A Comprehensive Guide on Fundraising Analytics for Managers

What is Fundraising Analytics?

Fundraising analytics is a process of discovering insights that make an organization more efficient and productive, resulting in lower costs and higher fundraising results. As with other applied analytics in other industries, technology is changing fundraising analytics' form and function. It now uses data science and artificial intelligence (AI). In this guide, you will learn about the application of fundraising analytics to key needs in fundraising.

Fundraising Analytics for Management

An executive like yourself has limited time, and often you can't give advance notice to your colleagues to pull the required information. Because of the time crunch and high-pressure situations, the chances of making mistakes are high. Automation via data science can help. Also, data and system analysts can offer incisive insights when building or customizing CRMs or databases. Simple tweaks here and there can create amazing advantages. 

Fundraising analytics help create efficiencies such as making existing processes uncomplicated and faster, reducing chances of errors, getting better data, and helping staff focus on more challenging issues. Another benefit of data science: offering approaches to increase fundraising. 

Thus, the two main benefits of analytics or data science are:

1. Improved processes
2. Increased fundraising

We will see examples of these in the following sections.

Fundraising Analytics for Campaign Planning

"We should think of starting another fundraising campaign, and it should be larger than our last one," says the board chairperson to John, a vice president for development at a nonprofit. John replies, "Yes, you may have seen some of our planning material..." The chairperson cuts the conversation, "Yeah, but what do you think we ought to raise?" These words rattle John. But he has a plan.

John works with his data scientists to find out:

1. The number of gifts and dollars raised at each giving level in the last five years.
2. The number of prospective donor households at each giving level. 
3. The giving likelihood of each of these households.

With these data points, John and his team calculate the averages as well as find out the prospect pool. And by multiplying these two, they present a range of possibilities: a low and high estimate. 

Since no one can predict the future and because larger gifts skew calculations, this simple method works well to generate projections for planning. 

Other methods:

• Ask fundraisers to rate the top 10% prospects with the expected gift levels.
• Use forecasting methods: simple three-year averages work well to establish a baseline.
• Use proposals or opportunities for projections.
• Use advanced simulation methods such as Monte Carlo methods.

Fundraising Analytics for Dashboards

"I don't know how to say this," the first line of your email reads, "but the charts we sent earlier contain some mistakes. We will have to re-do and resend them." 

You see this email five hours before your board meeting. 

This scenario is common. The inconsistency in data and its presentation puts you in a difficult place right before your important meetings. You know that your employees are doing their best, but every meeting time, you face a similar problem. 

This problem is avoidable. 

If you and your team can figure out the most important numbers you want to share, and you can agree on how to present them, data science can help you reduce one stressor from your life.

Here's one approach:

1. You meet with your management team and data scientists to discuss the five to six main things you care about regardless of the day or the month. Avoid any future frustrations by committing to these key things, also known as Key Performance Indicators (KPIs). It is also critical to agree on their definitions.
2. You ask your data scientists to create multiple mock-ups of dashboards without the real data. Avoid the temptation to give design feedback, but it is your dashboard, so you can select the designs you like the most. Simplicity makes things clear and attractive.

Once you pick the design, your data scientists can find a way to pull the relevant data, put it in an elegant format, and schedule the automatic delivery to you. 

Some Key Performance Indicators (KPIs) to consider:

• Current FY YTD Fundraising compared with last FY YTD Fundraising
• Comparison of five full years of fundraising
• Number of donors year over year
• Number of assigned or managed prospects
• Percent contacted during the previous six months
• Percent of donors retained year over year
• Number of gifts year over year over a fixed threshold ($1M or $100,000)

Fundraising Analytics for Fundraiser Metrics

"Who are your top performers?" You must have heard this question often.

The answer typically:

1. is based on "gut feeling" that is how well do you know someone's fundraising results either because of some large gifts or because of the critical relationships they hold, or,
2. uses complicated measurements or metrics. 

Occam's razor is a good guideline for us to follow while designing fundraiser metrics:

As one of my colleagues said once, "doctors need only four to five vital signs to assess the overall health of a person." If you require "CT scans," of course, you can prescribe them, but for consistent benchmarking and assessment, simple metrics convey useful insights.

You should select metrics that drive the behavior you want the fundraisers to practice. And the metrics that measure the most-valued activities for you and your organization.

You can also combine these metrics to come up with an activity score, which puts everyone on the same scale, as seen in the below chart of fundraiser activity and top performers.

Some metrics to consider:

• Number of visits/contacts
• Number of closes
• The number of $100K+ closes
• The dollar value of the closes
• Number of qualifications
• Some measure of portfolio awareness
• Some measure of planned solicitation activity

Books to consider:

Fundraising Analytics for Staffing

More fundraisers mean more asks, hence more fundraising.

If you are in a fortunate position of hiring more fundraisers, this question invariably crosses your mind: "how many fundraisers should I hire?" And if you work at a large higher education institution, you also have to think about their regional assignments.

Data science can help with actionable recommendations for both of these issues.

How many fundraisers should I hire?

Budget constraints notwithstanding, the number of new hires depends on how many unmanaged prospects you currently have. You also need to look at the prospects' wealth capacity as well as their likelihood to give.

Since these new hires have to do qualification work, they can go through many names quickly compared to the fundraisers with established portfolios.

Which regions to focus on?

Once you have the data on unassigned, potential prospects, your next questions will be: "what regional territories can I create?" and "what's the best, central location for my gift officers?"

Using a technique called clustering, your data scientists can find the best clusters from your prospects' addresses. You can even discover the central location in each of these regional clusters.

See the example below of a made-up dataset and note:

• the color-coded regions, 
• regional centers, and 
• the number of prospects in each region.

We used similar techniques at the University of Michigan to recommend staffing and portfolios for a program. The results amazed us: in two years, the team went from 1.5FTE to 6FTE, and in a few years, the annual revenue went from $6.2M to $75M, an 1100% increase.

Fundraising Analytics for Forecasting

"I would like to present a goal of $20 million for the next year to the board," says Pat, the VP of fundraising, to Sal. As the AVP of operations, Sal is responsible for providing accurate data and information, helping guide Pat's goals. Sal sweats but has the data to inform Pat's plans.

Sal presents three different forecasting results: a worst-case scenario, a moderate-case scenario, and a best-case scenario. After working with Sal for many years, Pat focuses on the projections rather than the techniques.

But Sal had worked with his analytics team to run various techniques and models. They had experimented with different options:

• ARIMA models with seasonality 
• Three-year average
• Simple discounting of previous years' average
• Linear regression models
• Future proposal or opportunity data
• Annual goal planning metrics

After going through all these results, Sal chose the simplest model, which was easy to explain and provided similar results compared to the other models.

Books to consider:

Fundraising Analytics for Prospecting

Most of the fundraising analytics efforts target prospecting. Artificial Intelligence (AI) also finds uses for advancement for prospecting. We can breakdown these activities in the following areas:

• Predictive Modeling: using statistical, machine learning, and artificial intelligence techniques, analysts can generate segments and leads for major gifts. Some typical terms you hear in relevant discussions: major giving scores, classification, prediction, likelihood, RFM, neural networks, and clustering. These methods have a singular goal: help you prioritize your prospects.
• Relationship Mapping: using vendor-provided tools, diligent sleuthing, or advanced network analysis methods, analysts can show the quickest path between you (via a volunteer) and the prospect. 
• Finding Interests: using vendor-provided and internal giving data and natural language processing (NLP), analysts can generate interest areas for prospects.

Books to consider:

Biased promotion: my colleague and I wrote a book on this topic: Data science for Fundraising. 

Fundraising Analytics for Annual Giving

"Our alumni participation rate has gone up by 3-percentage points," said beaming Megan, the director of annual giving. Julie, the VP for advancement, couldn't hold her excitement also, "That's amazing! Congratulations! How did you turn it around?"

Megan then described the tactical steps as well as the data-driven processes.

Annual giving offers ways to use data and analytics to improve results, which analysts can measure in a short period compared to major giving.

Megan and other directors use processes such as:

• Identify likely donors using predictive modeling
• Upgrade or "cross-sell" existing donors
• Identify community champions
• A/B test direct marketing efforts
• Customize giving appeals for the audience
• Test imagery and words
• Find clever ways of uncovering hidden donors
• Revive lapsed donors
• Save costs on direct marketing efforts (by cutting the lists)
• Monitor response data

Fundraising Analytics for Digital Marketing

"Mom, I gave my first gift to a charity today," said Adam with satisfaction. Mom replied, "That's great, honey. How did you learn about it?" Adam replies, "You know, how some ads follow you all over the internet. Well, a nonprofit specializing in providing IT training has been showing up in my browser. Their images spoke directly to me."

Wouldn't you love to get first-time donors like Adam?

We have grown used to seeing targeted ads, noisy (sometimes spammy) emails, even commercials on TV. For-profit companies and social media companies have amassed personalized data.

If used with care, tools offered by these companies can prove useful for nonprofit fundraising also. 

Many nonprofit websites seem stuck in the early 2000s, and they are missing inbound traffic, possibly warm prospects due to the lack of good Search Engine Optimization (SEO).

Some ideas to consider:

• Localized ads during the tax filing season
• Retargeting event attendees
• Customized video ads for specific segments
• Customized giving pages and options
• Tailored communication
• Manual outreach to consistent email openers
• Build websites/pages and content that targets desired keywords (a critical part of content marketing)
• Build chatbots to direct visitors to tailored giving options or other fundraisers
  
  

Books to consider:

Fundraising Analytics for Automation

What if you get an email with a few potential prospects along with their short bios every Monday morning? What if you get news alerts on your top prospects? What if you get emails when fundraising suddenly drops or increases in a day (along with the transactions)? What if you receive a detailed analysis of your fundraising programs?

How would you react?

All of these examples are real and in practice. Once created, they need minimum maintenance and keep generating results until we stop them. These automated activities save many human hours and have the potential to give you timely information to take action. 

Data science can create substantial efficiencies and save costs with automation. You can ignore all the other hype around data science and AI, but keep this one close to you.

Artificial Intelligence for Advancement

While you feel the buzz around AI for advancement and in general, you may wonder why this section is almost at the end. I am not a believer in sensationalism. While there are good reasons to believe that the modern AI tools can benefit all of us, we have covered most of the practical use-cases using data science for fundraising already.

Don't get me wrong: we can use AI's other applications for advancement, but we don't need them right now unless we see 10x improvement. 

Occam's razor applies to AI too.

That said, we should note some of the advances in AI and related technologies and see how we might use them.

• Natural Language Generation (NLG): this is an exciting and rapidly advancing field. Computer models can generate text whose author (a computer or a human) humans can't differentiate. While NLG models can generate coherent and amusing text, it is yet to be seen how we can use the technology with factual data. We may be able to automatically generate richer profiles of our prospects, but we still would need to verify their accuracy. (Note: we have been using mail-merge and templates for a long-time.)
• Natural Language Processing (NLP): NLP can help us identify entities such as organizations and persons, thematical topics, and analyze sentiment from given text input. If we have enough text data on our prospects, we can use these techniques, otherwise, we may find things that we already know.
• Deep Learning (and other advanced predictive models): deep learning, a complex form of neural networks, can create text, fake images, and videos, produce music, generate art, among other things. We can, of course, use these methods to make our donor identification models better, but imagine creating customized videos for our donors or prospects with this technology! There are many ethical and moral questions associated with this approach, and they will need to be addressed with transparency if one decides to use it.
• Augmented or Virtual Reality (AR/VR): I haven't read a lot about these technologies, so I can't say where they will go. But I see us building tools to excite our prospects and donors. Imagine presenting a VR proposal of a new research lab: the donor can feel how the lab would look like and how researchers would conduct their experiments. 

Books to consider (remember to tune out the hype):

Fundraising Analytics Resources

To come.