1-3 seconds. That's about all the time you get win a shopper's attention or get lost in the shuffle.

But what if you could predict what people are likely to see in that critical first moment of truth? And what if you had a way to evaluate how you - and your competitors - are standing out?

Heatmaps on Shelfgram let you toggle a photo overlay that predicts just that using the latest in artificial intelligence!

But first... here's a bit about visual attention models

In the last few years, visual attention modelling has started to move from research laboratories into commercial applications driven by major advancements in machine learning technologies.

Although visual attention models can exceed 92% of the accuracy of a human based-eye tracking study, they're not perfect. However, it's about the same accuracy as most small-scale human based eye-tracking studies because of the inherent inaccuracies in human to human variances.

Shelfgram Heatmaps use predicted human eye fixation patterns to highlight the distinct subjective perceptual quality which makes some items in the retail world stand out from their neighbours and immediately grab shopper attention. Our AI model is based on 330K images that 250k people looked at in an actual eye tracking study.

Quick tips

Here are few preliminary design best practices based on what we've noticed so far in Heatmaps:

• Faces (that aren't too far away) attract a lot of attention
• Less is more when it comes to battling visual clutter
• Contrast, shape, color intensity, and lighting are all very important
• If nothing stands out, attention defaults to whatever is at eye-level and shoppers default to reading left to right
• Beware of recurring patterns because they can create a camouflage effect that hurts attention

Heatmaps are now available to all Shelfgram users on both our web and mobile applications!