In the United Kingdom, Walkers is a popular brand of potato chips. Or 'crisps'. Whatever. You people and your languages. Anyway, potatoes like rain. Not too many other people like rain, but potatoes, which I've just realized that I've implied to be people, like rain. So Walkers, which pushes a potato-based product, in 2010 opted to base a promotion around rain.
Here's how it worked. When you bought a bag of chips, the bag came with a code. You were given a map of the United Kingdom. The map was divided into a grid as mapped out by the Met Office, the UK's national weather service, or at least based on what the Met Office was using. You were to select a location on the grid, and predict rain for that location (according to the Met Office standard of 1 mm) on a time and date of your choosing. If you predicted correctly- that is, if at least 1 mm of rain fell on your selected patch within a 4-hour window- you won 10 pounds.
When you're doing promotions with prizes at stake, a crucial thing you need to do is build in something sufficient to ensure you only are handing out a certain amount of prizes. Hand out too much, and you end up costing yourself more than the amount of the business you drummed up, defeating the purpose. A number of Kickstarter campaigns have unexpectedly gone down in flames this way when the creator of a project gave out too much in reward-tier and stretch-goal swag, spending money on rewards, and shipping rewards, that should have been going towards the project itself. This was Walker's first mistake: there wasn't enough of a limiter. They limited people to two predictions a day, and limited spots on the grid to one prediction at a time- but this was something of a skill competition, as enough knowledge will permit you to figure out where it's likely to rain tomorrow. And 10 pounds- about $16 at the time- is, of course, many times the cost of a bag of chips (40 pence in this case, with 100 pence to a pound). What needed to happen here was a lot of dry weather, or rainy patches of land going unpredicted.
From 2008-2012, the United Kingdom received an average rainfall of 1,220 mm per year. The United States, by comparison, got 715. It's not exactly tropical-level stuff- Costa Rica got 2,926, for example- but it's pretty rainy by European standards. As a result, there were quite a few correct predictions. And in an already-rainy country, Walkers happened to hit a wet spell during the contest period, which itself was held in the fall, also known as the UK's rainy season. At one point, David Spiegelhalter of Cambridge University calculated that, predicting randomly, one could expect to make 1 pound, 20 pence off of any given 40-pence bag of chips, which triggers the classic marketing blunder of offering a prize worth more than the cost of the product.
Walkers, at some point, quietly reduced the prediction allowance from two per day to one per day.
What happened next depends on what source you're going by. If you're going by the Uncle John's Bathroom Reader (Zipper Accidents, specifically) where I originally caught wind of this, Walkers saw an especially rainy week coming over the horizon and prematurely cancelled the promotion. If you go by Marketing Week, an angry player complained about the reduction in play allowances to the Advertising Standards Authority (ASA), who, the following March, agreed with the player and ruled that Walkers should not have altered the rules after the contest was launched, and that they may not run another promotion in that form. (These two versions of events aren't mutually exclusive, but they don't make any effort to confirm each other either.)
If you're going by UTalk Marketing, the whole thing was actually a great success. Which leads me to ask how many chips they bought.