A contact form on a business website, a comment field beneath an article, a publicly listed email address: each of these entry points can be targeted to test how resilient an infrastructure is to unwanted messages. Before launching an internal awareness campaign or a filtering test, you need to establish a clear technical framework to avoid having your main domain penalized by email providers.
Dedicated sending domain and authentication: the technical foundation before any test
Sending a test message from your website’s main domain puts the sender reputation of the entire organization at stake. The volumes sent through a newsletter, CRM, or online store are aggregated at the domain level: a poorly isolated test can damage the deliverability of legitimate emails.
The first practical step is to create a subdomain reserved for testing (for example, test.monsite.fr). Configure SPF, DKIM, and DMARC on it before sending anything. Since 2024, Google and Yahoo have required these authentication methods for bulk senders, and Microsoft followed suit in 2025. Without all three, the message goes straight to spam, skewing the test and damaging the domain’s reputation.
When you want to create an unwanted message online as part of a simulation, always use an isolated sending environment. The dedicated subdomain absorbs any complaints without contaminating the main email stream.
Configuration points to check before the first send
- The subdomain’s SPF record authorizes only the server or tool used for the test, not the entire infrastructure.
- The DKIM key is generated specifically for the subdomain, with a key length that meets current recommendations.
- The subdomain’s DMARC policy is set to “quarantine” or “reject” so you can clearly observe how recipient filters behave.

Internal phishing simulation: target willing participants
Sending a phishing test to all employees without telling anyone creates an immediate operational problem. “Spam” reports are sent to email providers as complaints. Exceeding a certain complaint threshold can trigger restrictions on the sending domain, even if the message is an internal exercise.
Working with a small group of willing participants helps control this risk. Select a test group, inform the IT department, and limit each wave to a few dozen messages.
Build a realistic scenario without triggering filters
Anti-phishing filters analyze several signals: shortened links, executable attachments, and sender impersonation. To genuinely measure human vigilance, the message should resemble a credible attempt without using techniques that would be blocked before reaching the inbox.
Choose a plausible subject line (order confirmation, password update), a link pointing to an internally controlled page, and a displayed sender address that matches the test subdomain. The destination link must never collect real data: it should redirect to an awareness page explaining the exercise.
The message format matters too. Real phishing emails often contain carefully chosen mistakes or create a sense of urgency. Reproducing these elements in the simulation helps measure whether employees spot the warning signs, provided the scenario is proportionate to the maturity level of the team being tested.
Unsubscribe rules and legal compliance for email tests
Recent email provider requirements include a one-click unsubscribe mechanism for marketing messages. This requirement also applies to test emails if they are sent to external recipients or to lists containing personal email addresses.
In France, marketing emails sent to individuals require prior consent. The rules are different for internal tests: employers can organize security exercises within the scope of their IT policies. The distinction lies in obtaining informed consent after the exercise. Informing participants after the test, explaining what was measured and how their data was handled, remains good practice and is consistent with the GDPR.
How the legal framework affects simulations
- An internal phishing exercise does not require prior consent from every employee, but the company’s IT security policy must explicitly provide for it.
- Data collected during the test (click-through rates, identification of users who clicked) is personal data subject to the GDPR: limited retention period, restricted access, and documented purpose.
- If the test involves external addresses (partners, service providers), prior consent is required again.
- The message must include an unsubscribe link if it uses a marketing channel, even for testing purposes.
Post-send monitoring and domain reputation metrics
Once the test is underway, monitor three key metrics: the spam complaint rate reported by providers, the hard bounce rate, and the subdomain’s reputation score using tools such as Google Postmaster Tools.
A complaint rate that rises above the threshold tolerated by providers indicates that the test has gone beyond its intended scope. If this happens, immediately stop any remaining sends and investigate the causes: excessive volume, uninformed recipients, or overly aggressive content.
The time between sending and measurement varies depending on the recipient’s email provider and the volume processed. For small test volumes, consolidated data generally appears within a few hours. You can then compare results across waves to refine the next scenario.
A subdomain’s reputation builds gradually. Launching a large first test from a new subdomain almost always triggers a block. Start with very low volumes, then scale up over several days. This warm-up protects the sending capacity of the main domain, which remains unaffected throughout the test campaign.



