How to Run a Phishing Simulation: Six Steps and How to Read the Results

A phishing simulation uses controlled, realistic phishing emails to measure how employees open, click and report. This guide covers the six steps, how to read the results, what Taiwan's rules say about frequency, and the difference between doing it yourself and outsourcing.

The short answer: a phishing simulation (社交工程演練) sends controlled, realistic phishing emails to see how employees react to a suspicious message. The aim is to find the situations that cause trouble and arrange training for them, not to find out who clicked. The usual flow is: plan, design the scenario, set up allowlisting, send, collect results, then analyze and teach.

Decide Four Things First

  1. Goal: is this a first baseline measurement, or a check after training? The goal changes how hard the email should be and how you read the result.
  2. Scope: the whole company or specific departments? Finance, HR and customer service see different kinds of email.
  3. Authorization: written approval from someone with the authority to give it, who knows the scope and the time window. A simulation without it can become an internal trust problem.
  4. Recipient list and personal data: the list (name, email, department) is personal data. Decide up front who can see it, what it is used for, and how it is handled afterward. Our practice: only authorized staff handle the list, within what the exercise needs, and it is destroyed after acceptance as agreed; individual behavior goes only to the client's designated contact, and statistics in the report are shown by department or group.

Six Steps

1. Plan and prepare

Fix the recipients, time window and contacts, and decide how to handle real incidents found during the exercise, for example a recipient who reports the simulation email as if it were a real attack.

2. Design the scenario

Match emails the organization really sees, such as parcel notices, account verification or shared documents. Choose the difficulty on purpose. Topics such as pay, bonuses or layoffs can affect employee trust; whether and how to use them has no absolute answer, so agree it with the client beforehand.

3. Handle allowlisting and the sending domain

Send from a test domain that is similar to real ones but identifiable, and agree the allowlist with the IT team beforehand. Without it, mail filtering may block the emails first, and the result then reflects the filter rather than the employees.

4. Send

Sending in batches and across time windows is closer to the pace of a real attack than sending everything at once. It also makes it harder for recipients to warn each other, and reduces disruption in any single time slot.

5. Collect results

Record who received the email, who opened it, who clicked a link or opened an attachment, who entered data, and who reported the message as suspicious.

6. Analyze and teach

Give anyone who clicked or entered data a short, non-blaming explanation at that moment of what to look for. Confirm to anyone who reported that they did the right thing. Then arrange focused training from the analysis and compare at the next round.

How to Read the Results

  • Use one denominator: calculate rates against the number of emails delivered, so rounds can be compared.
  • Account for difficulty: the click rate of an obvious email and a very realistic one cannot be compared directly. NIST SP 800-50 Rev. 1 (2024) notes that methods such as the NIST Phish Scale can be used to assess the difficulty of a phishing email in light of the recipient's context.
  • Do not look at clicks only: clicks and data entry indicate risk, but a report shows that employees know what to do. NIST SP 800-50 Rev. 1 also lists whether a user reports the email among the measurements a phishing exercise can collect.
  • Filter false clicks: mail scanners can open links automatically and create click records that are not human, so filter them out when analyzing.
  • Report by department, not by person: the point is to find what needs strengthening, so reports show departments or groups, not individual names.

We have not listed an "industry average click rate". Published figures vary widely and mostly cannot be traced to an original survey, so we have no figure reliable enough to cite.

How Often? What the Rules Say

  • TWSE/TPEx-listed companies: Article 31 of the Cyber Security Guidelines for TWSE/TPEx-Listed Companies (上市上櫃公司資通安全管控指引, amended 9 November 2026; formerly Article 30, renumbered with the wording unchanged) reads, in our translation: conduct email social engineering exercises regularly each year, give education and training to those who mistakenly opened an email or link, and keep related records.
  • Government agencies: under Article 10 of the Regulations Governing the Reporting, Response and Drills for Cyber Security Incidents (資通安全事件通報應變及演練辦法), the directly affiliated agencies of the Office of the President, the National Security Council and the Five Yuans plan and carry out cyber security drills for themselves and for the government agencies they supervise. The drills must include at least social engineering drills once every six months and reporting and response drills for cyber security incidents once every year. Under the same article, special municipality and county/city governments follow the same requirements for themselves, the agencies they supervise, and the township offices and representative councils under their jurisdiction. Article 16 separately provides that government agencies cooperate with drills planned by the competent authority, which may include social engineering drills. (Laws & Regulations Database of the Republic of China, amended 5 January 2026.)

This summarizes the original text and is not legal advice. Which provision applies to your organization, and what to do about it, depends on the authority's notices and your legal counsel. Exercise and training records can be among the records you keep, but that does not represent the result of any audit.

Do It Yourself With Open-Source Tools, or Hire a Provider?

Doing it yourself costs less and is flexible. The open-source tool Gophish is a common choice; see our Gophish tutorial (Chinese) for installation and setup. But you then have to handle:

  • the sending domain and its reputation, and allowlisting in mail filters.
  • scenario design and difficulty control.
  • personal data in the recipient list and how long it is kept.
  • analysis, management reporting and follow-up training.

If the organization lacks the staff for this, or wants an outside team to run it and produce a report, outsourcing is an option. When comparing, first ask: how many exercises are included, how mailbox tiers are counted, what the report contains, and how the list is handled. ThunderShield's plan and process are on Phishing Simulation, with prices on the pricing page; for what to ask about on any quote, see How Much Does a Website Security Test Cost?.

FAQ

Will a simulation disrupt daily work?

Exercises can be planned around the company's situation, balancing realism with low disruption, and produce risk data without affecting operations.

How often should we run one?

It depends on the rules that apply to you. Article 31 of the listed-company guidelines says regularly each year (see above). Some organizations also run smaller batches more often, so the exercise matches the rhythm of real attacks.

How is personal data in the list handled?

The list is personal data and should be handled only by authorized staff within what the exercise needs, not used for other purposes, and destroyed after the project as agreed.

Can results be used to discipline employees?

We advise against it. NIST SP 800-50 Rev. 1 (2024) says such activities should not be punitive, nor should any employee be called out for their response. The purpose is to find the situations that need strengthening and to arrange training. When results lead to punishment, people tend to hide mistakes, which delays reporting of real incidents.

Next Step

To see how ThunderShield plans a phishing simulation, read about the service and the plans, or contact us with your number of mailboxes and what you want to achieve, and we will reply based on the scope.

More cybersecurity news