By Sabine Schuller, Sr. Research Specialist, The Rotary Foundation @s_schuller

                                                                                                                                                                                                               

“It is a capital mistake to theorize before one has data. Insensibly one begins to twist facts to suit theories, instead of theories to suit facts.”
― Arthur Conan Doyle, from the Sherlock Holmes story A Scandal in Bohemia (1891)

Even though this quote is from more than 100 years ago, I think the attendees of the recent Do Good Data conference  would have been proud to put it on a T-shirt.  The conference’s aim is to help learn how data can and is changing nonprofit work.  You might have heard stories about big box retailers and credit card companies using customer information to sell more products. Walmart shipping pop-tarts to Florida just as a hurricane hits comes to mind.  However, there are some who see a different purpose for using data analytics.  Instead of selling more iPhones, they want to pinpoint which rural South African farmers would benefit from having cell phones to track the weather.  Rather than target which demographic will click on a banner ad, they want to identify which youth are at most risk for dropping out of school. 

Here are the highlights of conference presentations that caught my attention most.

Peer to Peer Fundraising

Traditional fundraising is usually seen as a one-on-one relationship between the NGO (non-governmental organization) and the donor.  The NGO gratefully receives the donor’s contribution supporting their work; the donor feels satisfied their resources are now being used for the greater good.  A peer to peer scenario is different. Some examples are the “fun runs” raising money for a cause from the athlete’s friends and family.  Girl Scout cookies  and the ALS Ice bucket challenge are other examples.  In those cases, the support depends more on the relationship between the participant and their donor, rather than the NGO or its cause.  In this fundraising paradigm, you would focus on the connectors to leverage their network rather than one large donor.

What if:  You were a disease fighting charity that used “fun runs” and individual volunteer fundraising pages as your main way to build support?  One runner, Mr. W.E. Coyote, secured one large $200 donation from The Acme Corporation.  Another participant Ms. Roadrunner, had 20 of her friends donate US$20.  In this scenario, cultivating a relationship with Ms. Roadrunner might bear more fruit.  That’s because her larger network could potentially grow exponentially in support of your charity, compared to Mr. Coyote’s one connection.  But in order to do that, which tools would you need to identify your “best” prospects and what information would you need?  Would knowing Ms. Roadrunner’s personal financial situation be less important than understanding the strength of her network?

One Well Presented Graph is Worth a 1,000 Word Report

There’s only so much data the human eyes and brain can absorb without exploding.  Pictorial representations of data, like graphs, are one way to tell a compelling story still based on facts.  One presenter at the Do Good Data Conference used her program evaluation, data analytics, and graphic design skills to explain how to best present hard earned findings in ways non-expert decisions makers could easily digest.

What if:  You presented a donor’s giving history in a pie chart instead of a table?  It would probably make it easier for someone to understand their primary philanthropic interests. 

What if: You organized donor information using hierarchical text for a front line fundraiser:

1.   Philanthropist Sells Kansas Farm: Donates US$7.5 Million

2.     Money will save the lions, tigers, and bears in Oz.

3.       Her favorite color is Ruby Red.

Rise of the Machine

If you’ve ever watched Netflix make their niche entertainment recommendations, you’ve seen an example of machine learning.  How do they do that?  The short answer is that clever data analysts create a mathematical formula by putting individual words associated with a movie into buckets.  By analyzing how many words go in each bucket (aka category), it predicts a result.  Let’s say you streamed Spiderman, Superman, and The Hulk all in a row.  It should come as no surprise Netflix recommends the newest Avengers movie.  The algorithm has picked up on keywords in the movie’s description or reviews like “super hero”, “villain”, “darkest hour” and predicted you’d like something similar.  In another example, this YouTube video shows an algorithm that “learned” the words “sweet” and “pleasant” are predictors of a “good” review.  

What if:  You could use machine learning to identify fundraising developments by analyzing Twitter, social, or traditional media?  If there’s a steady, growing mention of “#FabulousNewFundraisingTrend” maybe that’s a technique you should invest in.

If this piqued your interest, look ahead to the next Do Good Data Conference next year April 27-29, 2016.  I would like to thank my sponsors, the Strategy, Research, and Enterprise area at Rotary which paid my conference registration even though I work in a different department.  That’s how important they think building data knowledge is!  May many follow their example.