This article is part 3 in our year-end follow-up strategies series.
“Test Everything!” We all hear this over and over from the fundraising gurus. Clearly, it isn’t possible in terms of time or finances to actually test everything.
But don’t let the thought of testing overwhelm you. Just start small!
With some advanced planning, you can easily design a simple test for one of your year-end campaign activities.
Let’s say you are going to do a small scale phone follow-up to your year-end appeal. Before you begin, you’ll want to think through how you’ll know if the calls actually influenced your donors to give.
You can do this by comparing two groups of donors – one group gets a call and one group does not. You’ll want to see if the donors who got a call actually respond at a greater rate or with larger a larger gift than the donors who are not called.
First, divide your original call list (of non-responding best donors) into two separate lists split alphabetically by last name. This is a nice random division and an easy way to know at a glance to which group a particular donor belongs.
Group 1 is your control group and will not receive a follow-up call.
Group 2 is your test group and will receive a call or voice message.
Second, you’ll need a plan for how to track responses from both groups. Hopefully your donor database will temporarily let you tag the donors as members of one of either group. If not, no worries! Just create a simple spreadsheet for each group that you can manually tag when someone on the list makes a gift.
Third, after your phone-a-thon is complete and your campaign cut off date has arrived, calculate and compare the percentage responses and average gifts for each group
A mathematical reminder of how to calculate percentage change:
[(% response rate of group 2 – % response rate of group 1) / % response of group 1] x 100 = % increase in response rate.
Let’s look at a hypothetical example:
Whoville Children’s Literacy Council mailed their 2012 annual renewal appeal on November 26 to 4,000 donors.
3 weeks after their appeal was mailed, they pulled a list of their best non-responding donors, who they defined in the following way:
- No gift yet in 2012 AND
- At least 1 gift in 2011 AND
- At least 1 gift in 2010 AND
- Had a phone number in the database.
Let’s say this generated a list of 500 donors deemed worthy of a follow-up attempt. They divided this list of 500 donors in half (approximately) alphabetically, then held a phone-a-thon with board members on December 17th.
Group 1: (Last names A-K) contained 230 donors. 6 donors made a gift by the cutoff date of December 31st, two weeks after the phone-a-thon.
6/230 = .02608 x 100 = 2.6% response rate
They calculated average gift of Group 1 at $25.
Group 2: (Last names L-Z) contained 270 donors. 26 donors could not be reached by phone and were thus disqualified from our test group. This left 244 donors who received either a follow-up call or voice message.
15 donors responded with a gift by December 31st. Group 2 has a response rate of
15/244 = .06147 x 100 = 6.15% response rate
They calculate the average gift of Group 2 to be $45.
WCLC could now calculate the % increase in response rate for Group 2:
6.15 – 2.6= 3.55
3.55/2.6 = 1.365
1.365 x 100 = 136%
So, there was a 136% increase in the response rate for donors who were called. Not bad!
WCLC calculated the % increase in average gift for those who were called versus those who were not called…
[(Avg Gift Group 2 – Avg Gift Group 1)/ Avg Gift Group 1] (x 100) = % increase in average gift
20/25 = .8
.8 x 100 = 80%
So, there was an 80% increase in average gift for those who were called. Also pretty awesome!
WCLC documented everything they did in order to repeat this activity next year and compare results from year to year. Now they’ll be able to use this data to recruit more callers, include more donors, and raise more money.
It isn’t complicated nor is it expensive to design a simple test. The key is thinking through in advance what you’ll need to measure. Then you can design your activity so that it is easily measured.