Parkinson’s UK turns to AI to head off the ‘charity crunch’

The UK non-profit sector is bracing itself for a ‘charity crunch’ in 2021, according to a recent study conducted by research company Pro Bono Economics, in partnership with the Chartered Institute of Fundraising and Charity Finance Group. 

Demand for their services is surging, but income has dwindled, due to the cancellation of fundraising events and the closure of charity shops. While three-quarters of charities anticipate high levels of demand for their services in 2021, more than eight out of ten (83%) are braced for a continued decline in their income over the year, relative to pre-crisis expectations. More than two-thirds (69%) think it will take more than a year for income to return to pre-crisis levels. 

At Parkinson’s UK, Head of Data Strategy Dr Jesse Mears is looking to Artificial Intelligence (AI) as a way to keep fundraising levels as high as possible, while also better understanding the needs of the community that the charity supports, so that resources are allocated as efficiently as possible.

A rapid change of focus 

The investigation of potential uses for AI at Parkinson’s UK originally began some weeks before lockdown, when the charity began work with Greenhouse Intelligence, a small UK-based AI strategy consultancy. As is typical of a Greenhouse Intelligence engagement, the initial focus was on investigating the ‘art of the possible’, over a window of several years, and developing some quick wins and proof-of-concepts to build momentum and enthusiasm for AI at Parkinson’s UK. 

But as the serious impact of the coronavirus pandemic became clear, the charity found itself facing far more pressing issues, prompting a serious chat with Greenhouse Intelligence CEO Andrew Burgess, says Mears:

When the crisis hit, we knew we had to be there for our community. People living with Parkinson’s suddenly needed much more advice from us: they needed to stay updated on the changing guidance on Parkinson’s and coronavirus; they wanted to be able to manage their condition at home; they needed to maintain their fitness, physical activity and mental health under lockdown. It was really important that, as an organization, we were there, listening to the community and responding to its needs in the most coordinated and evidence-based way that we could.

So I said, ‘Look Andrew, what we’re doing here is great – but instead of having this longer term view of how AI can help our organization transform and do its best for people with Parkinson’s over the next three to five years, what on earth can we do right now, given that people with Parkinson’s are more isolated than ever and getting confused by all the conflicting information being flung around? And how are we going to make sure our funding doesn’t dry up overnight?’ And Andrew was fantastic, saying, ‘OK, cool. This is a great challenge - and I’m going to take that challenge.’

Top recommendations from Greenhouse

First, Greenhouse Intelligence recommended that Parkinson’s UK use a technology from a specialist AI vendor, Relative Insight, to track the topics and themes that communities were discussing online. By using a technique called comparative linguistics, this was able to identify how conversations on the charity’s forum, helpline and social media groups changed on a week-by-week basis.

The software was negotiated on a special short-term ‘crisis’ license to allow Parkinson’s UK to try it out, but also to benefit from it during the critical period. The team was able to identify a number of key concerns among people with Parkinson’s; for example, the importance of keeping fit came up again and again as a topic, leading to the charity offering physiotherapist-led sessions via its YouTube channel and other digital channels. Says Mears:

We could start to understand, for the first time, in a more robust manner, what language our community uses to ask and answer questions across different topics, and on different forums and social media sites. And also from the helpline, so looking at the language in the notes that helpline advisors take when answering calls from people with Parkinson’s, or their family, friends and carers.

The second recommendation from Greenhouse was to optimize fundraising campaigns by using machine learning to better identify potential donors. While the charity already used some more basic analytical methods to do this, machine learning delivered new insights on the best donors to target for specific campaigns, says Mears, and helped to identify regular givers who might be thinking of pausing their donations, or one-off donors who might be willing to upgrade to a regular subscription. The models can be updated regularly, as the economic impact of lockdown affects people’s propensity and ability to donate during the crisis. Says Mears: 

We’re really, really happy with the progress we’ve made so far. We didn’t have a year to sort out something immediately useful around our use of AI. We needed to know, straight away, how our community was coping and what it needed. And we needed to know where best to go fundraising, who to reach out to, rather than bombarding people who couldn’t donate with requests. There’s still a lot of work to do, in terms of our digital transformation and working to be fit for the future - but I’m proud that we’re an organization that sees opportunities and grasps them in an agile and proactive way. That’s definitely been the case here.