Secure in Theory, and How about Reality? Why Usability Is Part of Security
A security tool can be technically flawless and still fail to protect the systems it was designed to secure. When the tool is difficult to understand, confusing to operate, or does not match the way people actually work, even experienced professionals can make mistakes using it.
Security features may be misconfigured, assessment results misinterpreted, or important warnings overlooked. The consequences can range from security incidents and a false sense of security to regulatory non-compliance and legal liability. In other words, software usability is not simply a matter of convenience. It is an essential part of cybersecurity.
The Cybersecurity Certification and Assessment Tools (CCAT) project recognizes this challenge. CCAT translates cutting-edge academic research into four open-source tools (TLS-Scanner, SCRUTINY, ALVIE and sec-certs) that support the implementation of European cybersecurity regulations across both private and public sectors, while integrating usability into their development from the very beginning. Rather than treating usability as a final polishing step, the project follows a human-centered design approach that starts with understanding users, their tasks, and their working environment. Throughout the tool development process, usability experts collaborate with developers to evaluate prototypes, identify potential barriers, and iteratively improve the tools using established methods such as expert walkthroughs (performing typical user tasks by power users) and heuristic evaluations (judging interfaces against the established usability principles). The goal is to ensure that the tools are not only technically working as designed and implemented, but also intuitive and effective for the professionals who rely on them.
In the following interview, Mariia Bakhtina, a usability expert in the CCAT project from the Centre for Research on Cryptography and Security (CroCS) at the Faculty of Informatics, explains why human-centered design is becoming an indispensable part of cybersecurity, how usability can strengthen rather than compromise security, and what it takes to develop security tools that experts can use confidently and effectively.
Why is usable security often perceived as less established than secure software development?
I wouldn't say usable security is less established than secure software development overall. Both fields have mainly been established around general-purpose software used by non-security experts. For example, if a company builds a booking system, it's common knowledge today that the system should be designed securely, and that security features shouldn't get too much in the way of users completing their main task. Academic research, however, has paid much less attention to the usability of security assessment tools. When a tool is built to help technical experts assess the security of other systems, there's often an assumption that users — security engineers, certification experts, and the like — are technically advanced enough to figure it all out themselves, even if the tool wasn't designed with their actual tasks and expertise in mind.
What expertise do developers need to incorporate usability into security features, and at what stage of software development should human-centred security design be considered?
Usability should be considered from the very beginning of software development. Developers need to understand who their target users are, what level of expertise they bring, which tasks should be handled by the system and which by users, and continuously improve the design through feedback and testing. Throughout the process, developers should put themselves in the users’ shoes and ask: „Would my user know this concept? Would they understand what the system is telling them?“
Can you give an example where a secure system failed due to poor usability?
A well-known example is the landmark "Why Johnny Can't Encrypt" study (Whitten and Tygar, 1999), which examined how people used PGP, one of the first tools for email encryption, a technology that protects the contents of an email so that only the intended recipient can read it. Although the encryption itself was highly secure, the study found that even technically skilled users struggled to use the software correctly. The interface expected them to understand complex cryptographic concepts such as encryption keys, key servers and trust webs without explaining what these terms meant or what actions they needed to take. As a result, many users made mistakes or abandoned encryption altogether and sent their emails without this protection. It's a good illustration of the general pattern we see in our own field too: a security assessment tool can be technically correct and still fail its purpose if the expert using it can't reliably interpret or act on its output.
Who is actually responsible for usable security: developers, designers, product owners, or management?
All of them, together. Usable security should be woven into the entire software development process, just as development itself involves every one of these roles. Product owners need to understand users and how to make security features usable for them, then communicate that vision to designers and developers who implement it, while management needs to ensure the team has the time and mandate to treat usability of security as a non-negotiable part of every project.
How do you see AI and automation affecting usable security in the coming years?
I see it cutting both ways. On the one hand, AI can genuinely improve usability, for example, by translating dense technical output from a security assessment tool into plain language explanations tailored to the user's role, or by flagging likely misconfigurations before a user acts on them. On the other hand, there's a real risk of automation bias: if a tool's AI-generated summary is wrong or oversimplified, users may trust it without verifying, especially if they lack the expertise to sanity check it themselves. So the usability challenge doesn't disappear with AI; it shifts to designing interfaces that communicate uncertainty and limitations honestly, rather than just producing confident-sounding output.
How do we measure whether a security tool is usable?
Ultimately, by testing it with real target users. Before we get there, though, we evaluate the tool against established best practices (usability heuristics) and then design observational studies to see how users perform realistic tasks. We try to answer questions such as: “Did we correctly anticipate the knowledge, skills, and context of our users? Can they interact with the tool as intended? Can they achieve their goal? Where users run into problems, are those caused by poor interface design, weak documentation, or lack of support? Is the tool fast enough and does it tolerate mistakes gracefully?” We answer these by combining direct observation of task performance, timing measurements, comparisons with how developers expect the results to be interpreted, and users' own subjective feedback.
The most valuable feedback combines behavioral data (what users did, where they hesitated or made errors), outcome correctness (whether they reached the right conclusion or configuration), and subjective reflection (what they found confusing or expected to happen instead). None of these alone is enough: users often can't articulate problems they don't consciously notice, and self-reported confidence does not reliably show whether they used the tool correctly. Therefore, we rely on triangulation of observed behavior, task outcomes, and user commentary.

Have you already tested the CCAT tools in real-world settings with real users?
During the initial phase of the project, we identified 307 usability issues and systematically advised tool developers on how to improve their tools' usability. These improvements were subsequently tested at DevConf.cz, a free community conference for developers, DevOps engineers, and other open-source contributors, which took place in Brno this June. During the conference, participants tested two of the CCAT tools by performing tasks that reflected real-world security assessment workflows. The conference attendees represented a diverse group, including software developers and system administrators, project managers and product owners, and security experts. Some of them were potential end users of the tools, while others participated primarily as organizational representatives.
The two tested tools were SCRUTINY, a toolset for assessing cryptographic implementations in hardware (smartcards and TPMs), and software libraries; and sec-certs, a tool for exploring the Common Criteria, FIPS 140, and EUCC certification landscapes and analyzing relationships between certification documents and products. Researchers observed how participants interacted with the tools, where they encountered difficulties, how efficiently they completed their tasks, and whether they interpreted the results correctly. Interestingly, live user testing surfaced the same problems that had already been flagged through expert heuristic evaluation. Together with participants’ feedback, these observations provided valuable evidence for further refining the tools before the release of their new versions this September and their wider deployment.
Learn more about the CCAT project on our website and follow us on LinkedIn.

The project funded under Grant Agreement No. 101225878 is supported by the European Cybersecurity Competence Centre. Funded by the European Union (EU). Views and opinions expressed are however those of the authors only and do not necessarily reflect those of the EU or the European Cybersecurity Competence Centre (ECCC). Neither the EU nor the ECCC can be held responsible for them.


Photo: Martina Miechová
Authors: Martina Miechová, Mariia Bakhtina
- Web address
- Attachments