back to play arrow

Responding, fast and slow

Amber Brown
Upfront Analytics

Joe Marks
Upfront Analytics

Amber Brown and Joe Marks of Upfront Analytics discuss how using ‘real’ games can help better predict consumer behaviour

One of the things that makes understanding consumers difficult is the disconnect between self-reported and actual behaviour. An explanation for this is Kahneman’s dichotomy between instinctive, reflexive System1 thinking and rational, reflective System2 thinking — actual consumer behaviour is often governed by System 1, whereas self- reporting is usually more a function of System 2. However, by using gameplay, it’s possible to yield better insights than comparable survey data by tapping into instinctual System 1 thinking.

More recently, games have been proposed mainly as a better way to entertain and engage market research respondents, but by using real games, you can also generate data that is more predictive of actual consumer behaviour. The gamification fad, on the other hand, uses simple tricks of behavioural psychology to try to get people to do something they don’t want to do. The simple fact is that nobody needs to be tricked to play real games. Let’s take a look at a couple of examples:


Would you pay for Wikipedia?

We all use the non- profit website Wikipedia, but are we willing to pay for it? We investigated this question using our mobile survey and game app, The Pryz Manor. We asked a straightforward survey question to a subset of our US panel: “I would be willing to donate this amount annually to support Wikipedia: $0, $1, $5, $10, $50.”

This looks pretty good for Wikipedia: 43% of respondents said that they were willing to donate some amount to support its mission. The results are similar to a survey conducted by the Wikimedia Foundation, in which 33% of US respondents said they would donate.

But then we used another element of our app, a game called Slice of Life, to investigate further. In this game, another set of respondents were given a choice of two prizes they might actually win. In this case, they were offered a choice between some cash for themselves or a $50 donation to Wikipedia. The results were very different: 86% preferred a cash prize of $10, with only 14% being willing to forego the cash so that Wikipedia might get a $50 donation. When the ratio of cash prize to Wikipedia donation was increased in two follow- up tests, the percentage opting for the donation dropped even further, approaching zero when the prize and donation were equivalent.

The actual annual donation rate for Wikipedia users is less than 1% globally. The survey questions were obviously ineffective in predicting this kind of behaviour, but a game like Slice of Life emulates the actual decision with much higher fidelity. The Slice of Life game instance with the equal choice — $50 in cash vs $50 donation — is closest to the actual decision faced by people when prompted for a donation, and the behaviour in that game was very close to actual donation behaviour.


What is the perception of General Motors?

For a second example of how games can provide more predictive data, we considered the news reports in April 2014 about defective car parts and the total cost of the US government’s bailout of General Motors. We wanted to know if consumers were paying attention to these negatives around the GM brand. To answer this question we surveyed some US respondents: “Which of these words do you associate with General Motors: Buick, defects, Cadillac, Chevrolet, recalls, GMC, bailout, none of the above?” We also presented the same word selection to another set of respondents in Name Dropper, a wordguessing game inspired by the parlour game Charades.

The cluegiver selects which clues to send to the guesser to help him guess the given topic as quickly as possible. Clues have different time penalties, so the cluegiver is encouraged by the game mechanic to pick clues that are differentiating and useful. The available clues also include distractor words, to discourage mindless selection. The screenshot below shows the clues the guesser has received, along with a winning guess.

In the game, the negative attributes were selected at a lower rate, suggesting that consumers’ instinctive responses have not been greatly influenced by the negative publicity. And, indeed, sales of GM cars have apparently not suffered. At the end of the day, business professionals need high-quality information to help make more informed business decisions.

Through the use of real games, it is possible to gather data on awareness, sentiment and behaviour toward products and brands at a level of continuous engagement that would be impossible with traditional research methods.