Context is what separates a generic phishing email from a convincing social engineering attack. A contextual layer means the simulation incorporates OSINT-gathered intelligence about the target: their role, their recent public activity, their relationships, their communication patterns, their organization’s current events, to construct a scenario that feels specific and credible rather than generic and suspicious.

Ask whether their platform conducts pre-engagement OSINT reconnaissance. Ask whether simulation scenarios are customized to reflect the target organization’s actual internal context, a pending merger, a recent executive departure, a known vendor relationship, or whether they pull from a generic template library. Ask whether they can demonstrate what a contextually enriched simulation looks like versus a baseline campaign.

• Do you conduct OSINT reconnaissance before building simulation scenarios, or do you use templated content?
• Can you incorporate our organization’s current events, executive roster, and vendor relationships into the simulation narrative?
• Can the contextual layer extend to AI voice cloning, using the target executive’s actual speech patterns derived from public recordings?

There is a meaningful difference between a pre-recorded deepfake video and a live, interactive AI Avatar that can hold a conversation, respond to questions, and adapt in real time. Adversaries like UNC1069 have used synthetic video meetings where the “executive” on screen responds dynamically to what the target says. A static deepfake clip does not replicate that scenario. A conversational AI Avatar does.

Ask whether their Avatar can conduct a two-way interaction, not just play back a video. Ask whether it can be configured with a specific executive’s communication style. Ask what happens when a target asks an unexpected question mid-call.

This question sounds soft. It is not. There is a fundamental difference between a platform built by a product team that read threat intelligence reports and a platform built by operators who have actually run adversarial social engineering engagements against real organizations with real stakes.

Platforms built by product managers optimize for user experience, dashboard metrics, and SaaS scalability. Platforms built by practitioners optimize for adversarial fidelity: the degree to which the simulation reflects what a real threat actor would actually do. Ask who designed the methodology. Ask for their red team credentials. Ask which real-world attack campaigns informed the platform’s design decisions.

• Who specifically designed your simulation methodology, and what is their red team background?
• Can you share documented engagement results from real production deployments?
• Have your practitioners published threat research or spoken at security conferences?

A vendor that only simulates published, attributed attack campaigns is always one step behind the threat. By the time a campaign like Black Basta’s Teams-based social engineering chain was written up by CISA, threat actors had already evolved it. Vendors who conduct their own threat research understand where the attack surface is moving before the public advisory drops.

Ask whether they have a threat research function. Ask whether they have mapped AI-enabled social engineering TTPs that have not yet appeared in published threat intelligence. Ask whether their simulation scenarios are informed by their own engagement data, not just vendor blog posts and public advisories.

This is the most technically significant question on this list, and most buyers do not know to ask it. Multi-channel means the platform can send attacks across email, SMS, and voice. Orchestrated means the attacks are sequenced, conditional, and adaptive. Each step is triggered by the target’s behavior in the previous step, exactly as a real adversary would operate.

In a multi-channel campaign, all vectors fire simultaneously or on a fixed schedule. In an Orchestrated Phishing Simulation, the platform escalates intelligently: if the target opens the email, the vishing call triggers. If they engage on Teams, the deepfake meeting link follows. If they click, the campaign adapts the next stage accordingly. These are fundamentally different threat models.

• Does the next attack step trigger based on the target’s actual behavior, or does it fire on a fixed schedule regardless of engagement?
• Can a campaign escalate from email to voice to a deepfake meeting link as a connected sequence?
• Can campaigns be paused or modified mid-engagement based on real-time target behavior?

Most simulation platforms are designed to measure awareness at population scale: send a thousand phishing emails, measure who clicks, generate a dashboard. That is a fundamentally different capability from a red team engagement targeting specific individuals with OSINT-informed, bespoke social engineering attacks.

Ask whether their platform can be configured for named-target operations with individualized reconnaissance. Ask whether they can execute a full red team engagement, not just run a simulation template, against specific executives or high-value targets. Ask whether their practitioners can sit across from your red team and coordinate a joint human-technical operation.

Microsoft Teams is no longer a secondary communication channel. It is the primary collaboration infrastructure for most enterprise organizations, and it has become one of the most actively exploited social engineering vectors in documented threat campaigns. Black Basta affiliates used external tenant impersonation on Teams to deploy ransomware across hundreds of organizations. The vector is real, it is active, and most simulation platforms cannot test it at all.

Ask specifically whether they can simulate an inbound Teams message from an attacker-controlled external tenant impersonating your IT help desk. Ask whether they can generate deepfake video meeting links delivered inside Teams. Ask whether Teams simulation is a production capability or a roadmap item.