Charity Digital – Topics – How AI can improve your CRM

Artificial intelligence (AI) is used as a banner term for a large range of computer or robot-led capabilities that mimic human intelligence.

In the charity sector, when we talk about using AI, we’re often referring to generative AI tools like ChatGPT or Midjourney. So, what is predictive AI?

Like generative AI, predictive AI is part of an AI subset called machine learning. Machine learning algorithms can process large volumes of data and identify patterns.

In generative AI this is used to produce new content based on learned patterns. In predictive AI, pattern recognition and analysis is used to make predictions and recommendations.

Predictive AI tools can help charities with:

• Donor prospect modelling
• Donor database segmentation
• Lifetime value prediction
• Engagement analytics
• Sector-specific fundraising needs

Perhaps this doesn’t sound like anything new. Many fundraising teams already have the resources to do all of this in-house. The question is whether AI does it better.

This was the challenge that the founders at Dataro levelled at Greenpeace Australia in 2018. Could their tool make better predictions about giving behaviour and help Greenpeace raise more from its appeals?

John Roberts, Dataro’s Head of Sales explains, “every time a charity has an appeal, or indeed any kind of fundraising activity, the first thing they’ve got to do is figure out ‘who do we mail, who do we pick the phone up to, who do we steward for that major gift event,’ etc.”

“So…the charity built a list of the donors they thought they should mail using their own traditional methods. The three [Dataro] founders used machine learning to try and spot the kinds of patterns in data which would predict who they should include and they did a bit of a head to head. The bottom line is that the machine learning-based list significantly outperformed the list the charity was able to generate using their own internal methods.”

How do predictive AI tools work?

Predictive AI tools like Dataro, Donor Search, or Blackbaud’s Momentum identify patterns of giving behaviour on charity databases using a number of datapoints. Roberts says that at Dataro, “there are up to 400 different data points that we can consider.” The patterns identified are used to make predictions and suggestions on which donors should be included in upcoming campaigns.

Fundamentally, predictive AI applications like these are helpful with retention of existing donors rather than acquisition of new ones. Roberts says, “our job is to help the fundraising team understand what they need to do with that donor to see their lifetime value as great as it possibly can be for each and every supporter.”

Is using predictive AI a risk?

There is inherent risk with using any kind of AI. Even the experts don’t fully understand how AI algorithms are making decisions. This is referred to as the black box problem.

One of the risks that crops up frequently is bias. If predictive AI tools are fed demographic data, could they produce a result that is, in effect, racist or sexist, for example? Roberts says, “things like race, sexual identity, anything that could be construed as ‘special category’ under GDPR don’t form part of the [Dataro] process. So a lot of the inherent biases that AI can fall victim to never enter the modeling equation entirely.”

The approach predictive AI tools have to data privacy is also critical. The potential for data breaches or mismanagement of personal data is huge. Predictive AI suppliers should be able to explain how data is de-personalised, processed, and protected in accordance with GDPR and forthcoming AI legislation.

Are charities using predictive AI?

Pancreatic Cancer UK worked with Dataro on their end of year appeal in 2022. They mailed 65% fewer donors, based on the tool’s predictions, but saw a 111% increase in net income as well as making a saving of more than £12K on their direct mail cost.

How charities can prepare for using AI with their CRM

There are three things charities can do to get their CRM ready to use predictive AI.

• Develop a data management strategy that strives for clean, consistent data
• Look at how your CRM integrates with important tools for acquisition like your email marketing software and plan for scale 
• Budget for using predictive AI on the assumption that it will lead to increased return on investment and test this assumption in one or a series of pilot studies

Charities are rarely short of donor data, but Roberts says, “We’ve had a lot of experience in seeing charities almost fall into a bit of a trap where they don’t want to do anything with the data they have because they’re chasing perfect data.”

“Improving data quality should absolutely be a goal…but it should be much more about…ways of working that see data slowly but surely improved.”