Unilever is trialling in-store facial recognition technology that measures shoppers’ engagement with on-shelf displays, in an attempt to gather more detailed analytics about the physical retail environment. The trials, which have so far taken place in the US and Brazil, were revealed earlier today (7 March) at Marketing Week’s Insight Show.
In partnership with Mesh Experience and Emotion Research Lab, Unilever placed discreet cameras on shelves to analyse shoppers’ facial expressions as they approached its product displays, calculating the basic demographic profile of the individual along with the ‘noticeability’ of the display, and the levels of attention and engagement given to it. Shoppers’ images were not recorded in order to preserve their privacy.
Unilever vice-president of consumer and market insight BV Pradeep told the audience that the trial is intended to correct the lack of effort that goes into measuring the effectiveness of below-the-line marketing investment, which he believes is currently being neglected.
“If you look at the area of mass media communication – whether it is television or digital – there is a lot of focus on whether the quality of the asset is right, whether the length of the TV commercial or digital pre-roll is right, whether it is placed in the right place, what is the viewership action, what is the click-through rate and many more [metrics].
“Whereas when it comes to the investment that we do in-store, there’s hardly any measurement and we feel it’s fine and it should work. This is a bit like, to me, the syndrome of a drunkard who is searching for his lost key under a lamp post just because there is light.”
More ‘engaged impressions’
A trial with Unilever’s Knorr brand in Brazil involved testing a product display with and without a hanging shelf stopper, finding that noticeability jumped by 3.5 times when the stopper was used, although this declined in the second week that it was present.
The trial also found the display using the stopper was more likely to engage men and that it increased engaged impressions overall, though not the level of positive attention. The average attention devoted to it was three seconds, which dropped to one second during peak shopping periods.
When it comes to in-store, there’s hardly any measurement. This is a bit like a drunkard who is searching for his lost key under a lamp post just because there is light.
BV Pradeep, Unilever
According to Mesh Experience president Fiona Blades, brands could also use the technology to compare different in-store advertising formats, brand-building versus promotional messages, seasonal versus occasion-led displays, cost per engagement and the locations and days of the week when a display performs best. At present, the technology becomes less reliable if there are more than eight faces in the camera’s view, while distance also affects accuracy. It needs to be able to see the shopper’s eyes in order to record the data effectively.
Pradeep argued that marketers should be putting more effort into developing similar methods of measuring in-store assets, contrasting the relatively high engagement level of shoppers with digital audiences, where catching the attention is increasingly difficult.
“When you look at the in-store environment, there are people coming there very clearly in a mindset to buy certain product categories, buy certain brands. They are more open to receiving messages and they are more open to engage with communication and the impact it can create. However, we don’t seem to be focusing on developing any metrics.”
Blades told Marketing Week she hopes similar trials can be conducted with retailers in the UK.