Manual matching delays opportunities
Big Brothers Big Sisters of Puget Sound (BBBSPS) is dedicated to helping every child unlock their potential through impactful one-to-one mentoring. Traditionally, the organization faced challenges with a manual, time-consuming matching process—relying on memory and intuition, leading to delayed connections. With matches taking an average of more than three months and a waitlist that never dipped below 500 children, BBBSPS needed a tool that would make the entire process more efficient.
The AI matching recommendation engine: A solution built on data
Today, through the power of artificial intelligence, AIMRE (AI matching recommendation engine) is transforming mentorship—powering improvements that can help make every match more meaningful, every journey more hopeful, and every future brighter. Developed by KPMG on the Microsoft Fabric platform, AIMRE significantly reduced matching time, based on BBBSPS’s pilot results. It improved the mentor-mentee matching process by providing data-driven recommendations while maintaining the crucial role of human experts.
AIMRE works by ingesting and analyzing data collected during the Big Brothers Big Sisters application and interview process. The data contains a number of attributes across demographics, preferences, and fields of interest. Using an advanced algorithm, AIMRE identifies and recommends potential matches between a Big and a Little. The Matching Specialist then reviews those suggestions, and based on their analysis, decides whether or not to make a match.
With the flick of a switch, AIMRE increased BBBSPS’s speed, accuracy, and capacity. While most BBBS agencies are forced to add staff when growing their matching capabilities, AIMRE helped BBBSPS scale capacity without hiring more.
Finding the facts in successful matches
The data AIMRE gathered also dispelled some previously held assumptions and replaced them with data-based decisions. It had long been assumed that shared demographics were keys to successful matches, but our analysis showed that shared interests and life experiences were far more important. While closeness in age was thought to be a success factor, for example, data showed just the opposite: Littles did better when Bigs were significantly older; and had worldly experience to help guide the mentee through career and life choices.
KPMG and Microsoft: Collaborating to support improved matching processes for nonprofits
The alliance partnership between KPMG and Microsoft was instrumental in developing AIMRE, and the collaboration highlighted the potential for AI to solve complex problems in the nonprofit sector and beyond. Microsoft offered advanced AI and cloud capabilities, while KPMG brought deep expertise in business transformation, data strategy, governance, and implementation. With their complementary strengths and their long tradition of community involvement, they worked together and:
- Leveraged GenAI to analyze natural language from interview notes and extract relevant themes
- Employed machine learning to evaluate these themes and rank potential matches
- Implemented a scoring system that provides data-driven recommendations to Matching Specialists
- Ensured that Matching Specialists have the final decision-making power.
The project was a welcome opportunity for KPMG to leverage its resources for public good. Both KPMG and Microsoft are longtime supporters of BBBSPS, aligned with its commitment to supporting young people facing challenging circumstances, such as economic hardship or housing instability.
Idea to implementation in 16 weeks
From the early thinking stages through whiteboarding sessions, meetings with BBBSPS Matching Specialists, building the solution, and the program’s pilot launch, KPMG, Microsoft, and BBBSPS worked as one team, coalescing around a common goal—helping children achieve a better life. Microsoft, Microsoft Fabric, and Microsoft Azure provided the unseen digital infrastructure for machine learning models that help predict compatibility and long-term success. During this process, KPMG:
- Analyzed more than 10 years of anonymized historical match data
- Developed predictive scoring to surface top potential matches per mentee
- Built a user-friendly user interface and dashboard with ongoing input from BBBSPS
- Delivered strong change management assistance to enhance adoption.
Empowering the Specialists
In addition to speeding up the process, AIMRE has also made life easier for Matching Specialists, the linchpin of the program. Before AIMRE, Specialists had to sift through piles of questionnaires, weigh the dozens of factors that influence a match, and then make recommendations based solely on memory and pages of notes.
The fear of making a wrong choice often led Matching Specialists to adopt an overly risk-averse view of potential matches. Being human, they were also subject to subtle biases—the most common being “recency,” the preference for those they’d just seen over those who had presented earlier. AIMRE has addressed these issues in the following ways:
- All recommendations are supported by data rather than tradition
- Memorization is no longer required; all information is catalog and instantly searchable.
- With the benefit of AIMRE’s recommendations, Matching Specialists have more confidence and make more matches.
- Matching Specialists believe AIMRE enhances, not replaces, their judgment.
With AIMRE, Matching Specialists can greatly increase their efficiency. They are making more—and more successful—matches; and today BBBSPS, one of the largest chapters in the United States, supports more than 1,000 active mentoring relationships.
AIMRE also means less administrative burden for staff and volunteers, and more time to build strong, lasting relationships. For mentors, it reduces frustration and keeps them engaged. And for kids, it means a better chance of being paired with the right role model, sooner.
