Cover Image for Fundraising segmentation 101: the basics

Fundraising segmentation 101: the basics

Segmentation is a powerful tool for fundraisers but it is something that I haven’t often seen done well and something that there is very little information out there on that isn’t locked behind a paywall or an agency.

I’m changing that!

I aim to publish a series of three posts about my thoughts and experience on segmentation, each increasing in complexity and sophistication. This is the first one—this one—is covering my absolute foundations. The second will be about reporting and the third will be approaching building out a complex segmentation model.

Segmentation in fundraising is the process of dividing up a supporter database based on a specified data criteria in order to more effectively communicate with them and improve campaign performance and ROI.

Each criteria should have a hypothesis as to why that criteria should work and how those people will respond–the hypothesis could be simply that these people won’t give. Hypotheses need to be tested!

Segmentation can be really complex or really simple, it is really up to the organisation/team.

The foundations of great appeal segmentation are:

  1. Actually segmenting your supporter data
  2. Have an understanding of your segments
  3. Keep segments comparable between years and campaigns
  4. Think beyond just print / direct mail
  5. Make it implementable – There is no shame in keeping it simple
  6. Bonus: Track the supporters movements through them over time

Actually segmenting your supporter data

Sending a printed letter to an entire database is expensive. A great way to save money is to just send it to less people! But a great way to raise more money is to send it to more. What to do? Oh the dilemma!

You can do both by segmenting!

There will be people who have never responded in years to the type of campaign you are running—boom that’s a segment. It will probably cost more money to send a letter to these people than what you will raise (worth testing though!).

There will be people who always respond—boom another segment. You definitely have to send to these people.

It might be worth having a slightly different message for new people because they are new and special (probably not though, please test this)–yet another segment.

This group of people always respond to emails, it could be worth not sending them print, providing their email is opted in—oohhh a fancy segment!

Starting simple like this will help you begin to dial in the ROI of your campaign and mean you can justify sending to less people because you are sending to the theoretical right people.

Your database is not an amorphous blob, every supporter is different and would have a different data trail with your organisation. You may not have all the data in the one spot, but there is probably some data there that can help you segment your audience.

Have an understanding of your segments

I made a $2 donation online to a charity a year ago and I’ve got every single appeal letter with all kinds of random crap in it (letter, magnet, postcard, post-it note, bookmark and more). Four appeals later, I have not made another donation and can see my donation has been totally chewed up by print goodies.

I don’t think there is a great understanding of what the segments are at this charity.

As I’ve donated within the last year I’d say I’m in the ‘new donor’ category, probably based on a created date somewhere. However I only donated $2, is it worth sending me everything? Probably not.

I’ve seen a number of charities now where the segmentation is done by a dark data wizard in a cave somewhere scary that the fundraisers approach only when absolutely necessary. If the fundraisers are clever they send in the sacrificial agency employee first with any request to see how they respond.

You don’t need to know how to weave spells of SQL in order to understand your segments, but so often they are treated this way. Segmentation is a black box and the wizard tells you it works so it does (or they tell you it is impossible).

As a fundraiser you should really understand who these groups you are communicating with are as it should change how you communicate or approach them. It is important to understand specifically who is in and who is out.

If you have a data wizard, get them to write a short description of each segment in plain english for you. If they grumble, tell them to use ChatGPT to help them–wizards love new shiny technology.

I don’t think the charity mentioned earlier understood their segmentation because if they did I don’t think they’d see the value in sending so much to someone who only donated $2.

The messaging really thanked me for the huge impact I’d made and blah blah blah. I donated $2, it cost you that much to send me this. This doesn’t make sense to me as a donor.

New donors are typically special, but have a dollar threshold as well as a date range as part of your segment.

Keep segments comparable between years and campaigns

More often than not segmentation is approached anew for each campaign or mailing. It bugs me as it’s not building on or often taking into consideration the previous years/campaigns results.

Ideally segments are refined for effectiveness each year, yet still kept comparable. Doing completely different segments each year for the sake of it is not ideal. You learn from one campaign, tweak and refine, then learn etc.

Comparisons year on year will really help you understand if your segments are working and will empower you to make smart decisions in future years/campaigns.

Once you have multiple years of data for a particular segment—response rates, average donation value, primary response channel etc—you will be able to accurately predict appeal outcomes and better understand your supporters’ giving behaviour. It’s amazing!

More insights lead to a better understanding which leads to smarter fundraising.

Think beyond just print / direct mail

From my experience segmentation typically starts and ends with direct mail. The driver is effectively maximise ROI by reducing print expenditure. That is good, but you can do better—segment your whole database.

Segmenting your entire database and then looking at all the segments that responded may surprise you. The segment “These people will never ever respond” had 15 donations via email, interesting–why, let’s dig into it!

I’ve worked with a client and we segmented their database into 25ish different segments. The direct mail only went to the top 8 which were grouped into three separate packs. Here the segments were used to predict giving behaviour, the segments with the highest predicted giving got the print packs.

This way you can bring the segmentation into other channels like email and SMS too. I use it more as a tool to dial up or down frequency rather than a customised message.

I haven’t experienced much success with creating unique email sequences or dynamic content or customised emails per segment. Most often it is horribly time consuming and the juice is not worth the squeeze—but, please test it for yourself!

Make it implementable–there is no shame in keeping it simple

Your segments being implementable is the most important thing on this list. You may design the most comprehensive, brilliant model for your organisation, but if it can’t be implemented it doesn’t matter.

You may have technology limitations, old databases or database silos. However, the most limiting factor I’ve found is the skillset to translate a fundraiser’s ideas into a dataspec, then into the segments. Data wizards are rare.

There are tools out there that are quite limited, but it may be enough, something is better than nothing.

Talk to your IT/data people, try and understand what is possible and work with one another. IT/data people like to say things are impossible—some data wizards are in fact just sneaky data muggles—keep asking questions, ask why, push through the impossible, make them think, and find a solution that works for everyone.

Bonus: Track the supporters movements through them over time

How people move through segments overtime is an amazing datapoint itself. Are there patterns of movement? Is it related to tenure? Can you predict multi-year movement? Does this movement towards a net growth or decline in predicted income?

It isn’t something that I’ve seen come out of the box, but the possibilities that tracking segment movement over time yield excites me. It’ll help paint a better picture of people respond throughout the tenure of a supporter.

Not a fundamental, but something to consider if you are setting up segments now. Your future self will thank you!

My key takeaway is to utilise segments, understand what they are, make them comparable, think beyond print and make them implementable.

Keep reading