A phone script moved kids toward generous wishes far more than Santa's voice did.
Researchers ran a field experiment on a public German Santa Claus phone hotline, routing children's calls to one of four AI voice agents in a 2x2 design that varied persona (Santa, high authority, versus Helper, low authority) and framing (persuasive nudges toward prosocial wishes versus neutral conversation). Of 1,072 logged calls, 89 conversations with a median caller age of 6 met the study's inclusion criteria. Persuasive framing raised the odds of a child requesting something for someone else, rather than just for themselves, from 11.6% to 45.7%, nearly quadrupling the odds, a result that held up under statistical controls. The Santa persona itself showed a near-zero advantage in actually changing what kids asked for.
Persona still mattered, just not for persuasion: kids hung up on the lower-authority Helper within the first minute nearly twice as often as they hung up on Santa, 65% versus 39%. That split is the real finding. Once an agent already carries built-in legitimacy from context, like a hotline billed as Santa's own line, the script does more to change behavior than the costume does.
It's a small sample, 89 completed calls, but a rare one: real children on real phones, no lab consent form explaining the plot first.