Tackling the under-matching problem and personalizing college counseling with machine learning
Sign up for Student Union

The Problem


Each year, more than 400,000 qualified low-income students don’t attend college, and another 200,000 are significantly undermatched. Undermatched students are those who should attend more rigorous schools based on their qualifications. Instead, undermatched students attend colleges that limit their chances for success, which affects their long-term earnings and social mobility.

How we help


First-generation college applicants are often under-supported when it comes to the difficult decision of choosing a school.

First-generation college students are often under-supported when it comes to the difficult decision of choosing a school.

They may make decisions based on criteria such as ease of admission or where their friends are going. This leads to students matriculating at schools that often increase their likelihood of dropping out. Uptake gives low-income and first-generation college applicants the power of informed choice. Students can find schools that’ll help them reach their full potential and provide the support they need to graduate.

Student Union

Students enter their GPA, test scores and demographic information.

Student Union

Student Union finds school choices that are the best matches based on school graduation rates.

Student Union

Students can filter by preferences such as size, location and diversity.

Advantages of Student Union

Based on data

Many matching tools rely on rules-based models to place students into selectivity categories. Our algorithms analyze correlations found in historical data from first-generation college students to generate individualized predictions for each student’s best-fit schools.


Tailored to each student

We focus on first-generation applicants and students from low-income areas. Colleges are suggested based on where students with similar profiles have been admitted and have succeeded in graduating.


Improves continuously

Our models continuously improve through machine learning. The more data our models ingest, the more patterns and nuances they learn to recognize. The more that students use it, the better it becomes at predicting success.


Shares knowledge

Because the application is used by multiple organizations, we can predict acceptance, persistence and graduation better than any one organization can by itself.


Partner & scale

Current Impact

  • Focused deployment in three regions: Chicago, Southern Texas and Southern California

2018 Goals

  • Create the largest data set on first-generation college students’ success in the United States.
  • Introduce Student Union more broadly in Chicago, Southern Texas and Southern California.
  • Find data and implementation partners in the Bay Area, Michigan, St. Louis and New York.

To help every first-generation college student in America, we need a wealth of partners to adopt and distribute Student Union to their stakeholders.

Become a Partner

We’re seeking additional partners, including:

  • Nonprofits with a mission of helping first-generation college students graduate from college.
  • Philanthropic foundations with missions to uplift first-generation college students.
  • School networks that can spread awareness to high schoolers.
  • Examination organizations that reach millions of students.
  • Other tech companies that want to help support this project.
  • Government institutions that can contribute data or help scale the application.
  • Research institutions with extensive knowledge that want to support the project.