Standard appeals to civic duty had plateaued UK organ donor sign-ups for years, so researchers tested eight competing messages on over a million real website visitors — and the one that won wasn't about duty at all, it was a question.

the problem

a public appeal has stopped working, but nobody knows which specific alternative wording would perform better without testing it live

background

what everyone would do

The standard approach when a civic appeal has plateaued was to debate internally which value, social proof, loss aversion, altruism, ought to motivate people most, pick the framing that seemed theoretically most compelling, and swap it in for the generic civic-appeal message that had been running for years.

what they saw

The Behavioural Insights Team saw that internal debate over which persuasive value should work best was exactly the kind of argument that never resolves without a real test against real behavior, since reasonable people could argue for social proof, loss aversion, or reciprocity all sounding plausible in theory. Rather than guessing and swapping in a single new message with no way to know whether it actually performed better until well after the fact, they ran eight genuinely distinct framings simultaneously against real visitors at national scale, letting actual registration behavior settle the argument instead of internal persuasion about which framing ought to work.

the move

The Behavioural Insights Team ran a live, quasi-randomized trial between 24 June and 19 July 2013, showing one of eight message variants — a plain control plus seven theoretically distinct persuasive framings including social norms, a gain frame, a loss frame, cognitive dissonance, and reciprocity — to everyone in England, Scotland and Wales who renewed vehicle tax or registered for a driving licence online during the trial window. The reciprocity message read simply: 'If you needed an organ transplant, would you have one? If so please help others.'

why it works

Testing a plain control alongside seven theoretically distinct framings, social norms, gain and loss frames, cognitive dissonance, reciprocity, on more than a million real site visitors during the same window meant the result reflected genuine comparative behavior rather than any single team's prediction about which value would motivate people most. Because the trial ran live on GOV.UK's actual vehicle-tax and driving-licence renewal traffic rather than in a lab or small sample, the winning reciprocity message's 38 percent relative increase in registrations was a result the team could trust and scale immediately, projecting roughly 96,000 additional annual donor registrations once adopted as the new default. Running all eight variants simultaneously rather than testing one guess at a time also meant the team learned which specific framing won, and by how much, in a single trial window, rather than needing years of sequential trial-and-error to arrive at the same answer.

the payoff

Across 1,085,322 site visitors, the reciprocity message produced 4,256 registrations versus 3,085 under the control message, an odds ratio of 1.38 and roughly a 38 percent relative increase, outperforming every other framing tested including social-norm and values-based appeals. Scaled to the full annual volume of GOV.UK transactions, the team projected the winning message would add approximately 96,000 additional organ donor registrations a year over the previous default wording.

where it breaks

The mechanism depends on having enough traffic or participant volume to detect a real difference between multiple framings within a reasonable testing window, a smaller-scale appeal without GOV.UK's millions of visitors couldn't run an equivalent multi-arm test with the same statistical confidence, and might need to test framings sequentially instead, trading speed for the same rigor. It also depends on being able to randomize or quasi-randomize exposure to each message variant fairly, since a test where certain framings systematically reached different populations would confound the comparison and produce a misleading winner. And a message that wins in a single live test reflects what worked for that specific population and moment, not a universal law of persuasion, meaning the same framing that outperformed everything else in a UK vehicle-tax renewal context wouldn't automatically transfer to a different population, ask, or moment without its own separate test to confirm it.

what came after

The trial, published as a peer-reviewed study in the journal Trials in 2018, is cited in behavioral-policy circles as a rare large-scale, real-world demonstration that a simple reciprocity framing can outperform values-based and social-norm messaging for a civic ask, and the winning wording was adopted as the default message on the UK's national organ donor registration page.

references

1. [1]Effect of persuasive messages on National Health Service Organ Donor Registrations: a pragmatic quasi-randomised controlled trial with one million UK road taxpayersTrials (Springer Nature, peer-reviewed), 2018link.springer.com
2. [2]Applying Behavioural Insights to Organ DonationThe Behavioural Insights Team, 2013bi.team

keep it

same kind of clever

• 2010 · American ExpressAmex didn't advertise its card on the biggest shopping weekend of the year — it invented a whole new day inside it
• 1962 · AvisAvis told customers outright it was only #2, and the confession made everything else it said believable
• 2025 · AXA France (with Publicis Conseil)AXA covered domestic violence like a house fire by changing three words in a contract millions of people already had