Innocuous data signals predicting other behaviours

Dr Michael Housman, Chief Analytics Officer at Cornerstone OnDemand, pioneered the idea that people’s characteristics could be identified by their browser.

He analysed data from 50,000 people who his recruitment software company had helped find jobs and discovered that browser choice accurately predicted their performance. People who opted for a non-default browser, like Chrome or Firefox, lasted 15% longer in their jobs than those with a default browser, like Internet Explorer.

Housman attributed the difference to the fact that choosing Chrome or Firefox was an active decision – those workers were taking the effort to find a better browsing solution than the one pre-installed on their PC. That identified them as someone who wasn’t content with the default.

What’s the marketing application?

Clare Linford and I wondered if Housman’s finding could also be useful for marketers. Perhaps people who avoid the mainstream default browser choice, might do the same in other product categories?

We tested this hypothesis by questioning 22 lager drinkers about their brand of choice. When we split the results by their favoured browser the results were clear-cut. Only a third of lager drinkers who used Internet Explorer preferred a beer from outside the mainstream, top five lagers. However, 56% of those who didn’t use a default browser preferred a non-mainstream lager.

Browser choice predicted other behaviour because it recorded a tiny act of rebellion: somebody had rejected the default.

Interesting. Also catnip to analysts who can turn a click into a permanent theory of personality before the kettle boils.

Customising settings, importing data early or choosing an advanced route may signal effort and intent. Those actions may predict deeper adoption. They may also signal spare time, technical confidence, disability, workplace restrictions or an ancient laptop inherited from Accounts. The visible action is a clue. The mechanism is still wearing an electronic tag.

Installing a browser will not improve an employee by 15%. Please do not add “Chrome” to the leadership programme.

Defaults move. Products change. Populations change. A useful proxy decays, sometimes quickly. Worse, a pattern that helps us understand a journey can become unfair when used to allocate jobs, prices or opportunities, especially when the innocent-looking variable carries differences nobody meant to judge.

The interesting work begins with questions. What could plausibly connect this action to the outcome? Does the relationship survive later data and another context? Who is absent? What else might the signal stand in for? Use the clue to improve the experience or guide deeper research before awarding it power over people.

Predictive systems adore neutral-looking variables. Neutral-looking is a costume. Consequence is the test.

Excerpt from

The Choice Factory: 25 behavioural biases that influence what we buy

by Richard Shotton

View on Amazon More from this book

Behavioural principles

Behavioural ideas at play in this post

Short, plain-English explanations of the principles behind this post, with links to related books and examples in the archive.

Friction costs

Small hassles and extra steps can sharply reduce action.

More quotes

Overconfidence effect

We are often more certain than the evidence really allows.

More quotes

Innocuous data signals predicting other behaviours

Behavioural ideas at play in this post

Sharp thinking (my rants) in your inbox