Spoiler Alert: People Are Different

December 29, 2021      Roger Craver

 

So why do so many fundraisers ignore this reality and treat donors and prospects with the same one-size-fits-all approach?

One reason is that far too many fundraisers know very little about their donors and refuse to invest the time and money in finding out more. Consequently, stuck in the belief that determining who the donor is presents a seemingly impossible task,  they’re satisfied in delivering the same message to everyone.

Fortunately, there are those who refuse to accept the the status quo  and over the decades lots of time and money has been invested in trying to figure out ways in which to identify groups or segments of donors with similar characteristics in hopes of improving response rates, retention, and donor value.

A purist’s view of segmentation (and here at The Agitator we’re nothing if not pure) is that an essential element involves grouping donors by their identities/personalities/values and soliciting each group with a message specifically tailored to that identity/personality/value system.

In short, the ‘holy grail’ of segmentation is to be able to single out donors and communicate with them on a one-to-one basis. As Nick Ellinger noted in an earlier Agitator post  “if they can tell your personality from your writing, what makes you think that personality doesn’t matter when you are being written to?”

Human language reflects personality.  Guess which one came first?  The “101” in cause and effect tells us the only way one thing can cause another is if it precedes it.  Thus the fundamental importance of determining and testing personality when it comes to segmentation.

Decade by decade we’re  getting there thanks to technology and research.  In minutes and at low cost we can now apply a range of third party data, predictive analytics and modelling to arrive at selections.  But, we still have a long way to go before it becomes common practice to accurately and easily identify and communicate with donors possessing particular values and psychological characteristics – personality — that could generally be described as “segmentation.”

As Kevin noted in Are You Doing “segmentation” or Segmentation?  current practices –although high-tech and quite sophisticated – are still  little more than selections of portions of lists that attempt to pass themselves off as segmentation.

We know from the research and in-market testing in pilot projects at DonorVoice this process of segmentation is not only possible , it works.  The difficulty and disagreements arise (see Kevin and Nick’s ‘quibbles’ in this comments section) in the testing of the segments.

Nick argues that once all the algorithms, analytics and third part data have been applied and a selection made that everyone in that selection should get the same mailing.  Kevin argues that those machine-generated selections need to be further broken into distinct personality/value segments and tested individually. In short, these machine-generated groupings fall short of the one-to-one goal because they assume the group is monolithic and does not consider individual differences.  As Kevin has noted, “ If your testing doesn’t have (at least) two versions against the one-size fits all control then you aren’t being donor-centric.  You’re rearranging deck chairs”

To many, this ‘argument’ may seem esoteric, but in reality it’s fundamental when it comes to the future of fundraising.  Most sophisticated direct response organizations understand their problem is not that of identifying those likely to support it (the list industry nailed this issue years ago). Rather their pain point is to determine which of its donors is open to supporting particular parts of its program. And only attention and discussion on the pros and cons of various approaches to segmentation and testing can relieve that pain.

At this stage when it comes to segmentation we are certain of this:  All organizations have different donor segments but their ‘why’ for giving has zero to do with demographics, behavior, channel, or any other internally defined segmentation.  It also has zero to do with random, persona clusters that were created by throwing everything into a statistical blender.

Start with why people do what they do, recognize it isn’t the same answer for everyone while knowing groups of similar people exist.  Build a test aimed at a specific group but also include a group you don’t think it will work with to more fully establish cause and effect.

The worrisome fact that the charitable sector has not grown (by any useful measure)  since the 70’s is not unrelated to our sector’s failure to pay serious attention to donor selection and “segmentation.”   As a result the in-group of donors is getting smaller every day.

Roger

 

P.S.  If you want to dig into the issue of segmentation in more detail you’ll find these Agitator posts helpful:

Taxonomy of Donor Messaging. Too often we think we know very little about our donors. Consequently, we believe tailoring messages to who they are is seemingly impossible.   Sadly, this means everyone gets the same thing.

Taxonomy is the science of naming, describing, and classifying.  In most fundraising we put our respective versions of fundraising best practice into action against what’s known; usually only the brand and mission and programs of the organization.

One More Way I’ve Probably Ruined My Kids      

We’re  genuinely sad about what this chart suggests about politics and civic mindedness (and parenting).

This is why we spend so much time and energy promoting a better way forward for fundraising, one that recognizes and operationalizes the innate traits of people that turns donor-centric from slogan to strategy.

Sure, politics has made a vice of trashing the out-group and so it’s more divisive but no more or less of a fundraising dead end than one-size fits all.

Vest Size Fundraising  

We get rid of random by focusing on the ways we differ on the inside that cause our behavior.  Our personalities are a great example.  It’s inside stuff and we don’t all have the same one.

Fortunately, Personality can be reduced to a Big Five and from there, we pick and choose based on hypotheses about brand fit.

Is Your Testing Divorced from Reality?

This parody is divorced (slightly) from reality but it still lands dead-on in how 99% of our testing plans assume everyone is the same.  We ended an earlier  with a soft plea that the A/B, test with its random nth assignment of humans into one of the two conditions die a quick death.  We also lamented that’s unlikely to happen. So maybe this post is just more windmill tilting.

HOWEVER…. we tilt because we care.  We care that our sector wastes an enormous amount of time and money on testing to nowhere, with test ideas divorced from why people do what they do and methodology that mistakes the lead standard for the gold standard by using the random nth to split people into test and control groups.   In so doing, we violate a most basic maxim of humanity:  we’re different.

How to Have More Winning Tests

Framing matters.  But it matters differently for different people.  You do have different donor segments but their ‘why’ of giving has zero to do with demographics, behavior, channel, or any other internally defined segmentation.  It also has zero to do with random, persona clusters that were created by throwing everything into a statistical blender.

Start with why people do what they do, recognize it isn’t the same answer for everyone while knowing groups of similar people exist.  Build a test aimed for a specific group but also include a group you don’t think it will work with to more fully establish cause and effect.

The random nth should die a quick death.  It won’t. But it should…

Personality and the Words You Use

Why does any of this matter?   To quote our good friend and former Agitator editor, Nick Ellinger, “if they can tell your personality from your writing, what makes you think that personality doesn’t matter when you are being written to?”

Human language reflects personality.  Guess which one came first?  The “101” in cause and effect tells us the only way one thing can cause another is if it precedes it.  Your personality came before your Twitter account (though  social media trolls may be the exception…)

Transactional Segmentation is NOT Strategic Segmentation

Transactional analysis should not be the only segmentation you do nor should it even be the first.

Transactional analysis focuses on the wrong question:  “what people does my organization want to send this communication to?”  Rather, the first question you should ask is “what people will want to receive this communication?” or “how can I make this communication something these people want to get?”

This order is important.  You want the variables that make the most difference at the beginning of your decision tree so you can create different versions or include/exclude audiences based on the most important information.

Let’s take a trivial, simple example. The most important factor in determining if someone will respond to a communication is whether they are alive.  If they are, they can respond.  If they aren’t, they can’t.

If you were segmenting based on RFM + are they dead, it would make no sense to create 64 different variants of RFM, then apply the “are they dead” filter to every one of those 64 segments.  Rather, you first remove the dearly departed, then do the rest of the segmentation.

There are a few of these gatekeeper questions (e.g., “did they opt out of this channel? No”, “are they on the seed list? Yes”) that are the most important things.  They come first.

What comes next?  What makes the most difference in both who will get a communication and what it contains?  It’s not RFM segmentation.

A Missing Ingredient to Raising More Money –Donor Personality

People are different so we need to stop treating them the same.

Crazy, right?  If the choice was ‘past donor’ in our hypothetical then the fundraiser is most probably grasping at straws. On what basis could she/he develop two concepts aimed at two different audiences with only one audience selected?

Considering most charities have at least several known attributes about their donors (e.g. past giving, gender, age, behavior)  but still have only a single control or a single test idea it would seem that needing more attributes isn’t the barrier; it’s the understanding that goes with those different attributes.

 

4 responses to “Spoiler Alert: People Are Different”

  1. Nicholas Ellinger says:
    December 29, 2021 at 11:02 am

    Thanks, Roger! This one resolved some quibbles and raised some new ones.

    I’m glad you are clarifying that Segmentation doesn’t just mean A/B of versioning but can go beyond that. That was a primary motivator for my original comment, as I think our fundraising letters can be far more than that. We have one client (working to get permission to share in more detail) that mailed with a mail piece with more than 20 different points of data-driven personalization in one letter. They also celebrated hitting their December goals before they celebrated Christmas, so it seems to have been well received. The ultimate goal of segmentation, customization, and personalization is for everyone to have the communication that works best for them—that communication likely doesn’t look like any other person’s letter. That requires going beyond A/B and I’m glad this post embraces that. If you want more on that one-to-one future (cheap plug alert), you can check out our free Death to Buckets white paper at https://wearemoore.com/downloads/death-to-buckets/ about how segmentation can be far more than it is now.

    So let’s talk about the lower end of the spectrum—segmentation without customization. I seem to owe an apology to the reader: if you read my comments as once a selection has been made that everyone in that selection should get the same mailing, I clearly explained myself very poorly.

    Rather, I argue that segmentation *can* take place when people get the same piece. In DonorVoice’s work with Catholic Relief Services, which you can learn about at https://www.youtube.com/watch?v=69S-ygi5zpo , they found one predominating donor identity — Catholics living out their faith through their gift. They honed their message to that audience and selected for, in donor and acquisition mailings, audiences who would be responsive to it. That’s identity-based segmentation and it’s happening even though everyone is getting similar mailings.

    This is true because sending nothing is always a choice. Captain Ahab won’t join Greenpeace. Dr. Frankenstein won’t join PETA. Dorian Gray doesn’t want an AARP card. It’s best for both the organization and these people if they are segmented out of these organizations’ mailings. “Nothing” is what’s appropriate to these people’s donor identities or lack thereof.

    Would it be better if there were multiple messages to different types of donors? Absolutely. But I don’t find the purist’s view that there is only one true Segmentation persuasive or helpful.

    I’m Oliver Twist, Kylo Ren, or Samuel Gompers asking for more (or Andrea True asking for more, more, more). The goal is to be better today than you were yesterday. So if your segmentation goes from full file every time to modeled segmentation, that’s a great big step forward. Let’s applaud that and get ready for the next step. Like Roger and Kevin, I’d absolutely argue that that should be trying to determine different reasons for giving. But I’d also say that if what you can do tomorrow is make sure your donors in Hawaii aren’t getting imagery with three feet of snow in it this time of the year or to stop emailing those people who last opened an email in the Mesozoic, that’s also a pretty good tomorrow.

    If you want to see how this works, another Agitator post I recommend on the topic is
    http://agitator.thedonorvoice.com/what-do-you-mean-when-you-say-segmentation/ that talks through the different types of segmentation and how you can make these steps.

    So that’s that part. Now to the new quibbles with the statement: “All organizations have different donor segments but their ‘why’ for giving has zero to do with demographics, behavior, channel, or any other internally defined segmentation.” This isn’t true or productive.

    It’s not true because many donor segments come from things like demographics. In the CRS example above, “Catholic” is an identity. It’s also a demographic. Similarly, when Drezner et al looked at university giving, they tested four different reasons for giving: need-based, merit-based, first generation to go to college, and LGBTQ person cut off by parents because of LGBTQ status. Those who were LGBTQ themselves or part of another marginalized identity were more likely to give because of the person cut off by their parents. That’s a demographic that is also a productive identity-based segmentation. Even the classic “cat person” or “dog person” can map nicely to the demographic data points “presence of cat in the home” and “presence of dog in the home.” As for behavior, we know a lot of organizations have those who will advocate for political change and those who won’t and that the former respond better to a why of giving that includes advocacy and an opportunity to get involved. If someone takes an advocacy alert on your website (a behavior), might we not assume at least at first that they are in the advocacy group?

    Trying to disentangle identity from other data points can be counterproductive because it can’t be used until you are able to ask them about their status. The Chesapeake Bay Foundation wanted to expand the types of people who would think of giving to them. So it went after an animal lover identity. The mail package teaser read “Because your pets will drink any water they see… there is something you should know.” with a pet tag premium. This doubled the response rate in co-op lists and produced a 30% lift to animal lists.

    It’s a great example of Segmentation – different message to difference audiences. But it it was in acquisition. All they had to go on was demographics (animals in the home) and behavior (donation to other animal charities).

    Or take Kevin’s Conscientious example from the previous post. You can’t rent a list of conscientious people. You can only use proxies for it like political inclinations. At the point that you are using demographics as a proxy for identity.

    Yes, it’s always better to ask. But if you want to have different messages to different audiences and have it start at acquisition, you need to embrace demographic, behavioral, and channel proxies for identity. And that means it has far more than zero to do with the why of giving. To ignore that means that you are starting your donor’s journey with the same undifferentiated messaging you decry.

    Love the spirited debate on these points; let me know when you are ready to start a podcast where we yell at each other about the 1% of fundraising we disagree about!

    • Kevin says:
      December 29, 2021 at 11:21 am

      Nick,

      We do use third and second party data as proxy, so it has value but only because we understand the root cause of why people do what they do and as importantly, how to message to the resultant segments derived from proxy data.

      My major quibble that rises to level of strong disagreement is anyone who uses data by throwing it into a statistical blender and thinks the outputs are magic. If there isn’t strong domain knowledge on why people do what they do then you’ve likely got a faulty, false starting point. And no matter of AI or exponential increases in amount of data will change this. In short, data scientists or those pretending to be one need outside expertise in the psychology of giving to put that data to use in a way that reduces risk and increases the chance of repeatable, positive outcomes.

      The catholic faith example is one where we’ve gone beyond what you cite. We did deeper now within that Faith. A recent, similar example is with folks who Identify as Jewish. We found the source of that Identity is cultural, religious or ancestral. We ran different ads to these different Jewish Identities and we don’t even need much 2nd or 3rd party data. Let the message drive self selection.

      • Nicholas Ellinger says:
        December 29, 2021 at 12:04 pm

        Figured you had worked to subdivide the religious groups since that 2017 webinar I cited. A good example of how segmentation can always get better even starting from a place of understanding just one donor segment to start. So maybe there’s not a big-s Segmentation after all?

        And I agree that data scientists need psychological understanding. But would extend that to say that now those with psychological understanding need data scientists. We need to be open to both types of science: those discoveries that start “this supports my hypothesis” and those that start “huh, that’s interesting…” Because of the volume of data sets and power of machine learning, we’ll be surfacing more connections that don’t have an explanation at first. Those with psychological understanding then can figure out why something works, in the same way that human chess players now can learn to play better by seeing what machine learning has played.

        Or, as you put it, “let the message drive self selection” – that is, putting messages out there and letting people figure out where they best fit or react. That’s not capital-S segmentation, but it sure is segmentation and it sure is smart.

  2. Bob Hartsook says:
    December 29, 2021 at 11:11 am

    Good Morning. Of course, I deal in major gifts which allows you an even more precise understanding of the donor and their needs.

    THE #1 Failure of major gift officers is to transfer their value system to their donor. If the solicitor believes in having children and the donor doesn’t, they approach the childless couple with sympathy and regret. More and more couples are childless or one child have a different attitude toward this topic. If they are a Right to Work person verses and solicitor Union employee base, different approach. Our political attitudes are over lapping our philanthropic dramatically now.
    Views on reason for economic movement, homelessness, etc.

    Reaction to business decisions or family decisions are huge factors with solicitor view and donor view may not align. Even our tools taken literally, like “moves management” can be an incredible tool for one donor and not for another. More and more of our donor pool is taking a contrarian view. One reason the donor pool is shrinking is our marketing is from our point of view actually pushes people away from our cause. The point of this essay.

    Clearly, we have learned asking one person about how another will act in a selected situation more times than not, results in information about the the close friend’s view than the donor’s view.

    Listening is not, “not talking”. Curiosity is the approach to gain information. Even had organizations that “hail” their efforts to fulfill donor intent, who following the donor’s gift retain me to find ways to get around the donor’s intentions. Even worse when financial planners aggressively move donor funding of nonprofit to family members or even the financial planner’s direction.

    Another good essay. The integrity of gift planning is a huge reason nonprofits are not trusted. Happened to me with a life time gift just this week of millions of dollars.

Ask A Behavioral Scientist

    Behavioral Science Q & A

    The Agitator Tool Box

    Ideas, applications, tools, processes, and case studies of break-through solutions in fundraising, including: