Three Cambridge graduates working in consulting and advertising had spent six months and about £500 on fruit developing smoothie recipes in a rented kitchen. Market research could tell them whether people liked the taste, but not whether strangers would actually hand over money for it in a real purchase moment — and none of the three wanted to resign from stable jobs on the strength of a focus group's opinion.

The standard move for an unproven consumer product was more research: taste panels, surveys, a business plan pitched to investors before quitting anything. All of it measures stated preference — what people say they'd do — which is notoriously unreliable next to what they actually do with their own money and time.

Three friends sold smoothies from a stall at a London jazz festival in the summer of 1998, with a sign asking festival-goers to throw their empty cup into a bin marked 'yes' or 'no' in answer to: should we quit our jobs to do this full-time? They counted bins, not opinions — a real behavioral vote in place of a survey.

Let festival-goers vote with their cups: throwing an empty cup into a 'yes' or 'no' bin measured real behavior, not stated preference, giving a signal no survey could.

The cup toss was a real behavior, not a hypothetical, so it measured actual preference.

Hundreds of strangers voted with their hands, giving a large, unbiased sample.

The visible yes/no bins made the choice public and consequential, reducing flippant answers.

The low cost of the action still required effort, filtering out those who didn't care.

Before irreversible commitments, design a test where the answer is a real action with a small cost, not a stated opinion.

The 'yes' bin filled; only three cups landed in 'no'. All three resigned the next day. The company they launched, Innocent Drinks, grew into the UK's leading smoothie brand; Coca-Cola bought a roughly 20% stake for £30 million in 2009, raised it to 58% for about £65 million in 2010, and held over 90% by 2013 — sources vary on the exact 2009 percentage (10–20% per Wikipedia, 20% per trade press), so it's given as a range rather than a false-precise figure.

FOLLOW THE EVIDENCE

The sources

  1. Wikipedia — Innocent Drinks Wikipedia · en.wikipedia.org
  2. Gotta Be Marketing — Event Sampling: Innocent Smoothie & The Yes/No Bin Gotta Be Marketing · gottabemarketing.co.uk
  3. BeverageDaily — Innocent Drinks has no plans to crack US as Coke swallows new stake BeverageDaily · beveragedaily.com