How it usually happens
The pattern is remarkably consistent. A startup builds — or buys — its product fast. Sometimes an in-house team ships features every week; sometimes an agency or freelancers deliver the app and move on. Either way, nobody runs a security test, because nothing forces it yet.
Then a forcing event arrives: an enterprise prospect sends a security questionnaire, a SOC 2 or ISO 27001 audit starts, or an investor's due diligence asks for evidence of security testing. The question is always some version of "Can you share a recent penetration test report?" — and the honest answer is no.
If your software was bought, not built
Startups that commission their app from an agency or freelancers carry an extra risk: they often have no view into how authentication, access control and data handling were implemented. The delivery contract rarely includes a security test, and the people who wrote the code are gone when the findings arrive. Fixing someone else's code under deadline is slower and more expensive than fixing your own.
If this is you, test before you accept the delivery — or at the latest before your first enterprise customer does it for you.
What the bill actually includes
The pentest invoice is the visible part. It is rarely the biggest one.
| Cost line | Typical range | Notes |
|---|---|---|
| Manual pentest (web app + API) | $8,000–$25,000 for a typical SaaS scope | Vendor pricing guides; scope drives price more than company size. |
| Retest after fixes | Included, partial, or extra | Varies by firm — confirm in writing before you sign. |
| Engineering time to fix | Often the largest line | See the worked example below. |
| Delayed deal | Contract value × months of delay | The customer waits for a clean report before signing. |
| Worst case: a breach | $4.44M global average | IBM 2025; $10.22M average in the US. Mostly larger organizations — but the categories of cost apply to everyone. |
Worked example (illustrative — use your own numbers)
A first pentest on an app that was never tested returns 12 findings. If each takes one to three engineer-days to fix, test and deploy, that is 12–36 engineer-days. At an internal cost of €600 per engineer-day, remediation alone is €7,200–€21,600 — on top of the pentest fee — and those are days the team is not shipping product. Add a retest, and the calendar time from "the customer asked" to "here is a clean report" is easily measured in weeks.
Why fixing late is slower — and getting slower
Vulnerabilities found long after the code was written are harder to fix. The developer who wrote it has moved on to other work (or left), the insecure pattern has been copied into other endpoints, and other features now depend on it.
Industry data shows the same trend. Veracode's 2025 State of Software Security report found that the time to fix half of an application's security flaws has grown to 252 days on average, up from 171 days five years earlier — a 47% increase. Leading teams fix half of their flaws in about five weeks; the slowest take longer than a year. Veracode also reports that around 74% of organizations carry security debt (flaws left open for more than a year), and nearly half of that debt is critical.
The difference between those groups isn't talent. It's when testing happens: teams that find issues while the code is fresh fix them while the fix is still small.
What a "clean report" actually requires
A report your customer or auditor will accept usually needs:
- A defined, dated scope that covers the application the customer will actually use.
- No open critical or high-severity findings — or a documented, accepted plan for each one.
- Evidence that fixes were verified, not just claimed: a retest or a re-run of the same attack.
- For a formal compliance requirement, a report from the testing provider your auditor expects — confirm this with your auditor early.
The cheaper path: test from the first release
The alternative is not "never buy a pentest". It is to arrive at the pentest with an application that has already been tested many times:
Run automated security tests against staging on every release. Fix confirmed findings the same week, while the code is fresh.
Test the delivery before you accept it, and make fixing confirmed findings part of the agency's handover.
Clear the OWASP Top 10 basics yourself, so paid consultant-days go to business logic and complex attack chains instead of findings a tool would have caught.
Re-run the same attacks to verify each fix immediately, instead of waiting for a scheduled retest window.
Keep testing on every deploy so regressions don't quietly reopen what you already fixed.
The result: the first manual pentest finds fewer and less severe issues, remediation takes days instead of weeks, and the clean report arrives while the deal is still warm.
If a customer asked for a pentest report tomorrow
- Ask exactly what they need: a third-party pentest report, a SOC 2 report, or answers to a questionnaire.
- Run an automated test on the in-scope application this week, and fix every confirmed critical and high finding.
- Collect the evidence: dated reports showing what was found and that fixes were verified.
- If a third-party pentest is required, book it now — and confirm in writing whether a retest is included.
- Reply to the customer with a timeline, and the evidence you already have.
For the questionnaire itself, see how to answer a security questionnaire when you've never had a pentest.
Find it while it's cheap to fix.
Nautillo Pro runs automated penetration testing (DAST) against your web app and APIs and proves every finding with the exact request and response — so your team fixes real issues early and arrives at the pentest with a clean baseline. Free to start, no credit card required.
Sources
Related articles