Skip to main content
Legia

Cybersecurity

Social Engineering Simulation for Employees: How to Test

A social engineering simulation tests whether employees recognize and report fraud attempts. This guide covers the forms of simulation, the approvals needed, how to read the results, and follow-up that does not damage trust.

SEO and Content Specialist

8 min read Reviewed by Legia Cybersecurity Team

A combination padlock on a white keyboard

Short answer: a social engineering simulation is a controlled exercise that imitates fraud aimed at employees, for example through phishing email, phone calls, or text messages, to measure how often employees fall for it and how quickly they report it. A simulation is run only with written approval from leadership, results are read per group rather than per person, and it is not used to punish.

Even a hardened system can be entered through a single message answered by the wrong person. That is why many companies add a social engineering simulation for employees to their security program, whether to meet audit requirements, to assess training results, or to find the weak points before an attacker does. A badly designed simulation damages trust and makes employees reluctant to report. This guide covers the forms of simulation, the ground rules before starting, how to read the results, and the follow-up.

What forms does a social engineering simulation for employees take?

CISA defines social engineering as an attack that uses human interaction and social skills to obtain information about an organization or its systems, often by building false trust. A simulation imitates that attack within agreed limits. The three forms CISA names are the basis of the following breakdown:

FormChannelExample scenario for an office (illustration)What is tested
PhishingEmail or a fake websiteA “your password is about to expire” email with a link to a fake login pageWhether the employee clicks, enters credentials, or reports
VishingVoice callA caller claims to be from IT and asks for a verification codeWhether the employee verifies the caller through another channel
SmishingText messageA message with a “parcel held” link to a work numberWhether the employee opens a link from an unknown message

The scenarios in the third column are only examples. A good scenario is built from real processes in your company, not picked from a generic list.

Illustration: cybersecurity

What does the simulation actually measure?

The figure most often reported is the percentage of employees who click. It is easy to count, but narrow. The UK’s NCSC suggests measuring successful reports alongside clicks, because employees who report suspicious activity become an early warning system for the security team, whereas rewarding only zero reports pushes people to stay quiet.

Three more meaningful measures:

  1. Reporting rate: what percentage of recipients reported the email or call to the IT team.
  2. Time to first report: how fast the IT team learns something suspicious is happening.
  3. Click or credential-entry rate, read together with the difficulty of the scenario.

The third point is often missed. The authors of the Phish Scale paper, NIST researchers, state that click rates should be expected to vary with how difficult each email is for a given audience, so a high or fluctuating figure does not automatically mean training has failed. That difficulty is rated by two components: cues, the observable features of a message such as spelling errors, spoofed sender names, or urgent tone, and premise alignment, how well the message’s pretext fits the recipient’s work context. An email closely matched to the recipient’s job will be clicked more than a generic one, so the results of two simulations with different scenarios cannot simply be compared.

Approval and ground rules before starting

A social engineering simulation differs from system testing because its targets are people, and that needs extra care. Before starting, make sure the following is in writing:

  1. Written approval from company leadership, stating the purpose, timing, permitted scenarios, and the groups covered.
  2. Involvement of HR and legal. NCSC advises checking with HR before running simulations, because punishing people who click a simulated email can look like entrapment.
  3. A basis for processing personal data. Per-person results, such as who clicked or entered credentials, are employees’ personal data. Article 20 of Law Number 27 of 2022 on Personal Data Protection requires a data controller to have a basis for processing, among them consent or other legitimate interests balanced against the data subject’s rights. Decide the basis to use together with legal counsel, and tell employees that the simulation program exists. This is not legal advice.
  4. Scenario limits. Prohibit themes that touch sensitive matters such as terminations, bonuses, health, or family. This is a practice recommendation, not a legal requirement: such scenarios provoke emotional reactions and damage trust.
  5. An emergency contact and a way to stop. Someone with authority must be able to halt the simulation immediately if there is confusion or a complaint.
  6. Rules on captured data. In credential-capture simulations, set that real passwords are not stored. Just record that the fake login page was filled in.
  7. Limited notification. The helpdesk team needs to know a simulation is running so employees’ reports are not treated as real incidents that consume resources, while the simulation’s targets, of course, are not told.

A similar approval structure is used in other security testing. For a picture of preparation on the IT team’s side, see Preparing for Penetration Testing.

Designing scenarios that are useful

A useful scenario tests the processes that are actually vulnerable in your company. Some guidelines:

  • Start from real processes. Payment approvals, access requests, password resets, and document sharing are points fraudsters often exploit.
  • Raise difficulty gradually. Start with scenarios that are easy to recognize, then increase difficulty in later rounds, and note each scenario’s difficulty so results can be compared.
  • Test the verification procedure too. A call that asks for an important action should be verified through a second channel. Testing whether that procedure is followed is often more useful than counting clicks.
  • Provide immediate feedback. An employee who clicks should immediately see a short explanation of the signs that could have been spotted, not wait for results next week.

Reading results without blame

NCSC states that no one can be expected to recognize all phishing emails, and that blaming users who click does not change behavior, because clicks are influenced by personality and situational factors such as stress or being busy. Training and simulation are therefore one layer among several layers of defense, not the main defense.

Reporting practices consistent with that:

DoAvoid
Report results per group or departmentAnnouncing a list of names who clicked
Compare against scenario difficultyComparing figures from two scenarios of different difficulty
Highlight reporting rate and reporting timeReporting only the click rate
Give short training to those who clicked, without sanctionLinking results to performance appraisals
Celebrate correct reportsStaying silent toward those who report

Follow-up: fix the system, not only the people

NCSC describes four layers of defense against phishing, and a simulation touches only the second. Simulation results should be used to strengthen all four:

  1. Make it harder for attackers to reach users: set up SPF, DKIM, and DMARC and filtering of incoming email. The steps are in SPF, DKIM, and DMARC for Company Email.
  2. Help users recognize and report: provide a clear reporting button or address, and make verification through a second channel a habit for important requests.
  3. Limit the damage if someone is fooled: updated devices, restricted admin accounts, two-step authentication, and least-privilege access. CISA states that MFA makes users significantly less likely to be hacked, because one stolen factor is not enough. See What Is Two-Step Authentication.
  4. Respond to incidents quickly: logs and a practiced incident response plan. The guide is in Cyber Incident Response Plan.

Checklist before running a simulation

  1. Set the purpose: assessing training, meeting an audit, or finding weak points.
  2. Obtain written approval from leadership, and involve HR and legal counsel.
  3. Decide the basis for processing per-person result data and its retention period.
  4. Build scenarios from real processes and note their difficulty.
  5. Prohibit sensitive themes and set what data must not be captured.
  6. Inform the helpdesk and set an emergency contact who can stop it.
  7. Run it, then show immediate feedback to those who were fooled.
  8. Report results per group, with the reporting rate beside the click rate.
  9. Apply follow-up to all four layers of defense, not just training.
  10. Repeat periodically with new scenarios and compare fairly.

How Legia does it

In the Penetration Testing service, Legia offers social engineering simulation such as phishing against employees as an addition to website testing. The simulation is run only with separate written approval from company leadership, and is priced separately from the website test. The main testing itself focuses on websites and web applications, and no testing of people is done without that separate written approval. Testing begins only after a written authorization letter that states the scope, schedule, limits, and an emergency contact who can stop it at any time. If you are still weighing which type of security check is right, compare them first in Vulnerability Assessment or Penetration Testing. Other services in this pillar are on the Cyber Security page.

Frequently asked questions

Is a phishing simulation mandatory?

Basically it is the company’s decision. If an auditor or customer asks in writing for evidence of a security awareness program, follow that requirement and ask whether simulation is part of it.

Should employees be told a simulation will happen?

Employees should know that a simulation program exists, including its purpose and the rule that results are not used for sanctions. The only things withheld are the timing and content of each scenario, so the results reflect everyday behavior.

How often should simulations run?

The sources this article refers to set no fixed frequency. Choose a periodic rhythm with changing scenarios, and read the results together with other changes such as training or technical controls.

What do we do if an employee enters credentials?

Treat it as a simulation event: show the explanation immediately, remind them to change their password if their real password was used, and record it as group data. Do not punish. NCSC warns that fear of sanctions makes employees not report real mistakes.

Sources

Category Cybersecurity

Keep reading

Related articles

All Articles