Cybersecurity credentials
NCA registration
Registered cybersecurity service provider

ISO/IEC 27001
ISO/IEC 27001 certified
Side by side
The same output, produced by different work
Both end in a document listing security defects. What differs is how the defects were found, and therefore which kinds of defect can appear in the document at all.
| Option | How defects are found | What it proves | Cadence | Typical cost shape | Where it fails | Best for |
|---|---|---|---|---|---|---|
| Vulnerability assessment | Software compares what is running against a database of known issues, across everything it can see. | That you know your exposure to publicly known issues, across the whole estate, as of today. | Continuous. Weekly at worst, and in the pipeline for anything built in-house. | A licence and the staff time to act on it. Cost scales with asset count. | It cannot judge whether a finding matters in your architecture, and it produces more findings than any team reads. | Knowing what is on the estate and what is unpatched — the baseline everything else assumes. |
| Penetration test | A person attempts to reach something they should not, using whatever combination of weaknesses works. | What an attacker could actually do to a specific system, demonstrated rather than inferred. | Periodic. Annually, or on significant change, against a defined scope. | A day rate against a scope. Cost scales with roles, integrations and environment complexity. | It is bounded by scope and time, so it says what was found rather than what exists. | Answering an external party, and finding the defects that only exist as a combination. |
Sequence
Why running the test first wastes the test
A penetration test bought before vulnerability management exists spends its days on findings a scanner would have produced for a fraction of the cost. An unpatched web server, an end-of-life database, a default credential on a printer — these appear in the report, consume the tester's time and the reader's attention, and displace the findings that only a person could have discovered.
The reverse order is cheap and obvious once stated: turn scanning on, work through what it finds until the noise floor is patching rather than architecture, then commission a test and get a report about authorisation, business logic and chained access. Organisations that do this describe the second test as far better than the first, and usually attribute that to the supplier rather than to their own estate having improved.
There is one legitimate exception. Where an external party has set a deadline — a customer contract, a tender, a framework assessment — the test is the deliverable and the sequence has to yield to the date. In that case the honest approach is to run a scan first anyway, in the week before, and fix whatever it returns cheaply, so the paid days go somewhere useful.
By situation
What was actually asked of you
- 01
"Provide evidence of regular security testing"
Read the wording closely: "regular testing" is frequently satisfied by a documented vulnerability management process with evidence it runs, which you can build once and keep. Where the requester means a penetration test they usually say so, and often specify independence. Ask before buying — the difference is an order of magnitude in cost.
- 02
"We are launching a customer-facing application"
A test, because the failure mode you care about is authorisation between customers and no scanner finds that. Run the scan first as a matter of course, but the purchase is the test.
- 03
"We have three hundred servers and no idea what is on them"
Neither, yet. The first purchase is discovery and inventory, because both activities produce findings against an asset list and an incomplete list produces incomplete findings. This is the case where the honest recommendation is not to buy a test.
- 04
"Our insurer is asking"
Usually a scan plus evidence of a patching process, because what the insurer is pricing is the likelihood of a commodity attack succeeding. Read their form rather than assuming; some do ask for a test and name the scope.
What neither one is
Who issues the certification
What we do
- Say which of the two answers the question you were asked, including when the answer is that neither is needed yet.
- Provide a letter of attestation after a test, stating scope, dates, methodology and the status of findings at retest.
- Help design vulnerability management as a process you own, rather than selling it back to you as a service.
Where the requirement comes from a national framework, the assessor decides which activity satisfies the control and what evidence they accept. Ask them before scoping — the answer is frequently narrower and cheaper than the assumption.
Primary sourceCommon questions
What buyers ask
Suppliers keep quoting me "VAPT". Is that both?+
Sometimes, and sometimes it is a scan with a manual review of the output, priced as a test. The term is used loosely across this market and is not evidence of anything by itself. The way to tell is to ask what the deliverable contains: if every finding maps to a public vulnerability identifier and none of them describes a sequence of actions across two accounts, you were sold an assessment. Neither is wrong to buy — but you should know which one you have.
Can one supplier do both?+
Yes, and there is an efficiency in it, because the tester starts from a current picture of the estate rather than building one. The thing to avoid is a supplier who runs your vulnerability management as a managed service and then tests the result — they are marking their own work, and any assessor asking for independence will say so.
How much does each one cost?+
A vulnerability management capability is a licence plus staff time, priced per asset, and the licence is usually the smaller half — the cost that surprises people is the effort to work through the output every week. A penetration test is a day rate against a scope: eight to fifteen days for a mid-sized application with several roles, fewer for a narrow network scope, plus retest. Anyone quoting a test before understanding the role count and the integration surface is quoting a template.
In this cluster
A short conversation that identifies which activity answers the question you were asked, and says so even when the answer is that neither is needed yet.
Work out which one you need