Initially, we had created an aggregate scoring algorithm that would give a single value based on several different types of activities in Salsa – signups, actions, donations, emails, etc. That algorithm was configured as follows:

While this algorithm is good for simple targeting of the most active members, by lumping different types of activities together it doesn’t allow the NPA to do more sophisticated targeted, such as people who have taken many actions but have not donated. For that reason we recently implemented a system where we track scores on three different, separate algorithms for donations, actions and emails.

ACTIVISM ALGORITHM

This algorithm gives points for each type of action someone can take in Salsa and then degrades them after 1 year to ½ of their original value, but still keeps a residual value, allowing us to give more weight to recent action takers but still keep a pretty high value for past action takers.

DONATION ALGORITHM

This algorithm gives 10 points each time someone makes a donation as well as a point per dollar they donate. The goal here is to give a pretty high value to recurring donors who, while they may not have as high value of donation as some one-time donors, get 10 points each time their monthly donation is made.

EMAIL ALGORITHM

This algorithm is pretty simple – 3 points each time an email is clicked and 1 point each time one is opened, reflecting the relative importance of each activity.

Using these three algorithms, the NPA can target for fundraising those supporters who have a high activism and/or email score but a low (or no) donation score by setting up a query such as:

Supporters with Activism Algorithm > 10 AND Donation Algorithm = 0 OR Email Algorithm > 30 AND Donation Algorithm = 0.

Note that determining which scores to use for these queries is a little bit trial-and-error – we usually run them, see how many people we get back in the query results and them tweak the numbers to get an appropriately large universe.

We also use these scoring algorithms to produce two types of monthly reports for the NPA. One gives a breakdown of the number of supporters who have a particular score range that month and how that compares to the month before. For example (note – real numbers changed):

The second gives the score for each supporter on each algorithm, allowing them to see people who could be ripe for fundraising solicitations who have high activism or email scores but low donation scores.

Another great use of Scoring is to create an “embargo” group for new supporters receiving a welcome series

The National Physicians Alliance also wanted to implement a best practice for their new signups to their email list – start by sending them a series of emails introducing them to the organization before starting to send them the general stream of communication. Putting that in place involved a few steps:

-      Creating a set of email reply triggers with a condition of “Is New == 1.” That way only people who are new signups to Salsa will receive those messages.

-      Configure one of those triggers to be an immediate reply, and the other two to go 1 week and 2 weeks later, respctively.

-      Adding those three email triggers to every signup, action, donation, etc pages that they create

-      Creating a scoring algorithm that would give people 14 points when they first signed up and depreciate that value down to 0 after 14 days.

-      Add a condition to their email blasts of “AND Welcome Series Embargo Score = 0” to only capture those people who’s score had degraded to 0, thereby indicating that they had been in the database for at least 14 days.

That scoring algorithm looks like this:

WELCOME SERIES EMBARGO

Have more questions or need help making this happen for your organization? Contact PowerThru.

The post Evaluating Supporters’ Online Engagement with Salsa scoring for National Physicians Alliance appeared first on PowerThru Consulting.

There’s a ton of strategies to accomplish that. But savvy online organizers know that the only truly good strategy is one that’s been tested to prove that it works better than other strategies with your particular list of supporters. Doing something like subject line testing to see whether one subject line or another gets more people to open your emails is pretty easy, and commonly used by groups with even small lists. Unfortunately, organizations often lack a large enough list size of former donors to perform statistically-significant tests to see what will make your former donors give more or less. (According to M & R Strategies 2011 eNonprofit Benchmarks Study, the average donation rate on an email is .08%, meaning you have to email to tens of thousands of people just to get tens of donations — which is what you’ll need to see an significant difference from one approach to another).

At PowerThru though, we do have clients with large enough list sizes for us to do significantly significant tests. They were game for having us try to maximize their fundraising revenue, by testing our assumptions during the initial stage so we could use the approaches that proved to work the best for the remainder of the campaign.

What we wanted to test

1) It seemed obvious to us (and to many other organizations we think know what they are doing) that one of the best ways to get former donors to give was by including in their email a reference to the amount they had donated in the past, and to mention that they had not given this year but we were counting on them to do so. The question was — would that work, and increase both likelihood to give and amount given?

2) Everybody likes free stuff, right? The organization had previously produced a bunch of what is affectionately known as “schwag” (pens and such with their name on it). So we decided to offer those as a thank-you gift to people who donated. Would schwag make both past donors and people who had never donated more or less likely to give, and increase their donation amounts?

Skip ahead to read our conclusions, or continue to read the whole story about how we performed the tests.

How we tested it

First, we built a bunch of custom reports in Salsa that helped us pull the supporter keys and amount given by year for our past donors. That’s relatively easy to do (and if you either already know how, or aren’t a Salsa user and don’t care, click here to skip ahead) — build a custom aggregate report based on the donation table (and you might as well join it to the supporter table for a later step, although that’s not necessary), and then set up your columns so they include supporter_KEY, Transaction Date and Amount. Select the Group by Boxes first for supporter_KEY and then Transaction Date, and use the Function option to make Transaction Date be by Year and the Amount field calculate Sums. Save it, click Run and you’ll then see the total amount of donations each of your supporter has given by each year.

Or, actually, more than that. Because you’ll be including a lot of errored or failed donations. So you need to set up your Conditions tab to exclude donations that have particular Result and/or RESPMSG values. Those can vary depending on which Merchant Gateway you use and whether you also enter in a lot of offline donations into Salsa that you want to include in the giving history. My trick? If you’re not already sure what Result and/or RESPMSG field to filter out, re-jigger your report so that it is grouped by Result, then RESPMSG, and then either counting supporter_KEYs or donation_KEY. Run it and you’ll see the total number of donors or donations you got for each Result or RESPMSG and you’ll see you need to exclude things like Result = -10 and RESPMSG = Approved Testing.

Note – if you’re scared by the Salsa report tool you’re missing a whole lot of the usefulness of Salsa! I have some online training videos on our blog, or you can read Salsa’s documentation.

Once we got the data, we exported it into Excel and then did some manipulation using filters and sorting, and then cutting and pasting so each supporter (identified by supporter_KEY) would have a series of columns with the total amount they gave in 2009 and 2010. (Still too detailed for you? Click ahead to see what we learned). Note we treated people who gave in 2011 differently — for them we got the average amount they donated per donation in 2011 so we could ask for them to make one more donation around that size). We then used used the “AND” function in Excel to compare the two columns to see which was greater (formula of “=AND(B2>C2)” will give you a “TRUE” for all supporters where the number in column B is greater than in column C and FALSE if not) and then sorted the rows appropriately. We then created two new columns where we multiplied the greater of those numbers by 1.5 and 2, since our strategy was going to be to ask people to give 1.5 or 2 times the amount they had given in which ever year they had given more. We then created new custom fields in Salsa for 2009 donation total, 2010 donation total, and 2011 ask amount 1 & 2, and imported that data back in.

Another quick Salsa aside — we had previously tried doing something like this using the Conditional Content feature in the Salsa email tool. That nifty feature allows you to create “SalsaScript” that does exactly what we had done using reports, Excel and custom fields — merge into an email a supporter’s total amount donated in a prior year and then multiply that amount by to get a new “ask” amount. The problem? When testing we found that there were some supporters who had negative donation amounts for the year, if they had received a refund in Jan for a donation they had made in Dec. While it was an edge case, it was going to look really silly to have people receive emails that said “You gave $-10 in 2010, would you consider giving $-15 or even $-20 in 2011?” So we went the extra mile and did the work manually, where we knew we could spot any errors like that and fix them.

Okay, enough looking under the hood at how we geeked out in Salsa and Excel. Back to more of a summary:

• We had about 7500 past donors that we divided up in 3 different groups:

1. Donated in 2010, but not 2009 or 2011
2. Donated in 2009 and 2010, but not 2011
3. Donated in 2011 but not in December

We then divided each of those sets into 4 different test groups of 25% in each group based on whether they were going to be receiving email with the different things we were testing:

1. Past giving history and a “schwag” offer
2. Past giving history and no “schwag” offer
3. No past giving history and a “schwag” offer
4. No past giving history and no “schwag” offer

The emails with past giving history contained “ask strings” like “You donated a total of $[[donations_2009]] in 2009, $[[donations_2010]] in 2010 but haven’t yet donated in 2011. Can we count on you to give at least $[[EOY2011_ask1]] or even $[[EOY2011_ask2]] to help us achieve our goals for next year?” The ones without history just asked for the 1.5 or 2 times level. The next paragraph either offered a gift if they donated or not (and included a call-out picture of the gift if it was offered).

Finally, we created 2 segments for the hundreds of thousands of non-donors we had on the list (since they had no giving history):

1. “Schwag” offer
2. No “schwag” offer

So, if you’re keeping track that was a total of 14 segments. But that’s what we needed to actually have this test work.

How we analyzed the results

For this experiment we were looking beyond simple things like open and click-through rate — we actually worked on minimizing those factors by making the email subject lines and content virtually identical (the only difference being the emails with “schwag” offers had a subject line and content calling that out). Instead we looked at:

• Was there any difference in the number of donations and total amount raised per send for each different segment, based on whether we showed their history and/or offered schwag?
• What % of the amount we asked for did donors tend to give? Since we knew how much we were asking for (1.5 or 2 times their past donation amount), we could compare that to what they actually gave to see if there was a lift or not.

What we determined

• For former donors, the better performing segments in terms of amount raised or number of donations per email sent tended to be the ones where we showed history and did not offer schwag, although there was some variation in the results. Former donors however did very clearly tend to give more than we asked for when we showed history and did not offer schwag:

• For non-donors, the number of donors and amount donated per email sent was clearly significantly higher when we offered schwag than when we didn’t – 160% and 205%, respectively.  And % given of amount asked was also clearly higher:

Conclusion

In the end, while this was a lot of work it felt like work worth doing, because it taught us several important things:

- Your former donors are more motivated to give and to give more when you tell them what they gave in the past, remind them that they haven’t given since then, and ask them to please give again.

- Your former donors aren’t helped much in their giving likelihood or amount if you offer them a free gift. They do fairly well if you offer them a gift and DON’T show their history, but worse if you both show them their history AND offer them a gift. Therefore we thought it would be best to just continue to ask them for a donation based on past giving and not offer them any gift in return.

- Your new donors are the ones who are significantly motivated by a free gift, both to give at all and to give more than you are asking for. So start ordering little bits of schwag with your group’s name on it, but perhaps only get enough for your prospective new donors.

Have more questions or need help making this happen for your organization? Contact PowerThru today!

The post What really works for end-of-year fundraising? appeared first on PowerThru Consulting.