For many businesses, the choice of heatmapping tools available, providing predictive performance insight across their marketing campaigns, can be overwhelming. With many tools offering solutions that provide data to help inform and optimise marketing or creative direction, it’s a tough choice to select the heatmap tool that is right for your business. Unlike many heatmap tools out there, Dragonfly AI uses a blend of AI and neuroscience to show instantly want your audiences see first, across any physical or digital channel. In this article, we take a deep dive into the science behind our technology and to explain the value it adds to our brand partners…before they launch their campaigns.

How is Dragonfly AI different from other heatmaps?

Image 1: Scrolling Heatmap

Image 2: Predictive heatmap

Heatmap data visualisations are generally used to illustrate historical data – in eCommerce, for example, this data might normally include mouse clickpaths, hovers and scrolls. This type of basic heatmap is known as a descriptive heatmap as it describes an event that has already happened.

At Dragonfly AI we offer a different kind of data visualisation, we create what we call predictive heatmaps which show you where the attention of the human eye will be fixed in the future, i.e. when a potential customer sees your product for the first time. Although descriptive heatmaps are useful to provide a broad sense of engagement, our predictive heatmaps and metrics help you to iterate, improve and optimize any content for human attention (from product and packaging, to instore, online and above the line) all before your content is launched into the real world.

How does the algorithm account for different demographics?

Our AI is based on a biological algorithm that simulates the way the human brain subconsciously (system 1 thinking) prioritizes visual information before the conscious mind (system 2 thinking) kicks in. This biological analysis is universal to human beings, which means that our predictions apply to any audience you might be targeting. Our algorithm models the pre-cognitive phase in the brain and predicts where consumers will look before their preferences (emotional response, ‘likes’ and ‘dislikes’) and other learned behaviours start whirring into action.

So, we can show you exactly where a customer is likely to look first, and we can apply our analysis universally across all visual media (digital, static, instore, POP and video). Our heatmaps can also be applied direct to your physical product and packaging, plus once the product is optimized for human attention, we can test how it performs contextually, on say, a shelf in a physical supermarket aisle, or how it performs on a web page, or in a video. Essentially, if you have something that you want customers to see, our heatmaps will show you exactly where they’ll look and if it’s not in the right place, we’ll show you how to iterate the design until it’s totally ‘human ready’ before you commit to publishing your content.

What do the numbers mean?

Our AI measures the visual saliency of content.  Something is considered salient if it is easy to spot. The more something sticks out and grabs your attention, the more visual saliency it has. Our AI determines the saliency of each element in a creative execution, it then quantifies the specific saliency and visualizes it in the form of a heatmap, showing you clearly which elements of the design people see first (red – hottest) and what they miss (blue – coldest). Our AI scores saliency in a range from 0 (no salience) to 100 (the highest salience) to give you specific, quantified metrics for any given area of your image.

If you want to understand particular areas of Interest (AOI) such as a promotion or health benefit in packaging; or how well your product SKU is going to perform on Amazon against a competitor, we can score those areas for you.

Dragonfly AI can also tell you if your brand name receives attention ahead of your key visual, or if a ‘Vegan option’ banner is stealing the show. We can also calculate probability of perception (PoP) which helps assess visual communication hierarchy and understand which key elements of your content are being seen by a customer in the first “zero moment of truth”.  Our share of attention (SoA) metricpredicts how 100% of a customer’s eye fixations are distributed across your content.

To summarise, probability of perception focuses on one fixation; share of attention analyses all fixations; and together they help you to optimize the design and hierarchy of all of your areas of interest.

Does your AI use machine learning?

Machine learning (ML) is a common technique employed for training AI. Let’s say an eCommerce site wants to build an AI to predict what product a customer is likely to buy next; ML trains the AI to file through all of that customer’s previous purchases, aggregating the data, identifying commonalities, then predicting from these assumptions. ML is prone to bias towards historical data that might no longer be valid.

Dragonfly AI, by contrast, foregoes ML and instead employs our own bespoke, coded biological model which simulates how the human brain processes visual information, i.e. it prioritises key visual elements based on their saliency in the same way that our brains process visual stimuli. Our technique allows us increased explainability over ML – that is to say, we can explain clearly how and why our AI reached a conclusion, and we can show you the exact drivers behind it (like larger font size, more colour, higher contrast, increased brightness, etc). ML is a bit harder to explain as it’s typically a black box of data that doesn’t tend to articulate how it reached its conclusion.

Isn’t this a blinding glimpse of the obvious?

Dragonfly AI provides invaluable visual data that can validate your current design as well as revealing inspiring new insights. Our predictive heatmaps often catalyse constructive discussion and debate between stakeholders and designers; and brands and their creative agencies leading to increased creativity and the generation of more efficient designs.

Dragonfly AI also takes the guess work and subjectivity out of the creative process showing what consumers will see.  Sometimes our heatmaps can reveal entirely unexpected data, showing that a certain element of your design is distracting from your core brand messaging, Dragonfly AI equips you to address message sequencing. It also shows you how and where to improve consistency across all of your media and platforms by creating heatmaps for all of your different designs, from digital pack shots using our Chrome extension, to instore point of sale using our App – all providing instant insight in real-time. Helping you to improve and optimize all your collateral until your communications hierarchy and design goals, are all met.

Interested in finding out more about how Dragonfly AI can help you optimise your content?

Similar Articles


How to improve the shopper experience with predictive visual analytics

The use of predictive visual analytics processed by Artificial Intelligence (AI) allows brands and retailers Read More

8 September 2021


3 ways to optimise your digital fundraising

To suggest that 2020 was a challenging year for charities is something of an understatement. Read More

20 August 2021


How To Use AI To Inform Design

AI, or artificial intelligence, is quickly sweeping across every aspect of our lives and businesses, Read More

3 August 2021