Mocke simulates cold email campaigns with your actual customers, so you can see your potential reply rate in one minute - without launching a real campaign. The question we get most often is: how does it work? In this article, I’ll explain the underlying mechanism of Mocke.
When provided with a list of leads and email content, Mocke simulates each recipient’s email behavior independently. Put simply, we create a unique world view for each real person, factoring in both personal and professional backgrounds. Upon receiving an email, every individual is prompted, often subconsciously, to decide what action—if any—to take. It’s almost like building a simulated world within a game.
Below, I’ll walk you through an example of a simulated email experience. Keep in mind, this is a highly simplified version of the implementation. Achieving realistic simulation requires a significant amount of engineering and prompt development to build what we call the “Large World Model” (LWM).
I’ll use myself as the hypothetical prospect for the mock email campaign: https://www.linkedin.com/in/thisismattsun/, and the sample email looks like this.
Subject Line:
Owning the market before it matures
Email Body:
Matt,
Sam Altman has a line: “In the early days, growth solves (nearly) all problems.”
The trick is — you don’t get that kind of growth from product alone. You need a way to repeatedly put it in front of exactly the right people, in exactly the right context, before the market decides the winner.
That’s my specialty:
→ AI-powered targeting of your best-fit accounts
→ Messaging that feels handcrafted at scale
→ GTM systems designed to be an unfair advantage
Should I send you a short GTM plan for Paigo showing how you can win now instead of playing catch-up later?
– Charles
Using the LinkedIn URL, Mocke’s system devotes significant time and resources to building a complete mocked version of “Matt,” which we internally refer to as a world view. This world view includes all the real-life factors that could influence email behavior, whether in major or minor ways. For example, below is a high-level subset of the dimensions we simulate in the Large World Model:
In this simplified example, the mocked “Matt” is fully constructed with all aspects of his work environment simulated. To make the mock as realistic as possible, we’ve configured nearly 200 dimensions in the world view to capture the complexity of real life.
Next, he moves through several steps in the Large World Model.
To make the simulation as realistic as possible, we break down the email-receiving behavior into very granular steps. Here are the core stages:
In Mocke’s Large World Model, this process is like peeling back the layers of an onion. Not every email will reach the later stages of the simulated decision process—just as in the real world, some emails go unopened, are ignored, or are archived. As a result, the decision process forms a funnel. Only the most effective cold emails make it to the end, resulting in a positive reply.
The example above illustrates how one lead is likely to respond to an email. Mocke runs this simulation independently for every single lead using a shared-nothing methodology, which means one person’s reaction to a cold email does not affect any other simulated recipient. After all email behaviors are modeled, Mocke aggregates the total number of replies and other metrics, such as unsubscribes, spam reports, and expressions of interest.
However, the process isn’t finished yet. At this stage, the mocked result is still not identical to reality, because other major factors—especially deliverability—can impact the final outcome.
If you’re not familiar with deliverability, it refers to how likely your emails are to reach recipients’ inboxes. Deliverability is an extremely challenging technical problem, which I won’t cover in full detail here. However, there are certain aspects we must simulate to achieve a more realistic model.
First, we simulate mailbox reputation. Every mailbox has a reputation with major email service providers like Google or Microsoft Outlook. Reputation significantly influences whether an email lands in the inbox or is filtered as spam. To replicate this, we allow users to specify their historical open rate. While this isn’t a direct measure of reputation, it’s closely correlated.
Next, we simulate spam filtering by scanning the email content and comparing it to patterns associated with spammers, phishing attempts, and poorly rated email servers. This process scores how likely an email is to be marked as spam. Most senders are familiar with avoiding certain trigger words like “credit card,” “free money,” or “lottery” to reduce spam risk.
One of the key advantages of our mocked email campaigns is the access to hidden campaign metrics and actionable insights. Here are just a few examples:
This information is derived from the simulated world view of each lead, as described above. In the Large World Model developed by Mocke, we simulate and record the internal thought processes of each mocked individual, allowing us to trace why they reacted a certain way to a cold email. When AI analyzes the overall campaign performance, it reviews the internal thoughts of all simulated leads collectively to identify significant common patterns. Since each simulation is independent, recurring patterns across many leads are strong indicators of cold email quality.
We understand the complexity of real-world campaigns, and mocked results will never perfectly match actual outcomes. To provide more value, Mocke offers a predicted performance range for reply and open rates, rather than single-point estimates. This range accounts for all factors—including deliverability, randomness in email engagement, and each simulated world view—and is presented with a 75% confidence level from our analytical AI.