There’s a vision of AI (Artificial Intelligence) I (jokingly) referenced a few weeks ago: that it is coming to take our jobs and, in extreme Skynet scenarios, our lives.

But a more realistic, and optimistic, scenario for the long-term is that we will all be enhanced by technology.  Computers are very good at repetitive tasks and following a framework; humans are better at creativity and imagining things as they are.  When they combine with each playing to their strengths, a human-computer hybrid can outperform either humans or computers alone.  Even in chess, thought to play to AI’s strengths, a combination can work best.

That’s great and all, but you don’t have IBM’s HAL sitting in your office.  What leaps forward can be taken now to get more out of fundraisers?  There are a few examples (and I’d love to hear your thoughts):

Automated communications based on activity triggers.  In this 101Fundraising blog post, Eva Hieninger and Gregor Nilsson talk about the perils of uncoordinated messages.  I’ve seen this working with nonprofits as well – when you actually  map the communications a donor gets, those in the room are shocked even though they are usually the ones responsible for the volume.  Different departments and systems send messages on their own schedule; that schedule is when it is good for the department, not the fundraiser.

Automation can be part of the solution for this.  It does require getting your data into one coherent data system (more on this coming up in future blog posts).  Once there, communications can be sent out based on constituent’s triggers, based on the conditions you set.  As Eva and Gregor say:

“In this way, complex, cross-channel donor journeys can be automated and integrated online and offline.  Although setting up such a journey is more time-consuming than with the  simple one-time mailing newsletters, the return on investment is significantly higher for three reasons:

1. The campaign can run for several months and thus continuously brings new leads or donors.
2. The individual steps are triggered by the donor’s trigger. They are therefore a direct reaction to the person’s behaviour and hence have a much higher relevance than classic campaigns based on the scattergun approach.
3. Online and offline fundraising no longer live in different worlds. A digital donor journey can also be used to trigger an individualised print mailing and vice versa.

The result is better and more long-term relationships with the donor, which means higher revenues and lower acquisition costs.”

Automated email drafts for personal fundraisers.  One of the first things I heard in training for major gift fundraising is that a major gift officer should only have a portfolio of 150 people.  This aligns closely with Dunbar’s number of 150: because of our brain size, we can maintain 150 stable relationships at once.

What if AI could help a major gift officer remember the things they need to remember and automate some basic tasks?  A company called Gravyty is employing this human-machine hybrid.  Its First Draft tool learns your writing style to create template emails you can send.  Further, it interfaces with your donor records to customize the email to the person receiving it.  The major gift officer can tweak the email and is in control of hitting send.

The advantage of having the machine remember your donors’ details and give you a template for communication is one of time and customization.  In one case study using Gravvty’s First Draft the University of Delaware’s fundraisers were able to expand their portfolios a whopping 150% by making these routine tasks easier.

Automating post-interaction communications.  Wait, you say!  We already do this with our auto-responders!

Chances are, you do, but it is barely personalized to the donor or the situation.

These communications are like the hot-air dryers in airports – they meet the legal and societal requirements for a job, yet you leave with your hands still wet (and a wall full of bacteria; learn more from the Mythbusters here).

So we’ve recommended looking at commitment versus satisfaction of donors and treating each quadrant of that 2×2 square differently.  Much better!

But what if you customize more deeply?  Amnesty Belgium took commitment and satisfaction and incorporated them as part of a predictive model to see who would lapse.  Commitment and satisfaction were the most predictive variables in the model (this is why models based on demographic and transactional variables will always lose to  models that can incorporate these data, by the way).  But like AI working along with a human, a model that incorporated user-generative information plus traditional measures worked better than either alone.

They then did automated communications based on whether someone was more likely to lapse, with humans entering the picture when someone was valuable, likely to lapse, and having a fixable problem.  As a result, they cut their donor attrition in half.  More details here.

These are just some of the ways AI/human hybrids are helping organizations increase their productivity (plus chatbots – another topic for another time!).  Others you are hearing?

Nick