A question of bias: Understanding shopping behaviours online

x Sponsored content on Research Live and in Impact magazine is editorially independent.
Find out more about advertising and sponsorship.

Releasing a new product is always fraught with risk. Will consumers buy it? Will it challenge competitors? And will it be able to maintain its market share in the long term? Traditionally, the purchase funnel has been used to explain how people shop: awareness of a product or service; interest in the type of product; desire for a particular brand; and taking action to purchase it. But does this model fit online shopping?

Google and The Behavioural Architects worked to update this business model for the 21st century, to understand what consumers considered when they searched for a new product. The aim was to see how those decisions could be influenced and what techniques companies should consider using to help consumers make the right decision.

Alistair Rennie, research lead at Google, who helped lead the project, says the tech giant has seen search data identifying changes in how people shop online. “The internet is no longer a tool for comparing prices – it has become this tool for comparing everything, as people no longer want cheap, they want the best,” he says. “That inspired us to think about what ‘best’ means to someone – it might mean best performance, value, quality, most popular or all of those things.”

To explore these changes further, Google and The Behavioural Architects decided to investigate shoppers’ behaviour in real life. They collected screen-capture video and audio of shoppers in action, with participants asked to provide a “stream of consciousness” of their thoughts and emotions over video, which was then analysed by The Behavioural Architects.

This led to the creation of a model of how people make a purchasing decision – what Rennie calls the “messy middle”. The model starts with a ‘trigger’ for a purchase and ends with the acquisition of a product. In between, an infinity loop swirls constantly between two concepts: evaluation and exploration. “It’s symbolic of the infinite potential of the choice, but also a nice way of representing the cyclical relationship between the two mental functions,” says Rennie. “It was an elegant solution to the issue of how you demonstrate the non-linear complexity of decision-making without making a complex model.”

Once Google built the model, the team prioritised six key biases that influenced shopping decisions the most:

1. Category heuristics – mental shortcuts that help us to make a quick and satisfactory decision within a particular category. For example, when buying a camera, people will focus on the number of megapixels it has to help streamline their decision and reduce the amount of information influencing the purchase.
   
2. Authority bias – the tendency to alter opinions or behaviour to match someone considered to be an expert in the subject, such as an economist or a scientist.
   
3. Social proof – copying the behaviour and actions of other people in ambiguous or uncertain situations, such as following reviews or four- or five-star ratings.
   
4. Power of now – people want to have things immediately, rather than waiting for them. For example, instant downloads or 24-hour delivery.
   
5. Scarcity bias – the rarer the item, the more desirable it becomes. This could be either time limited, quantity limited or access limited.
   
6. Power of free – demand for products is greater at £0 than almost any price above it.

The Behavioural Architects and Google then tested those six biases with consumers. To do so, they created a shopping simulation, with shoppers asked to share their first and second favourite brands from a selection within a specific category. Consumers then accessed a site displaying each of the brands’ logos and supplementary information. The supplementary information was changed to reflect the six biases.

In total, 31,000 shoppers were involved, with 1,000 people recruited for 31 different product categories, ranging from whisky to car insurance. The participants were online shoppers who had recently used Google and Amazon to browse products and not yet completed a purchase. The researchers found that deploying the six biases, either individually or all together, led to a switch from the first to the second-choice brand.

To further test the effectiveness of the biases, the team created a fictional new product and set it in competition against established brands. “We took the brand away, made a product up and saw how it did,” says Rennie. “We weren’t expecting to disrupt preferences the way we did.”

The fake products were priced at the market average, as price is a major driver of shopper behaviour outside the control of behavioural science. The team found that when the fake brand was “supercharged” – meaning it had all six biases applied to it – it was able to attract customers who previously opted for their first-choice brand. For example, the first-choice mobile phone brand had a 72% market share. This reversed in the study when the second-choice brand was “supercharged”, and the fake brand was even able to gain 50% of the market when all six biases were applied.

Rennie acknowledges that the fake product had the advantage of not needing to be created and marketed in reality. But he says the results show how brands could seek to gain – or retain – market share by understanding more about how their consumers shop.

“When people have so much choice and information, it is about understanding what is really important to a shopper when it comes to making a decision,” he says. “If you really understand the customer, you can help them make a really efficient decision. And by making an efficient decision, you should end up with some really efficient marketing.”

For businesses, the model is a starting point to adapting their marketing, with the aim of securing a greater share of the market or cementing existing hegemony. But Rennie says this should be done with the consumer in mind. The research highlights three implications for brands: to interact with customers while they are evaluating and exploring potential purchases; to apply behavioural biases to give shoppers the information and reassurance they need to complete a purchase; and to optimise site speed, user experience, and onsite messaging to shorten the distance between trigger and purchase.

“The best way for a brand to build value from this insight is to use it to help shoppers and consumers,” Rennie says. “We firmly believe if you get that right, it is a win-win – more effective decision-making, and more effective marketing.”

This article was first published in the January 2021 issue of Impact.