Creating and Implementing a Connection Score


By Marisa Ontko, Technical Analyst at Bentz Whaley Flessner, and Lindsey Nadeau, former Director of Research and Relationship Management at George Washington University 

In May 2017, George Washington University (GW) completed their one billion dollar fundraising campaign. After remarkable success in this campaign, a question arose: “What is the giving potential of our database?” BWF, a premier fundraising consulting firm, engaged in a project with GW to answer this business question. One component of the engagement was the development of a connection score. In conjunction with capacity estimates, the connection score allowed GW to focus portfolio optimization efforts on roughly 8,000 high potential prospects.

What Is a Connection Score?

A connection score is a rule-based model that quantifies the affinity constituents have for an organization. This score is created by applying a ranking system to variables from the database, with points applied to constituents for having specific activities, attributes or giving variables tracked on their record. When all the selected variables used in a score have been assigned point values, an overall score is generated for each constituent. Scored constituents can be segmented into groups based on these scores in order to better understand the connection of all constituents and tailor philanthropic strategy accordingly.

Applying a connection score to an entire population to highlight the most engaged constituents is a powerful tool, especially when applied in tandem with capacity estimates. When partnering with GW, the goal of the connection score was to assess the potential of the entire database and to begin the identification and prioritization of constituents in the prospect assignment process. The customized connection score for GW was an efficient way to identify the 15,000 most-connected constituents with high capacity from a database with roughly 560,000 records. Even this top 2% of high-capacity, highly-connected constituents contained many who had never been assigned or reviewed. These records were identified and then considered for portfolio optimization by GW.

Variables to Consider

When creating a connection score, there are a number of variables to consider. When available, BWF reviews both giving and involvement, or non-giving, data points. Utilizing these two types of data allows for all constituents to be scored on the same score, rather than prioritizing donors and non-donors separately. For example, many constituents show their affinity by attending an event and making a donation. Other constituents have attended galas, signed up for emails, and provided a phone number and address, but not given — yet. Creating a score where events, phone, address, and giving history are awarded points aids in identifying connected prospects.

When choosing variables related to giving history for a connection score, it is important to select variables that are relevant to both major donors, base donors, and non-donors, in order to ensure the entire population is reviewed. This can be done by using overall giving, planned giving, planned giving club membership, consecutive giving, and giving amounts around the major gift threshold. When choosing involvement activities for a connection score, consider variables that focus on the natural constituency of the organization, board members, and community members. A few variables that are commonly included in connection scores are participation in a board or committee, event attendance, volunteering, reunion participation, and different methods a constituent can be contacted.

When creating the GW score, BWF considered over 70 variables by reviewing the distribution and clusters of the population. Many of the variables used had a long history of being tracked consistently in the database, which is very helpful to the accuracy and precision of the finished score. Recently tracked social media and reunion attendance variables were included in the score because a specific percentage of constituents had this information tracked on their records. After the variables were selected for the score, each variable was assigned a numeric value. Weighting the scores allowed for specific variables to be assigned more points based on level of giving or level of involvement.

After each variable was assigned a numeric value, BWF added all the scores together for each constituent to create a composite score. BWF then created five groups of constituents based on percentile ranges. The connection score point ranges were converted to descriptive segments: very connected, connected, somewhat connected, modestly connected or not connected. To highlight how effective a connection score can be at identifying high potential prospects, very connected and connected segments made up 6% of the database, or about 33,000 of GW’s constituents. Focusing on the top 33,000 constituents, BWF then paired the connection score with capacity estimates. The strike zone, constituents with high levels of connection and capacity at the major gift threshold, consisted of 15,000 constituents. These top records revealed that roughly 8,000 high potential prospects were not assigned and ready for prioritization.

For a full list of the variables used in the GW score and possibly points attributed, please click here. 

To open the quiz in a separate window, click here

Looking to experiment with your own connection or engagement score? Get your feet wet with examples from Data Science for Fundraising: Build Data-Driven Solutions Using R from Apra’s very own Ashutosh R. Nandeshwar and Rodger Devine.

Rightsizing and Goal Setting

The most important part of working with an outside partner is the internal rightsizing of recommendations. At GW, the thought of optimizing prospect assignments for 8,000 unassigned high-capacity, high-connection (HCHC) strike zone prospects was mind boggling.

Institutionally, GW has a limit of 120 prospects per gift officer portfolio and specific guidelines vary from about 75-120 prospects, depending on management responsibilities, tenure at GW and fundraising area. Existing gift officers could manage approximately 4,000 prospects. Separate from the unassigned 8,000 HCHC prospects, GW already had HCHC assigned within portfolios, who shouldn’t be optimized out. So could be done with 8,000 prospects?

GW tightened its focus on the best-of-the-best strike zone prospects. GW overlaid its institutional knowledge of its record structure to account for house-holding (coupling spouses and closely related constituents), remove disqualified and recently deceased prospects, and implement a preference hierarchy that favored gift officer rating first and then research ratings, research qualifications (quick capacity estimates), and finally, unverified vendor ratings. This narrowed the strike zone to 2,700. GW applied the same right-sizing to its low-capacity, low-connection (LCLC) segment and identified the 650 prospects who should likely be optimized out of portfolios. Two goals emerged to guide the team’s portfolio optimization work:

1) Assign the 2,700 unassigned HCH prospects

2) Unassign the 650 LCLC prospects from portfolios, where appropriate

Training and implementation

GW conducted a workshop for the prospect development team on the composition of the score and how to integrate it in portfolio reviews with gift officers. A consultant also led a training on persuasive conversations with gift officers. For our bi-weekly, team-wide dedicated prospecting day, each analyst began their prospecting from the list of 2,700 unassigned HCHC prospects. The division’s intranet shared background information about the score, our goals and provided progress updates on the division’s success as prospect development began optimizing assignments via quarterly portfolio reviews and ongoing prospecting.

One of the most important training tactics used was underscoring the importance of meaningful conversations with our gift officers before optimizing portfolio assignments. Gift officers often know more than a score about a prospect’s wealth, so using the score as a starting point of a conversation was key to GW’s success.


When we compare outcomes from the 2,7000 unassigned HCHP prospects with outcomes from the 650 assigned LCLC prospects, the contrast is stark.


Additionally, three major gifts yielded from the list of 650 assigned LCLC prospects. Because of the strong partnership between GW’s prospect development team and their gift officers, these three donors were never unassigned from the gift officer’s portfolios.

Where Does GW Go From Here?

GW again called on BWF to develop a custom alumni engagement score, delivered to GW in Spring of 2019. GW continues optimization work using this new score. To better track the velocity of scored constituents, GW hopes to develop a canned report to measure attainment and pipeline development activity more easily and isolate significant outcomes across different segments.


Apra BoK Data Analytics icon.pngThis article relates to the Data Science domain in the Apra Body of Knowledge.

Apra University has more resources for measuring connection. Check out this webinar, "Measures of    Affection: Developing & Applying a Data-Driven Engagement Index."


Recent Stories
Apra Connections Chapter Capture — Q2 2022

Prospect Identification as a Tipping Point to Hire for Research

Willkommen, Bienvenue, Welcome, C'mon In: Onboarding for Success and Retention