An Introduction to Nonprofit Communications Segmenting

data-scientists

We often hear from small and medium size nonprofits the lament regarding being too small or not sophisticated enough to perform any type of communications segmenting.

Nothing could be further from the truth!

Before we explore why that is the case let’s peek into why this is so important to do.

The Benefits of Communication Segmenting

Basic segmentation allows for personalized and targeted communications to occur, while opening the door for continuous improvement through multi-variate testing. The long-term impact upon your communication results can make a huge difference in your fundraising success. Among the numerous benefits are:

  • Improved prospective donor response rates
  • Improved existing donor response rates
  • Lower new donor acquisition costs
  • Higher donor retention rates
  • Improved fundraising results
  • More successful events
  • Lower communication costs

The benefits above would be a joy to discuss in any staff or board meeting!

It is Easy to Begin!

The processes we discuss below are simple enough for any appeal of acknowledgement.

The first step is to create some sort of segmentation grid. These segments can be used to enable different aspects of the communication to be created and tested. This is vital since the test may bear out slightly different results, or even opposite results, for certain segments.

The grid below is an excellent first step in segmenting your. From it, you can tailor both appeals and acknowledgements.

four-segment-grid

With this segment, you are effectively breaking your donor database into four groups based on frequency of giving and whether or not they give above average (this requires you to calculate your average gift amount).

You could even create a segment for each fundraising channel, like so:

segments-by-channel

With this grid, you can segment a multitude of communications, including gift acknowledgements and appeals, by format and content.

For example, you could segment a gift acknowledgement in the following way:

  • new donor, above average: phone call from board member
  • repeat donor, above average: handwritten note
  • new donor, below average: thank you letter
  • repeat donor, below average: phone call from fundraising staff

You could also have four types of the same format:

  • new donor, above average: thank you letter #1
  • repeat donor, above average: thank you letter #2
  • new donor, below average: thank you letter #3
  • repeat donor, below average: thank you letter #4

Here, the content of each is tailored to that specific type of donor. The true value of segmenting is that you aren’t acknowledging every gift the same way.

There’s no shortage of ways to customize each thank you letter:

  • A different style or tone
  • A different author
  • A different handwritten P.S note.
  • A hand-addressed envelope
  • A totally handwritten letter
  • A letter + a phone call
  • A letter + a personal visit
  • A letter + a private event invite
  • A note or picture from the recipient of the service provided by the previous gift

Testing the Content Within Each Segment

Once you have your segments for each type of outgoing communications and you have verified your database can easily facilitate the use of the segments, you are ready to pick one or more segments for your first test. If this is your initial testing effort, you should probably just do one test.

After you pick the segment you want to test, further divide it into two equal and random groups for an A/B test.

For each of the two subgroups, try to test a single variable so you can find out which group based upon the use of that variable has the best result. For example, try an envelope with a handwritten address vs a typed address.

If you want to go for an A/B/C/D test, you can further divide the groups into four and add another variable like a handwritten address.  The four groups would therefore be:

  • Typewritten envelope with content #1
  • Typewritten envelope with content #2
  • Handwritten envelope with content #1
  • Handwritten envelope with content #2

The key to both type of tests is having two-to-four truly random groups of nearly equal size for each segment. It is probably also best to have at least 50 – 100 recipients for each group so the results are meaningful. (This may mean A/B tests only for small databases.)

As the results come in for any appeal test, make sure and record the exact sub-group the gift came from in your database. After all results are in, use your database reporting engine to see whether the A or B subgroup performed better or in the case of the A/B/C/D test, which of the four performed better.

If your overall mailing list is only a few hundred in size, segmentation may not be needed. You can just create two random sub-groups and perform an A/B test as outlined above. Just like with various segments, keep doing different A/B tests every time you communicate so you can keep improving results over time.

Going Forward

The resulting winning subgroup should be used for that particular segment until a future test beats that method in a head to head match-up. As you can see, over time you are perfecting your communication strategies based upon proven winners in scientific verifiable tests.

Such basic testing can be quite valuable to funding your mission over time. In addition, such testing can be fun and bring about numerous surprises in what works and what does not!

Do you segment and personalize your donor communications? Let me know in the comments below.

If not, please give it a try and let us know how it turned out.

Jay Love

Jay Love

Co-Founder & Chief Relationship Officer at Bloomerang
A 30+ veteran of the nonprofit software industry, Jay Love co-founded Bloomerang in 2012. He currently serves on the board of the Center on Philanthropy at Indiana University and is the past AFP Ethics Committee Chairman.
Jay Love
By | 2017-06-10T18:23:23+00:00 June 27th, 2016|Donor Communications, Donor Management|

One Comment

  1. Heather June 30, 2016 at 7:24 am - Reply

    Hello! I’d love your thoughts on how small of a group you can realistically use for segmenting. If we are starting with a list of 300, does segmenting in the above-described way make sense? Thanks so much!

Leave A Comment