct workload patterns. | Identity. Enumerating who can assume what, and where a role chain grants more than intended, is the bulk of the work. | Treating it as equivalent to a posture scan. A tool finds misconfigurations; a tester finds paths between them. | Estates where most of the perimeter is now identity rather than network. |
What moves the number
Nine factors, and the direction each pushes
Listed with the effect stated, so you can work out roughly where your own scope sits before anyone quotes you.
| Variable | Effect on price | Why |
|---|---|---|
| Number of user roles | Multiplies the authorisation work | The single most underestimated driver. Testing four roles is not twice the work of two — every privileged function has to be attempted from every lower role, so the matrix grows faster than the role count. |
| Authenticated versus unauthenticated | Substantially increases days | An unauthenticated test looks at what an anonymous attacker sees. Most serious findings in business applications are behind the login, so an unauthenticated-only test is cheaper because it is smaller, not because it is efficient. |
| Business logic complexity | Increases days, hard to estimate | A multi-step approval workflow, a pricing engine or anything involving money movement needs a tester to understand the intended rules before they can be broken. This is manual time no tool substitutes for, and it is where the findings that matter tend to be. |
| Retesting | Adds a fixed block | Verifying fixes is a separate piece of work. Check whether it is included, how long the window is, and whether it covers only the original findings or a regression pass — an unlimited retest window with a two-week limit on it is a common source of surprise. |
| Reporting depth and audience | Adds days at the end | An engineering report with reproduction steps is different work from an audit-facing report with risk ratings and an executive summary. Needing both is reasonable and should be priced, not assumed. |
| Environment stability | Can add days unexpectedly | Testing against a staging environment that is being deployed to, or one that does not match production, wastes tester time. Agree an environment freeze and confirm parity before the window opens. |
| Urgency | Raises the rate, not the days | A test that has to start next week is priced against tester availability. Booking three to four weeks out is normally the difference between a standard and an expedited rate, and there is no quality argument for the rush. |
| On-site requirement | Adds cost, sometimes required | Internal infrastructure testing and some regulated environments require the tester on the premises. Where remote access is acceptable it is cheaper; where it is not, that is a constraint to state at scoping rather than discover at kickoff. |
| Compliance driver | Changes the deliverable | A test run to satisfy a specific framework may have required scope, tester independence or evidence-format conditions attached. Establish what the assessor will accept before scoping — a technically excellent test in the wrong format gets rejected. |
Ask any supplier quoting this
- How many tester-days is this, and what is the seniority mix across them?
- Is the testing authenticated, and across how many distinct roles?
- What proportion of the work is manual versus automated tooling?
- Is retesting included, for how long, and does it cover regressions or only the original findings?
- Will the report include reproduction steps an engineer can follow without contacting you?
- Who specifically will perform the test, and can we see a redacted sample of their work?
- What happens if a critical finding appears on day two — is there a defined disclosure path?
Reading a quotation
Where a cheap quotation is genuinely cheaper, and where it is smaller
Some price differences are real efficiency. A supplier with strong tooling and reusable methodology covers the same scope in fewer days than one starting from a blank page, and a firm with lower overheads charges less for the same seniority. Neither of those is a compromise, and a buyer who assumes the cheapest quote is always the weakest is leaving money on the table.
The differences that are not efficiency show up in three places. Day count against the same scope: if one quotation is materially shorter, ask what has been dropped rather than assuming speed. The manual-to-automated ratio: an automated scan with a report attached is a legitimate product and a much cheaper one, but it will not find authorisation flaws or business-logic abuse, so buying it while believing you bought a penetration test is the actual risk. And seniority: a quotation that will not name who is doing the work is quoting availability, not people.
One further check worth making: ask for a redacted sample finding. A good finding tells you what the issue is, how to reproduce it, what an attacker gains and how to fix it — specifically, not by reference to a generic remediation library. Report quality is the part of the deliverable you actually consume, and it varies more between suppliers than the testing does.
Common questions
What buyers ask
Why will you not just publish a price?+
Because we have no approved price list, and a figure invented for a web page would be worse than none — you would anchor a budget to it and then receive a quotation that did not match. The model above is more useful in any case: it lets you estimate your own scope in days, which is the number that actually determines cost, and it lets you tell whether two quotations are for the same work. Send us the scope and you get a fixed quotation rather than a range.
Is a vulnerability scan the same thing, and much cheaper?+
It is much cheaper and it is not the same thing. A scanner finds known vulnerabilities in known software — missing patches, weak configuration, recognisable injection points. It cannot tell that a user with a viewer role can reach an admin function by changing an ID, or that a discount can be applied twice, because those require understanding what the application is supposed to do. Both have a place: scanning is a continuous control, testing is a periodic examination. Buying one and reporting the other is the mistake.
How often should we budget for testing?+
Annually is the common baseline for a stable application, and it is a weak control on its own for anything changing frequently. The more useful trigger is change: test after a significant architecture change, a new authentication mechanism, a new integration exposing data, or a major release. An application shipping every fortnight has been substantially rewritten between annual tests, so a test-on-change policy plus continuous scanning usually gives better coverage than a larger annual engagement.
Does a penetration test make us compliant?+
No. A test produces evidence that a control was examined; whether that evidence satisfies a given framework is decided by the body that owns the framework and, in practice, by the assessor reviewing your submission. Some frameworks specify scope, tester independence or evidence format. Establish what your assessor will accept before scoping the test — reworking a completed engagement to fit a requirement nobody checked is the most avoidable cost in this category.
Can the test be done remotely, or does someone need to be in Riyadh?+
External web, API and cloud testing is normally performed remotely with no loss of quality. Internal infrastructure testing generally needs either an on-site tester or a controlled device placed on the network, and some regulated environments require physical presence as a condition. Establish which applies to you before comparing quotations, because a remote quote and an on-site quote are not comparable figures.
In this cluster
Get a scoped estimate
A scoped estimate: tell us the application count, the role count and whether an authenticated or infrastructure component is in scope, and you get a day estimate and a fixed quotation rather than a range.
Start the conversation