Turning an Anti-Vax Challenge into a Fundraising Breakthrough Using AI

Decision Science | Concept blocks representing fundraising and AI
Decision Science - Filipe Páscoa | Senior Partner Consultant, Brazil Decision Science - Bernard Ross | Director
An interview between Bernard Ross and Filipe Pascoa on the Gustave Roussy project — an impressive case study showing the impact of applied AI and Decision Science on fundraising campaigns.


Introducing Gustave Roussy

Bernard Ross (BR): Filipe, you’ve been working on some fascinating projects at the intersection of fundraising, AI, and behavioural science. One of the most striking, I think, was with Institut Gustave Roussy in France. For those who don’t know, can you explain who they are?

Filipe Pascoa (FP): Of course. Gustave Roussy is Europe’s leading cancer research centre and the largest in France. They’re not just a research institute — they also run a major hospital. They treat thousands of patients, carry out surgeries, and at the same time advance some of the most important cancer research happening anywhere in Europe. But to do that they need significant money. And they need to raise that from a range of sources.

Finding the Challenge

BR: So, they already had a serious public fundraising operation?

FP: Yes, very solid. I worked with their individual donor team. Let me be clear they have skilled staff, clear strategy, strong systems, and they collaborate with several fundraising agencies. They weren’t looking for someone to fix a broken system — they wanted to take something already successful and push it further.

BR: What was the specific challenge they wanted to tackle? More donors? Different donors? Higher gifts?

FP: The brief was to find new audiences and new donors, especially for their annual appeal. It’s a campaign that already works well, but they wanted to diversify audiences, and their extend reach.

BR: How did you begin?

FP: Our process usually starts with a discovery step. First, we map all available data — internal records, donor insights, previous campaigns. This lets us know what is working and provides benchmarks. It also creates a knowledge base to inform the AI.

Then we add AI and social listening, mapping the whole conversation about cancer in France — not just about Gustave Roussy, but across the ecosystem. We take data from X, Facebook, Instagram etc.to discover what do people really think about cancer and cancer research.

From this we produce an Inception Report. In the Gustave Roussy case this reinforces some things we already considered. One major finding was the prominence of Octobre Rose, France’s national breast cancer awareness month. Everyone knows it. It shapes the conversation about cancer. We knew we had to use this as a key channel.

We also found out what the focus was in health terms. From the narrative analysis, we saw which cancers people talk about most — breast, prostate, lung, bowel.

This was important. We know that there’s often there’s a gap between what an organisation wants to prioritise and what the public conversation is about. For example, in Australia, we supported Cancer Australia’s Daffodil campaign. Strategically, the focus for them was bowel cancer. But the public conversation was dominated by prostate and breast cancer. When we encouraged the campaign to pivot to those narratives, fundraising results improved dramatically.

In France, we found a different challenge. Much of the conversation was less about the disease itself and more about prevention and research. This was a natural fit for Gustave Roussy.

The Anti-Vax Insight

BR: You moved to focus on the most discussed cancers and research?

FP: Well, that might seem like the obvious move, but we discovered something else. The scale of the anti-vaccine community in France was striking. Around 17% of the conversation — that’s one in seven voices — was anti-vax. People we’re questioning science, criticizing investment in vaccines, spreading doubts.

At the same time, people online were talking about Gustave Roussy researchers developing a ‘lung cancer vaccine’. Technically, such a treatment is immunotherapy, not a vaccine. But language matters. For the public, ‘vaccine’ was the word. We had to encourage Gustave Roussy to keep things simple. From a decision science point of view we wee keen to reduce ‘friction’ in language and find a first step point of engagement

BR: so how did you tackle the anti vax message?

FP: We knew we couldn’t convert hardcore anti-vaxxers certainly not overnight. Instead, we designed messaging for two outcomes:

  1. Reinforce the beliefs and attitudes of pro-science supporters — give them tangible reasons to donate. This built on the behavioural principles of confirmation bias heuristic and the availability bias
  2. Create empathetic touch points with the undecided — not attacking their beliefs, but showing the human face of research. We shared insights about messengers and identifiable victims

The campaign put researchers in white coats, shared stories of children, highlighted prevention breakthroughs. The overall message was simple: Despite the noise, science is advancing. We are making progress. And you can be part of it.

We underpinned this with some other decision science frameworks and used AI to identify how to frame the messaging and to whom.

The Results

FP: The outcome was outstanding. The campaign achieved about a 60% increase in donations and new donors — and this was already a successful campaign.

BR: How would you summarise the messages for the organisation?

FP: For me, the lessons were is clear:

  • Be open to new insights, even uncomfortable ones that don’t ‘fit’ with your ideas
  • Take calculated risks and communicate to your key supporters first. Reinforce them.
  • Trust the assets you already have — knowledge, research, stories. Use them wisely.

Lessons for Fundraisers

BR: How about key general lessons for fundraisers?

FP: This case shows how forward-thinking charities can use AI and behavioural science not just to improve fundraising, but to turn challenges into opportunities:

  • Listen broadly, not narrowly.
  • Align with narratives people already engage with.
  • Embrace the non-obvious.
  • Turn risk into opportunity.

For Gustave Roussy, that approach didn’t just drive income. It positioned them as a trusted voice of science and hope — exactly the role a world-leading research institute should play.

To find out more about how you can use Decision Science and AI in your campaigns contact Bernard Ross bernardross@mc.consulting or Filipe Pascoa filipe@enlaight.ai.


Institute Gustave Roussy

The Challenge

Listen to what the audiences we are addressing are saying in order to better understand the main issues surrounding cancer.

Gather content from social conversations to refine the mobilization strategy for the target audiences, ensuring we address their concerns more closely.

Understand the expectations of these audiences and the cultural movements to transform them into concrete ideas and content to be delivered to them.

To create increasingly engaging content capable of mobilizing new audiences and retaining current donors of Gustave Roussy.

Screenshot of the Gustave Roussy case on the Bee Relevant website

The Insight

Through social listening, we identified a significant cluster of general anti-vaccine misinformation. While not directly targeting Institut Gustave Roussy, this wave of disinformation posed a risk to public trust in scientific advancements.

By combining Social Listening with AI, we were able to sift through thousands of documents and uncover something remarkable: researchers at Gustave Roussy were making groundbreaking progress in the field of lung cancer vaccines.

That’s why we proposed a fundraising campaign with a dual purpose: to fight misinformation with knowledge and to raise funds to accelerate this vital scientific research.

Screenshot of the vaccine initiative on the Gustave Roussy website home page

The Results

# ONE-TIME DONATIONS +74% vs N-1

$ ONE-TIME DONATIONS +67% vs N-1

 

This website uses cookies for collecting data through forms and our mailing list. Learn more.