By: Elizabeth Martin, Phil Nadeau and Katharine Bierce
As Artificial Intelligence (AI) is increasingly used across sales, marketing, and customer experience departments in the for-profit sector, the question becomes how nonprofits and educational institutions can take advantage of these same tools. While AI can seem like a remote concept, tools are already or soon becoming available to make teams more effective across fundraising, programs, student success, and engagement. From built-in tools that score donor and student activity, to advanced analytics platforms that allow for data-scientist-like discovery, this blog post will arm you with a high-level overview of the tools available.
Ways Nonprofits are Using Artificial Intelligence
One easy way to use AI in your nonprofit is to look for tasks that you have in common with many other nonprofits. Here are a few examples of ways to start out, with examples of how Salesforce enables them.
Nonprofit Fundraising: Increasing major giving
A challenge many fundraising teams face on the major giving side is identifying the right individuals to cultivate as potential major donors. As nonprofits are often strapped for resources and time, targeting the right individuals is critical. Even in conjunction with wealth screening tools, it can be hard to understand which individuals will have a strong enough affinity to an organization to make a major gift. Consulting services are available that combine data on your constituents’ giving history with externally-available data from wealth screening tools to provide predictive insights, but these services can be costly and take weeks to complete.
Imagine if you could have these tools built in to your organizations’ CRM and run the analysis with a click of a button. Going forward, tools such as Einstein Prediction Builder will allow just that.
Einstein Prediction Builder will allow admins to create custom AI models on any custom Salesforce field or object to predict outcomes. In the case of major giving, field values on the Contact object can aid in predicting a potential donor’s likelihood to major gift. Models can be created based on all Salesforce fields related to the Contact, such as their giving and volunteering history, event attendance, affiliations, relationships, and even data from wealth screening tools. Using a declarative, point-and-click setup tool, admins can define what they’d like to predict and the fields they’d like to use for the model. The results can be saved on the Contact object and can then be used in filters, reports and list views, as well as direct mail and marketing campaign segmentation.
Built-in predictive insights will help major gift officers better understand how to focus their time by providing additional color into what factors contribute to a potential major gift and which individuals to focus on. These built-in tools will enable fundraising teams to be more effective in obtaining resources to advance a nonprofit’s mission, without having to invest in expensive consulting services or specialized data analysis software.
Fundraising is time-consuming, and you want to help everyone feel special – whether they’re donating $10 or $10 million. Often, however, email blasts and communications aren’t personalized, and donors disengage, resulting in lost opportunities to engage them (and raise more money).
What if you knew that some of your constituents were unlikely to open an email, but would positively respond to a text or a Facebook advertisement? What if you could then segment constituents based on these characteristics, placing them on different journeys based on these predictive insights?
Marketing Cloud Einstein allows just that. It uses machine learning to analyze patterns and predict the actions that individuals will take, such as open, click through, or unsubscribe when sent an email. The results can then be used by marketers to build smarter segments and journeys. A constituent unlikely to open an email can be sent a text, as an example, if this would result in a higher likelihood of engagement.
By automatically optimizing segmentation, tools like Marketing Cloud or Gravyty First Draft can save marketers at nonprofits and educational institutions time, while simultaneously resulting in more successful campaigns.
For organizations looking to improve volunteer engagement, AI can offer useful solutions as well. One way to improve volunteer engagement is to use AI to determine what kind of activities each constituent prefers.
One tool, Summery.ai, uses an entertaining interactive questionnaire to determine a user’s Giving Profile, which characterizes their preferred philanthropic activity. There are ten personality profiles, from the enthusiastic Pathfinder to the individualistic Pioneer. Their typology is based on a study by experts in psychology over many respondents in a typical corporation.
It’s also possible to use collaborative filtering to drive volunteer engagement. This type of filtering is ubiquitous in e-commerce, where a learning algorithm aggregates the shopping behavior of many users to predict future purchases. In other words, if Alice buys a flashlight and then buys a pack of batteries, then if Bob buys the same flashlight, he’s likely to buy the same pack of batteries. This article from the IEEE reviews the history of collaborative filtering at Amazon.com.
Salesforce.org is developing Philanthropy Cloud in partnership with United Way for better volunteer engagement and more. Get a sneak preview in this Dreamforce 2017 keynote, around minute 50. You can also read more in this announcement or check out these CSR tips from some of the product managers of Philanthropy Cloud.
In addition to enabling efficiency across fundraising, donor engagement, and volunteering, predictive analytics tools can also enable programs teams to be more effective. Understanding the factors that impact program outcomes can aid program managers in providing recommendations, arming them with powerful tools to influence results.
Social services agencies and universities alike have already been using machine learning to find at-risk client or students and prioritize them for services or additional help. This technique has proven useful across many different non-profit sectors:
- Medicine: The paper “Machine Learning for Social Services: A Study of Prenatal Case Management in Illinois” describes an experiment where machine learning prioritized high-risk pregnancies for treatment and achieved a 36% improvement over the paper process.
- Employment: Bayes Impact has built a system to aid job seekers in Paris through the use of interactive questionnaires.
- Education: CollegeForward uses machine learning to identify coaching methods to prepare at-risk high school students for college.
Previously, extensive resources would be required to analyze program data and identify correlations. For some organizations, the funding to hire data analysts and license the software for this purpose simply wasn’t available.
Predictive analytics tools built into a nonprofit’s CRM can fill this gap. Tools such as Salesforce Einstein Discovery can analyze billions of data combinations in minutes to surface predictive insights and provide recommendations, without the use of additional software or statistical models managed outside your CRM.
Einstein Discovery conducts statistical checks to confirm the models are valid, and is able to generate answers, explanations, and recommendations in a way that is easy for business users to understand, without having a data scientist on staff. Watch a demo of how Einstein Discovery can be used to identify trends and improve outcomes for drug rehabilitation programs.
Beyond the Basics: Next-Level Artificial Intelligence for Nonprofits
Once you’ve integrated the use of across fundraising, engagement, volunteering, and programs, here are some more advanced use cases that nonprofits are considering.
Intelligent Case Management
When using web-to-case or email-to-case, it can sometimes be time-consuming to read and sort through the comments in individual cases. This is another area that AI can help improve efficiency. Using machine learning, Service Cloud Einstein can automatically classify and escalate cases as appropriate. Additionally, knowledge articles which may be relevant for agents working on the case can be automatically surfaced, saving agents time and allowing for an improved constituent experience.
For cases coming in via live channels, chatbots can be used to answer routine questions and gather information to appropriately route cases. Chatbots are now commonly used across the commercial sector for eCommerce and customer service. We’ve previously mentioned Raheem.ai, a chatbot for anonymously reporting on interactions with police, and have seen several projects that use chatbots to establish contact with clients and solicit initial information. You may have heard about Adidas’ use of chatbots with Salesforce at Dreamforce; we hope to see this expand across the nonprofit sector as well.
Tools such as these do not only improve agent productivity, they allow for better experiences on both the employee and constituent sides.
Intelligent Social listening tools
As constituents increasingly move to social media channels to connect and share with others, social listening and social customer service become increasingly important to better understand and engage with constituents. Additionally, as social media has become a more visually-driven medium, brand identification in images has become more important. Previously, marketers would have to manually sort through social posts to review language and images to gain insight into their constituents’ intent, identify influencers, and classify posts for follow-up. Artificial intelligence tools built into social listening platforms now automate these processes.
With Einstein Vision for Social Studio, marketers can uncover images relevant to their brand through the use of image classification tools, allowing for automatic tagging of posts with an image of their brand’s logo, even when their name isn’t mentioned. Einstein Social Insights also automatically classifies post sentiment and intent using natural language processing, allowing for automated classification, and helps to identify influencers to better prioritize posts for follow-up.
Building Your Own
Many of the most common use cases are currently (or will be) supported by tools built directly into your CRM. If you have more advanced or non-standard use cases which may not be supported by out of the box features however, Salesforce offers several resources to support custom use cases. From image recognition and object detection to sentiment and intent, myEinstein offers a number of APIs and tools built into the Lightning platform to support custom applications. You can learn more about the Einstein Intent and image classification APIs on Trailhead, the fun way to learn Salesforce.
Need pro bono help with data science? If you’re using Salesforce, check out our Pro Bono Program that you can apply for to get employee help on Salesforce for your social good organization. DataKind, Delta Analytics and Bayes Impact also provide free data science volunteers for nonprofits.
In closing, we’d like to remind you that beneficial AI is only going to happen if people like you demand it. Whether you plan on implementing AI at your organization or not, you can help by encouraging other organizations, governments, and private industry to develop principled AI. Remember our ‘Nine Positive Steps to Keep AI honest‘, and please share your thoughts on Twitter by mentioning @SalesforceOrg too!
Got half an hour? Earn your first badge about Salesforce Einstein and see how predictive analytics can help make your organization smarter, faster.
About the Authors
Elizabeth is a Senior Solution Engineer at Salesforce.org, where she helps nonprofits advance their missions through the use of Salesforce. She lives in New York City. In her free time, she volunteers regularly with ScriptEd, which works to equip students in under-resourced schools with the coding skills and professional experiences that together allow for access to careers in technology.
Phil Nadeau is a lead member of the technical staff at Salesforce. In 2017, he was a Salesforce.org Technology Fellow in Artificial Intelligence. He started using Linux 25 years ago and has been working in software development for almost as long. Phil has written tens of thousands of lines of code with the LAMP stack (Linux, Apache, MySQL, Perl), Java, C, and a variety of other languages. In 2012 he graduated from Western Washington University with a Masters of Science. He enjoys helping make sense of the Internet, using Java, Scala, Spark and Python primarily for his work in search engineering. One of the highlights of his career was as a programmer on an experiment in machine vision at Bell Laboratories for controlling video games using a motion capture system made from vintage Silicon Graphics workstations and old analog video cameras. For previous work, see Phil’s blog, Why You Should Care about AI and Where Did AI Come From.
Katharine serves as editor-in-chief of the Salesforce.org blog and helps create e-books and other digital content at Salesforce.org. She is a lifetime member of Net Impact, a StartingBloc fellow, and has volunteered in producing “tech for good” events and content with the SFTech4Good Meetup (a NetSquared community) since 2014. A self-described “full-stack human,” she is an avid meditator and yogi. When she’s not managing digital content, you can find her teaching or taking yoga classes around the San Francisco bay area. Follow her on Twitter: @kbierce