All articles
Cyber security

Our in-house phishing testing system is now live and fully operational

Published5 September 2026
Reading time5 min read
AuthorJames Powell
In short

i90's in-house phishing simulation platform is now live and fully operational, offered free to clients on a managed service agreement, because phishing remains one of the most common ways businesses get compromised, and staff awareness training is one of the cheapest, most effective defences against it. Here's exactly what happens, end to end, when a simulation runs.

Why we built this ourselves

We already manage email for many of our clients as part of their existing managed service agreement, hundreds of IMAP email accounts, alongside Microsoft 365 and Google Workspace. Phishing simulation is a natural extension of that, testing the same inboxes we already look after, so we built it as an add-on to the relationship rather than a separate paid product.

Commercial phishing-simulation platforms exist, but most charge per seat, which would make offering this free, at scale, across many clients, unworkable. Rather than build a sending and tracking engine from scratch, we evaluated the leading open-source option, GoPhish, an actively maintained, proven framework for exactly this kind of testing, and built the parts that actually matter around it: the branding, the client onboarding process, the safety controls, and the debrief experience.

How a simulation works

A realistic pretext email lands in someone's inbox. For our pilot, this was a routine-looking "your mailbox is almost full" notification, styled and worded like a plausible, everyday IT message, not an obviously fake test email.

It's sent properly. The email goes out through a correctly configured mail relay with full SPF, DKIM and DMARC authentication set up on a dedicated domain used only for this purpose. Technically, it behaves exactly like a real email would, no shortcuts that would make it obviously fake, because the whole point is a realistic test.

If someone clicks, they land on a fake but harmless sign-in style page, mimicking a typical login prompt.

Every stage is tracked per person: whether the email was delivered, opened, the link clicked, and whether the form was submitted. What's never tracked or stored is anything actually typed into a password field. That was a deliberate, non-negotiable design decision, tested and verified before this ever went near a real client. A training exercise capturing someone's real credentials would turn a test into an actual data-security incident, so it simply doesn't happen.

Immediately after submitting, the person lands on a purpose-built debrief page explaining, plainly and without judgement, that it was a test, why that particular email was designed to be convincing, and what to look out for next time. The tone is deliberately warm rather than shaming, the goal is a teachable moment, not embarrassment.

Results feed back into reporting, giving a clear picture of how a group of staff responded. Useful for a client's own security awareness, and, in regulated sectors, as supporting evidence for frameworks like Cyber Essentials Plus or ISO 27001 audits. If you want a refresher on what to look for yourself, see our guide on how to spot a phishing email.

Phishing simulation How a test moves through six stages
  1. 01 Email sent

    A realistic pretext lands via an authenticated mail relay

  2. 02 Opened

    Delivery and open events are tracked per person

  3. 03 Clicked

    The link is followed through to a landing page

  4. 04 Landing page

    A harmless sign-in style page, no passwords stored

  5. 05 Debrief

    An immediate, non-judgemental explanation of the test

  6. 06 Reporting

    Results feed into client reporting and audit evidence

Because this involves sending deceptive email and measuring the behaviour of named individuals, the legal, ethical and consent framework around it was treated as a first-class part of the build, not an afterthought.

No campaign runs without sign-off. Nothing is sent to a client's staff without signed authorisation from an appropriately senior contact at that client first.

Only the data needed for the training metric is collected. Delivery, opens, clicks and submissions, never the content of anything typed.

The debrief experience is built to educate, not to catch people out.

Client organisations remain the data controller for their own staff's information; i90 acts as a processor under the existing managed services relationship.

Why this is different from buying it in

Most commercial phishing simulation platforms lock clients into per-seat subscription contracts, often on multi-year terms, so you're paying a fixed cost regardless of how much you actually use it. i90's approach avoids that entirely: it's included free as part of your existing managed service agreement, no separate contract, no per-seat cost, no paying for months you don't use.

Built with Claude as a genuine technical partner

Worth being upfront about: the whole platform, from initial requirements and architecture through to a working, tested, client-ready service, was built using Claude as a hands-on technical collaborator throughout, drafting the legal and ethical framework, working through server setup, debugging real infrastructure issues, writing the email and landing page content, and building the client onboarding process. It's a genuine example of an MSP building new internal tooling with AI properly in the loop, not just as a text-drafting tool.

Where things stand right now

We've run a full internal pilot against our own team first, proving the entire flow end to end before it went anywhere near a client, followed by real client testing carried out with proper senior-level authorisation. The service is now fully operational: client reporting, a broader library of pretext scenarios beyond the original mailbox-storage test, and a repeatable onboarding process are all part of how it runs today, alongside the safety and consent framework described above, which hasn't changed and never will.

Common questions

Will staff get in trouble if they click a simulated phishing email?

No. The whole design of the debrief experience is built around education rather than punishment, a warm, non-judgemental explanation of what happened and what to look out for next time, not a way of catching people out.

Is any real personal data or passwords captured during a test?

No. The platform deliberately never stores anything typed into a password field, only whether an email was delivered, opened, clicked, and whether the form was submitted. That was a non-negotiable design decision, tested and verified before any real client test.

Who has to approve a phishing simulation before it runs?

No campaign runs without signed authorisation from an appropriately senior contact at the client first. It's never run against staff without that sign-off in place.

Can phishing simulation results help with Cyber Essentials or ISO 27001?

Yes, in regulated sectors particularly, the results can serve as supporting evidence of staff security awareness for audits under frameworks like Cyber Essentials Plus or ISO 27001, alongside your other controls.

Want to know if your team would spot a phishing email?

If you're already on a managed service agreement with us, phishing simulation is included at no extra cost. Get in touch to find out more.

Get in touch