Skip to content
Technology transformation expertise firm · ARDIANE Group Innovate, secure, succeed

Engineering

Application security, question by question, for a management team that makes the calls

In brief

Application security brings together the practices that prevent software from being diverted from its use, from design to operation. Management Square reviews here the questions a management team asks about the security of its applications (scope, risks, timing of controls, responsibilities, applicable texts, evidence) and what a verifiable answer must contain.

Page contents

Application security is the set of practices that prevent an application from being diverted from its use, from design to operation. It concerns what an organisation designs, writes, assembles and configures: the code, the third-party components it embeds, the access rules, the deployment settings. The protection of networks, workstations and servers falls under other cybersecurity disciplines, which complement it without taking its place.

Before an executive committee, the subject comes down to six questions: which applications to deal with first, which defects to fear, when to act, who is accountable, which texts apply and on what evidence to judge. None of them requires reading code. Each calls for an answer that a manager can write and that a committee can verify.

Which applications should application security cover first?

An identical effort across the whole application estate is diluted. The expected answer is a ranking based on three criteria: exposure (application open to the Internet, to partners, or only internal), the data handled (personal, financial, health) and the role in the business, that is what a shutdown or a falsification would cost.

This ranking relies on the application map, whose upkeep is described on our IT architecture page. A management team that cannot obtain the list of its exposed applications, with an owner against each one, has found its first piece of work.

Each class then receives a level of requirement. The OWASP Application Security Verification Standard (ASVS), version 5.0.0 of which was published on 30 May 2025, divides its security requirements into three levels, from a first adoption tier up to applications that require high assurance. The project page states that the standard can serve as a basis for the verification requirements written into a contract, which applies to applications entrusted to a service provider.

What defects are we talking about?

The OWASP Top 10 provides the vocabulary. The OWASP Foundation presents it as an awareness document, which expresses a consensus on the most critical security risks for web applications. The most recent edition is that of 2025.

It puts Broken Access Control (A01) at the top, followed by Security Misconfiguration (A02) and Software Supply Chain Failures (A03). It also includes Insecure Design (A06).

For a committee, the lesson lies in the nature of these categories. A user who views someone else’s file, a default setting left open, a library compromised before being integrated, a business rule that can be bypassed: these defects arise from a design, assembly or deployment decision as much as from writing the code. An answer that merely cites a code analysis tool therefore covers only part of the list.

At what point in the lifecycle should you act?

Each stage calls for a different control, and each control has its blind spot.

Moment Control What it finds What it lets through
Design Threat modelling, written security requirements Design flaws, before they are built Implementation errors
Writing the code Static analysis (SAST), secret detection Known dangerous code patterns, credentials left in the repository Logic errors, such as a forgotten access control
Assembly Third-party component analysis (SCA) Published vulnerabilities in the libraries used A vulnerability not yet published
Before go-live Dynamic analysis (DAST), penetration test What is exploitable on the deployed application, configuration included What was not covered during the test
In production Logging, alerts, vulnerability monitoring Attack attempts and flaws revealed after delivery What nobody is monitoring

Tooled analyses (SAST, SCA, DAST) run in the delivery pipeline, on every change or every deployment. How they fit with other tests is described on our software quality and testing page. Logic errors are for a human reviewer, according to the organisation set out in our article on code review.

The question to ask concerns the distribution. A set-up concentrated on the final penetration test discovers late what is expensive to fix: a design flaw found at that stage means reworking what has been built on top of it.

This table stops at detection. What to do when a flaw is exploited nonetheless (continuity, recovery, crisis management) falls under our cyber resilience page.

Who is accountable for the security of an application?

An answer such as “the security team” signals an organisational problem: that team cannot review every change to every application. A workable split distinguishes four roles.

The team that develops the application is accountable for what it delivers. It receives the alerts from the tools, sorts them and fixes them. The security team sets the rules, provides the tools and expertise, and commissions penetration tests. The product owner decides where fixes stand against features, within a period agreed for each severity level. A named person, on the business side, accepts in writing the risk of a go-live with a known vulnerability.

To place the organisation as a whole, the OWASP SAMM maturity model describes fifteen security practices divided into five business functions: governance, design, implementation, verification and operations. It is used to choose where to progress first, practice by practice.

Which texts apply?

Directive (EU) 2022/2555, known as NIS2, includes among the risk-management measures expected of essential and important entities “security in network and information systems acquisition, development and maintenance, including vulnerability handling and disclosure” (Article 21(2)(e)). Its scope in France depends on transposition.

For financial entities, Article 25 of the DORA Regulation, Regulation (EU) 2022/2554, lists among the tests in the digital operational resilience testing programme vulnerability assessments and scans, source code reviews where feasible and penetration tests.

Regulation (EU) 2024/2847, known as the Cyber Resilience Act, applies to products with digital elements made available on the market, where their intended or reasonably foreseeable use includes a connection to a device or network (Article 2). It sets cybersecurity requirements for their design, development and vulnerability handling (Article 1), according to a staggered application timetable (Article 71). It concerns an organisation as a manufacturer, when it places software on the market itself.

Determining whether an entity or a product falls within the scope of these texts is a matter for legal analysis.

On what evidence should you judge?

Four measures, presented at regular intervals, give an overall view:

  • the share of applications in the most exposed class covered by each control in the table;
  • the time taken to fix vulnerabilities, by severity level, compared with the agreed time;
  • the number of known vulnerabilities in production accepted by exemption, and the age of the oldest;
  • among the findings of the last penetration test, those that a control placed earlier should have picked up.

A raw number of vulnerabilities detected says nothing on its own: it rises when the tooling improves.

At the end of the session, one point falls to the committee itself: the name of the person authorised to accept, in writing and for a limited period, the go-live of an application with a known vulnerability that has not been fixed. As long as that name is missing, acceptance is likely to happen anyway, without a trace, within the team under pressure to deliver.

FAQ

Frequently asked questions

Another question? Ask us

What is the difference between application security and cybersecurity?

Cybersecurity covers all the means that protect an information system: networks, workstations, servers, identities, staff awareness, incident response. Application security is the part of it that concerns the software itself: its design, its code, the third-party components it embeds and its configuration. A vulnerable application remains exploitable behind a well-protected network as soon as it is open to its users.

What is the OWASP Top 10?

The OWASP Top 10 is an awareness document published by the OWASP Foundation, which lists ten categories of security risks for web applications. Its 2025 edition puts Broken Access Control at the top. It is used to name and prioritise risks; for a list of verifiable requirements, the same foundation publishes the Application Security Verification Standard (ASVS).

What do SAST, DAST and SCA mean?

SAST (static application security testing) analyses source code without executing it, looking for known dangerous patterns. DAST (dynamic application security testing) probes the running application, as an attacker would, without access to the code. SCA (software composition analysis) inventories third-party components and flags those that carry a published vulnerability. The three complement one another, and none of them detects a logic flaw specific to the business.

Is a penetration test enough to establish that an application is secure?

A penetration test records what specialists managed to exploit on a given version, within limited time and scope. Its report is valid neither for subsequent versions nor for what was not covered. It measures the effectiveness of the controls placed upstream and does not replace them.

What is DevSecOps?

The term DevSecOps refers to the integration of security controls into the delivery pipeline and into the daily work of development and operations teams. Analyses of the code and third-party components run on every change, and their results go back to the team that wrote the code. Security ceases to be a validation step placed at the end of the project.

A project to bring to completion?

A first conversation is enough to understand your challenge, identify the right people and tell you frankly what we can bring.