Mixilo and Continuous Education at Tecnológico de Monterrey: Case Study

‍

Executive Summary

For the Continuing Education division at Tecnológico de Monterrey, every lead represents a purpose: to specialize, change careers, or broaden one's impact. With that conviction, it set out to make its digital campaigns do more than generate interest. They wanted them to help connect each person with the program that matches what they're looking for. To achieve this, it partnered with Muttdata to implement Mixilo AI Conversions, an artificial intelligence model that estimates how likely each lead is to convert into an enrollment.

The solution was validated through a 29-day controlled experiment across two academic programs. The group optimized with Mixilo AI Conversions outperformed the control group on every indicator: 41% more interested people and 7% more enrollments. More people took the step toward the next stage of their professional lives.

Success Story: Continuing Education at Tecnológico de Monterrey

The Continuing Education division at Tecnológico de Monterrey drives the development of individuals and organizations through diploma programs, courses, and Executive Education programs designed for every stage of professional life. It takes Tecnológico de Monterrey's academic excellence and innovative spirit beyond the classroom, so that each person can reach their potential and make an impact in their environment.

The Challenge

In Continuing Education, choosing a program is a carefully considered decision. Each person ties it to a goal (specializing, taking on a new challenge, changing direction), compares options, and takes their time. Weeks or even months can pass between first contact and enrollment. That's why a completed form is only the beginning of the story. What really matters is that each person finds the program that will take them where they want to go and decides to start it.

The Continuing Education division at Tecnológico de Monterrey has a broad and diverse portfolio, in which each program serves different profiles and purposes. This created an opportunity: traditional campaign optimization gives every lead the same weight, while a model capable of recognizing the fit between each person and each program could bring each individual the right offering.

The question that guided the project: how can more people find and choose the ideal program for their professional moment, while maintaining campaign reach?

The Solution: Mixilo AI Conversions

Digital platforms offer their own value-based optimization, such as Value-Based Bidding on Meta and Google, which learns from the events each organization configures. In this case, that event was the form fill.

A completed form is an initial signal of interest, but it doesn't always predict an enrollment. When platform leads were compared against those confirmed in the CRM, a considerable gap emerged. The challenge, then, was to get campaigns to learn from the decisive moment: when someone chooses to start a program.

Muttdata proposed a different path: Mixilo AI Conversions. The solution works in three steps:

1. Individual Analysis. The model examines each new lead based on its source campaign, age, geographic location, and lead score. The lead score reflects each person's level of interest and combines signals such as their profile, their engagement with content (email opens, site visits, replies), and the decision stage they're in. In this case, the score came from Continuing Education's CRM, (when an organization doesn't have one, Muttdata can also build it.)
2. Probability Assignment. Based on these characteristics, the model estimates the probability that each person will enroll.
3. Expected Enrollment. The sum of the day's probabilities becomes the new metric for guiding campaigns.

Why predict on leads?

The model predicts on leads (not enrollments) because the actual volume of enrollments is usually too low to train a statistically valid model. Leads, on the other hand, make up a significantly larger sample and serve as a high-volume proxy for learning which characteristics predict an enrollment.

The core difference from other solutions is this: Mixilo AI Conversions is not a layer of rules on top of the platforms' native optimization. It is a proprietary model, trained and validated against the client's actual business outcome.

Implementation

Continuing Education and Muttdata designed the implementation as a controlled experiment with a clear purpose: to isolate the model's effect from any seasonal or channel variation and obtain solid evidence for decision-making.

The approach brings to digital campaigns the same rigor with which Tecnológico de Monterrey generates knowledge: a hypothesis, careful measurement, and well-founded conclusions.

• A/B test. A test group optimized with Mixilo AI Conversions, and a control group optimized for lead volume.
• 29 days on Meta and Google. Attribution was measured against Continuing Education's CRM, not against platform-reported conversions.
• Continuous validation. The model's predictions were compared with actual results over an attribution window defined from enrollment history, which made it possible to measure accuracy in near real time.

Key Results

The model's accuracy was confirmed against CRM data.

The group optimized with Mixilo AI Conversions grew at every stage of the decision process, while the control group added leads but kept its enrollments at levels similar to its previous period.

‍

‍

Note: "Test vs. its previous period" and "Control vs. its previous period" compare each group against its own historical baseline. "Test vs. Control" compares both groups during the experiment. Because they use different baselines, the figures cannot be directly added together.

The most relevant signal is the trend: compared with its own previous period, the test group kept up its progress all the way from first interest to enrollment.

‍

Mixilo's Impact

• Each organization defines its own success. Mixilo, Muttdata's platform, optimizes toward the goal each institution considers a priority (enrollments, appointments, or activations), going beyond the metrics platforms measure by default.
• A model that gets validated. Predictions are regularly checked against actual results at the campaign level. This makes every performance claim traceable, with a level of detail that goes beyond aggregate averages.
• It adapts to complex institutions. The same service ran two experiments in parallel, with different programs, channels, and decision timelines. This shows its flexibility for institutions with multiple units and audiences.

‍

"At Continuing Education, we measure our campaigns both by the interest they generate and by the people who decide to enroll. Muttdata built the optimization around what success means to us, rather than what the platform measures by default, and supported us very closely with valuable insights and recommendations.” – Demetrio Morales, Paid Media Manager, Continuing Education at Tecnológico de Monterrey

‍

Technology in Service of Learning

For the Continuing Education division at Tecnológico de Monterrey, innovation is meaningful when it brings each person closer to the learning opportunity that drives their aspirations. This project shows how artificial intelligence, guided by clear goals and validated with evidence, helps each program reach the people who find their next step in it: people who want to grow, transform their career path, and make a positive impact on their organizations and communities.

‍