Six Tips to Maximise Appeal Returns & Build Better Connections With Your Donors

The basic formula for direct mail appeals has been unchanged for years. But what if there were simple steps charities could take to raise more funds and improve the donor experience? Fundraising leaders across the globe are now testing new strategies using AI-driven tools to get the best bang for their buck from these old media channels.

Here’s a wrap up of some of the clever ways charities have used Dataro’s donor scores to cut direct mail costs and increase donations.

1. Forget about segments and your ‘active’ universe and select from your entire database

The conventional wisdom is that more recent donors are more likely to give. Selection strategies therefore focus on an ‘active’ universe of 12 or 24 months and build ‘segments’ around these groups. In general this trend is correct. But by limiting selections only to those cohorts, we’ve found charities are missing out on a lot of donations both from older ‘lapsed’ donors and from donors who slip through the cracks in the segmentation approach.

Royal Flying Doctor Service Victoria put this theory to the test in their 2020 Tax Appeal. Using predictive modelling, they found nearly 1000 additional high probability donors that were ‘missed’ in the segmentation approach. These donors ended up contributing more than $26,000 to the campaign. As we can see, by widening the net with machine learning, charities can abandon artificial constraints like the ‘active’ label and score each donor on their chances of giving, taking into account all of their interactions with the organisation.

 2. Decrease the size of your Wave 2 or reminder mailing (at the very least)

A typical outcome from using machine learning is that charities raise more money while contacting less people. However, we’ve found a lot of organisations at first are reluctant to reduce the size of their mailing out of a concern they might miss donations.

Royal Flying Doctor Service Victoria also found a great solution to this problem. They still mailed the entire original selection (plus extra donors we found) in Wave 1 – even the donors that Dataro predicted were unlikely to give. But they significantly reduced the size of their reminder mailing to only target the most likely givers, immediately saving $9000. You can read more about it in this case study.

 3. Review propensity scores again before sending your reminder mailing and add high probability donors

CRM integrations mean that charities can now get updated predictive scores every week. But traditional campaign plans haven’t adapted to this faster tempo. Instead of simply sending wave 2 to everyone who received wave 1 (minus people who responded or returned to sender), that means charities can now introduce new donors who only receive the reminder mailing. We’ve found that often there will be new additions to the database between Wave 1 and Wave 2, or else a donor’s predictive score might change based on their other interactions with you. If these donors are now likely to give, it makes sense to ask them for a gift and increase overall returns.

 4. Optimise your ask amounts and make sure you test the results

We’re often asked if Dataro can predict how much money a particular donor will give. While this is something Dataro can do, it is not actually what you want. Instead it is better to use a better, more tailored ask strategy that takes into account everything about your donors to predict the best ask strategy.

Instead of simply applying standard ask strings we recommend testing which strings work the best for your donors. Predictive modelling can assist by showing which donors are likely to give more and which donors are likely to give less, to help inform whether you ask them for a large increase, a small increase, or no increase. This can also be used in conjunction with advanced modelling for premium packs, discussed below.

 5. Send your premium packs to the right people

How do you determine which donors will receive a ‘premium’ pack. Putting together higher cost mail packages is a risk, because the cumulative amount of money spent can grow very fast. One option is to use a separate predictive model that looks at which donors are the most likely to give larger amounts. Instead of just predicting the likelihood of giving, these models can identify who might give about a particular threshold. Our standard threshold is $500. By using a filter like this, high value packs can be directed towards the individuals more likely to give a larger financial return, hopefully encouraging more of them to do so without breaking the bank by sending high value packs to too many people.

 6. If you want to send to more lapsed donors, use the right model

Often charities are faced with a difficult decision when it comes to mailing ‘lapsed’ donors. They can either include everyone up to a particular lapsed threshold or none of them. There’s know way to reliably narrow down the group. Using predictive modelling, however, we can create specific models that will only look at likelihood to convert lapsed or long lapsed givers. Typically these campaigns return a much lower response rate, however this donor scoring has the advantage of including the most likely givers from across the entire lapsed cohort. It also allows you to project in advance how the group is likely to perform, so charities can enter the campaign with realistic expectations. When compared to new acquisition groups, these reactivations cohorts can still perform well and with the added benefit of no additional data costs.