Fundraising with Machine Learning

In brief

Although loyal donors have a much higher lifetime value than first-time donors, most charities have a transactional approach to fundraising.
As machine learning technology becomes more accessible, fundraisers can use it to better understand - and improve how they engage with - high-value donors.
There are several machine learning platforms - both fundraising-focused and general-purpose - that some charities are leveraging with astounding success.

From Transactional to Relational Fundraising

Acquiring donors is expensive. So expensive that the cost of attracting a new contributor may be twice as large as their initial gift. One might live with this if most first-time donors gave repeatedly.

However, very few do so. About a third of first-time contributors renew their gift to the same organisation. In contrast, charities retain around sixty percent of donors who have given for two or more consecutive years. Knowing this, one would expect organisations to spend a significant amount of time trying to retain loyal donors.

Percentage of first-time donors retained

Percentage of repeat donors retained

The reality is that the not-for-profit sector tends to have a transactional approach to fundraising. Most organisations focus on refilling a leaky bucket, instead of plugging the holes. As a result, they fill their book with low-value donors while losing sight of loyal members who have greater potential.

Of course, balance is important. Charities need new contributors to grow and replace the donors who lapse. But they also have to understand where the money is - the median donation is $2,500 for gifts above $1,000 and $20 for gifts below $1,000.

As digital technology becomes more cost-effective through cloud and open-source software, fundraising teams have an opportunity to embrace advanced analytics.

Fundraising with Machine Learning

By collecting and aggregating data from email campaigns, social media channels and websites, organisations can build dashboards to answer questions such as:

Donors

How can we meaningfully segment our donors?
Who are the most valuable segments in our file?
How many contributors from high-value segments did we acquire last month/year?

Acquisition

What are the most successful pathways to acquire high-value donors?
How much are we spending on each channel?
What was the ROI per channel last month/year?

Retention and Upsell

What is the optimal message content (ask, information, style) and timing for each contributor?
Who in our high-value donor file is on the cusp of lapsing and what is the next best action we can take to retain them?
What was our high-value contributor retention rate last month/year?

Having aggregated fundraising data for reporting purposes, charities are set to go a step further using machine learning technology.

Tools and Technologies

Many large companies such as banks and retailers are using algorithms to personalise their customer interactions, with astounding success. These algorithms are helping them to win mindshare by recommending content that appeals to each individual's identity and interests. Likewise, charities can leverage machine learning to analyse every donor's history of engagement in order to find causes that they care about and determine how best to win their attention.

The Rainforest Action Network, a not-for-profit focused on preserving the environment, achieved an uplift of almost 900 percent in the conversion of first-time to monthly donors using machine learning to personalise emails.

There are several options available to charities interested in exploring fundraising applications of machine learning.

These range from sector-specific solutions to general-purpose platforms. Fundraising-focused products such as Accessible Intelligence, Dataro and Keela require less customisation and are easier to learn, but they also tend to be less flexible and may be perceived to be less secure. The general-purpose machine learning platforms offered by Amazon Web Services, Google and Microsoft allow organisations to customise them for many purposes and bring world-class security, while involving a steep learning curve. We recommend evaluating these options in terms of their alignment with the organisation's technology strategy, their product roadmap, and their cultural fit.

If you need help to evaluate and implement the right machine learning solution for your organisation, please contact our team at Cognis, who would be pleased to help.

In summary

There are several mature and proven machine learning platforms available to fundraising teams, which could help to deepen their engagement with high-value donors. Machine learning technology represents an unprecedented opportunity for the not-for-profit sector to shift from a transactional to a relational approach to fundraising.