The recent announcement of Professor Richard Thaler’s Nobel Prize in Economics traveled quickly through University of Chicago (UChicago) digital alumni channels. A chorus of congratulations sounded from e-newsletters, Instagram and Facebook posts. Magnifying the moment, short video clips showed the popular professor recounting details from the 4 a.m. award-winning call from the Nobel committee.
Thaler’s Nobel announcement casts light on engagement in the digital realm. A growing ecosystem of email tools, social sites and online alumni professional communities are fertile ground for building social relationships, lifelong learning and professional networks.
Digital touchpoints ― clicks, tweets and comments — comprise a category of alumni and donor engagement called “Connect.” Connect builds upon the “Go, Give, Help” paradigm of engagement that emphasizes event, volunteer and philanthropic engagement.
To meet alums in the digital realm — and measure connection ― we must develop systems that capture digital data streams. Gathering digital data is often discussed, but not widely adopted when setting engagement goals because of the challenge of collecting data, setting up metrics and linking the data to individual alumni profiles.
Overcoming the data storage challenge is only the first step. A digital engagement strategy that considers the data, systems and emerging analytics is vital to tracking progress, measuring campaign success and gaining insights from the data. Digital engagement metrics can support better communication and engagement strategies and also offer a window into philanthropic interests — in turn, supporting an alumni experience that keeps pace with alumni interests and career milestones, meeting them where they are.
Background: Rich Data, Disconnected
We are living in an era of digital engagement, enhanced user experience and, of course, data.
More than 168 million emails are sent every 60 seconds worldwide . Last year at the University of Chicago (UChicago), more than 10 million clicks, opens and unsubscribes accumulated from emails sent to nearly 200,000 alumni and friends. And in the social realm, tweets, likes, shares and comments pile up as communications teams use Instagram, Facebook and Twitter to tell stories, share news and fundraise.
Outside higher education, our alums are increasingly accustomed to personalized content and well-timed messages, as responsive marketing experts nimbly digest and strategize based on clicks, preferences and shares. We have a finite pool of constituents, and they are more than customers — they are alumni. As such, they expect that same excellence throughout the alumni experience as from the student experience. Alumni have come to expect that we keep track of them over time: from addresses and degrees to events, digital touchpoints and interests — and offer aligned programming, content and giving opportunities.
Universities are challenged to meet alumni expectations for connection and quality within the context of modern online experiences. UChicago, like others, is grappling with what to do with — and how to make sense of — an ever-higher mountain of data in the digital realm, which has been disconnected from our traditional databases, reporting and analysis.
Our focus on digital engagement was spurred by an ambitious engagement goal. In 2014, UChicago launched its Inquiry and Impact campaign, with the dual goals of raising $5 billion and engaging 125,000 alums (over 80 percent of our alumni base) over a five-year period.
Tracking these new forms of digital engagement efficiently, charting engagement progress and understanding alumni social media behaviors became a challenge. The alumni engagement goal needed a digital engagement strategy that covered new communication channels and outreach to alumni in new cyber spaces — and with it a more sophisticated data platform.
The Big Data: Digital Engagement
First, a word about the data. To address challenges discussed above, a digital engagement strategy was proposed that incorporated diverse forms of data, including emails, social media, online learning and career services, webinars, and websites used to engage alumni and collect data. Collectively, this was our “big data.” Even though the amount of data we captured was still relatively small, its growing size, complexity, structure and variety prompted us to look beyond our traditional data warehouse for storage.
It is well worth mentioning that prior to gathering and using new data, we developed strong polices to maintain the confidentiality, privacy and security of alums; these are a work in progress as we adapt to new technologies, data and legal frameworks.
Email marketing remains a vital tactic because it is easy, effective and inexpensive (in production cost). Through emails, UChicago reaches more than 200,000 alumni, parents and friends annually. Newer email marketing tools, like Marketo, make the automation of testing, content curation and data collection easier for small shops and nonprofits.
In recent years, outreach and marketing in higher education have significantly expanded to social media platforms, including Twitter, Facebook, YouTube and Instagram. Like many peers, UChicago uses social media accounts to solicit alumni, promote events and share news.
Online Learning and Career Services
Higher education institutions also use social learning platforms that offer continuing education services and student career services to promote alumni engagement.
For example, UChicago’s AlumniU learning platform allows alumni to participate in exclusive online courses, engage in discussions with faculty, find curated learning content and join interest groups to connect with fellow alumni.
Likewise, many schools utilize online career mentorship and networking applications. At UChicago, we use Wisr (pronounced “Wiser), an online platform that makes it easy for alumni to connect with current students. This tool captures alumni-student relationships, career and biographic information.
Cultivating and positioning digital engagement platforms is crucial for UChicago to present itself to both prospective students and alumni, but we have often been in the dark when it comes to assessing activity and performance in a meaningful way.
Big Data Analytics: Transforming Traditional Metrics
As a starting point, we thought about the questions we wanted to answer with our new data. Working backward, we mapped our needs onto the data and tools at hand. Here is a sampling from our email data brainstorms:
Alumni Relations and Annual Giving
- Who are our most engaged alums online?
- How can we better target audiences for major events?
- What is the correlation between email clicking behavior and giving behavior?
- How does tweet/comment sentiment vary by generation?
- What are the send volume, rates for opens, clicks and unsubscribes:
○ by time of day?
○ for solicitation emails versus all other emails?
○ by recipient age? Other major demographic categories?
- How much email do specific groups receive throughout the year? Is it causing fatigue?
- What email chain activities precipitates unsubscribes?
- What does online engagement tell us about interests as we prospect?
- Do digital engagement metrics help uncover any rising stars?
- How does digital stewardship fit in?
Google Analytics and Facebook dashboard metrics are a helpful start in answering these questions. They chart post performance, single campaigns and likely visitors. With these, we are able to track and measure digital engagement, which are essential to providing insights about campaign performance.
However, the simple Google Analytics dashboard measures offer little analytical flexibility, and serve up social and email performance in a vacuum, disconnected from giving activities and other forms of engagement.
To gather insights from our data for prospecting, alumni engagement and communications, we needed to extend beyond each tool’s dashboard metrics and enter the realm of “big data analytics” ― that is, connect and analyze our granular digital data with our alumni data.
The table below sheds light on the differences .
Big Data Tools: Unified Data Architecture for Analytics
To integrate our digital data, UChicago chose to build a big data platform that augments our existing data warehouse. Our architecture is a traditional pathway for housing big data, but there are others emerging. As handling larger volumes of data becomes the norm, there will likely be an increasing number of options for storing and processing big data.
We utilize Cloudera Hadoop as our platform of choice. It is a framework for storing and processing big data. The platform is highly scalable and designed to process very large datasets.
Below is a sketch of our environment. The bottom layer shows the data sources feeding in; the middle “Hadoop” layer is responsible for data storage, processing and integration; and the top layer describes the analytics, reporting, and visualization avenues and applications.
For starters, we have focused on integrating email and social activity. Some of these data come in the form of text files/Excel files (from legacy systems) and others from APIs (that’s streaming data that flows in from sources like Twitter or e-marketing applications).
Within the Hadoop ecosystem (Hadoop, MapReduce, HDFS, YARN, Impala, HBase, Spark), several tools work in conjunction to create an environment that can store and process both structured and unstructured data efficiently. A variety of tools are useful for querying data (Python, SQL, Hive and Pig).
For analysis and visualization, we are currently using R for modelling and exploring Tableau for data visualization. With these tools, analysts or scientists can explore text mining and a variety of advanced analytics and visualize results. Some of these methods are still developing; measuring sentiment within Tweets and comments is a challenge.
Big Data Analytics: In Action
With our Hadoop platform in place, we turned to analysis and answering our questions. Analytics can have a wide variety of meanings. In the examples below we reference a few different types of analysis, from basic descriptive statistics — that is, what happened? — to prescriptive stats ― what should we do? — to illustrate the impacts that digital engagement strategy brought to University businesses, ranging from engagement to development efforts.
Alumni Relations and Annual Giving
At the core of our alumni work is engagement. Adopting a digital engagement strategy during 2017, we were able to round out our “Connect” bucket — and now count MOOCs, Tweets, clicks, AlumniU and other social directory activity as points of engagement. Last year’s focus was measuring breadth of engagement and securing data. This year we pivot to depth of digital engagement and taking stock of what’s most meaningful.
During the seven-month period we analyzed, more than nine million emails activities were generated among more than 120,000 alumni. There were two million opens and 191,000 clicks. The overall open rate is 23 percent and the click-through rate is approximately 2 percent, but combining the data in new ways, we are digging beyond click rates. A few of our initial investigations centered on understanding volume and how giving relates to communications.Because of the volume of the data and the rigidness of our current database system, we were unable to explore these patterns in the past. The Hadoop platform gave us the flexibility we were lacking.
Are we sending too many emails? We sought to understand volume and its relationship to performance. The calendar and fiscal year end are among the busiest times of the year in terms of email deployment, but — like many would suspect — email engagement proportionally rise with emails sent. Next, we turned to assessing how many emails per month certain audiences receive. A deeper dive here revealed some nuances: Sure, some audiences are saturated (receiving an email per day), but others receive relatively little mail (less than one per week).
How does engagement vary by day of the week? Looking closely at email activity by day of week, we observe that while many emails were sent on Tuesday, Saturday and Sundays have higher significantly click and open rates.
Which campaigns have the highest gift conversion rate? Attaching appeal codes to campaign emails and gifts, we are able to track the gift life cycle as well as measure the effectiveness of email campaign in terms of gifts (see chart below: Email Performance).
How can we better target event invites? Clustering techniques are used to identify segments of event email recipients based on alumni age, school, event attendance, giving and email engagement behaviors. We’re exploring new ways to target content that appeals to each segment. The following are four groups of recent graduates who are engaged with event emails: termed Loyal MBAs, Proud Alums, Active Networkers and Engaged Grads (see chart below: Cluster Analysis).
Beyond engagement and communications, we’re exploring the integration of digital data into prospecting and research, using the click data to reveal interests. We began by looking at the gift planning pool and assessing engagement with Building for the Future, a gift planning newsletter.
The previous gift planning model contained traditional variables, including: age, years of consecutive giving, marital status, number of children, etc. The idea of including the email data stemmed from an analysis requested by the Office of Gift Planning (OGP) in conjunction with the Communications team, in hopes of getting a better understanding of who was clicking through their newsletters. By clicking through the newsletters, we assumed the recipient had demonstrated an interest in planned giving — the more email activity, the higher the potential interest.
With this in mind, we developed the Gift Planning Email Score. Alumni were given points for opens and clicks, aggregating scores across all Gift Planning emails that were available in Hadoop. While certain usual suspects rose to the top, such as those in our gift planning society, the Phoenix Society ― there were others who had not yet made a planned gift but were active (see chart below, GP Email Score).
After the score was completed, we were curious to see if it proved significant in predicting gift planning prospects. It turned out to be a strong factor. Combined with other traditional variables that are useful in predicting the likelihood of making a planned gift, the email score proved to be a significant predictor. Introducing click data contributed to the generation of more than 600 new individuals for further consideration by our prospect research and prospect management teams.
The alumni experience is increasingly digital: Robust peer-to-peer connections and online professional networks create opportunities for lifelong engagement with the university community. Next-generation systems and metrics must enter the digital realm. Advancement teams should consider big data storage, strategy and metrics, while staying abreast of increased regulation and security concerns around data privacy.
At UChicago, Hadoop created a bridge between traditional and digital engagement by integrating data sources within a powerful platform. With it, we can enhance the digital engagement experience within the alumni and donor lifecycle — and continue to meet our alums, wherever they may be.
 Fundraising Fundamentals, Section 1.2 http://www.case.org/Publications_and_Products/Fundraising_Fundamentals_Intro/Fundraising_Fundamentals_section_1/Fundraising_Fundamentals_section_12.html
Mirabai Auer is the director of analytics and reporting at the University of Chicago. She leads a team of analysts and architects to help grow fundraising and engagement using data, technology and reporting.
Xiaohong Zhang is a data architect on the Analytics and Business Intelligence team at University of Chicago. She is an analytics and IT professional focused on developing the platforms and tools that enable data-savvy teams.
Looking to delve further into data analytics? Check out the Apra Data Analytics Bundle 1 for five educational recordings that will help you dive into analytics, with titles such as "Competing in an Analytical Environment," "Database Mining" and "Thinking About Data Modeling? Five Key Questions to Consider before Starting."