Expert tips to unlock behavioral AI KPIs in pack testing

AI-powered insights promise speed and scale, delivering outputs filled with behavioral KPIs and performance metrics. They’re great at identifying patterns and surfacing anomalies - but not everything that stands out matters, and not everything that matters stands out.  

Without human interpretation, contextual knowledge, and industry experience, AI can mislead what is the most important - action and decision making. It takes real multidisciplinary expertise and cross-industry knowledge to understand the “hidden insight treasure” behind the numbers and turn data into decisions that actually move the needle.

That’s why I put together this short step-by-step tutorial based on our Predictive EyeTracking, AI-powered methodology we leverage in our Pack Testing solutions. Its goal is to clearly differentiate between the various KPI outputs - but more importantly, to show you how to move beyond the numbers and uncover insight that becomes truly actionable when combined with your expertise.

The Fabulous 4 of pack AI KPIs: Visibility, Attention, Scan Path, and Heat Map

This is an example of Design 1 and the predicted results you can get. You can learn which areas are the most visible and engaging, as well as the most probable order of noticing different pack areas.  

• AI Visibility shows you which pack areas are most probable to be seen (shown in predicted % of consumers).  

• AI Attention highlights which pack areas are most likely to engage consumers (shown in the predicted number of seconds spent there by those who have seen the areas).  

• Scan Path points out which predicted selections of the top 4 zones are noticed the fastest and their ranking.  

• And AI Heat Map is a visual representation of the most visible and engaging pack areas.

However, these numbers will get their true value only once they are compared with the KPIs from another design with a different scan path route and looked through the category and the research goal lenses. It goes without saying, the more in-context data and comparisons, the better, as you always want to compare with other potential versions or competitors.  

Know what and why to compare when working with behavioral AI KPIs

It is no secret that every insight is first and foremost the answer to the on-point question, one that originates from the brand’s main tactic goal. If, for example, we compare two packaging, we want to understand how the pack elements compare across designs and which arrangement works best. Again, what works best isn’t a universal truth - it depends on your specific goals and what message you want consumers to engage with.

In this case, the Scan Path reveals that Design 1 helps product benefits to be noticed earlier, while Design 2 leaves one important benefit out of the initial browsing pattern. Brand name and product image are noticed the fastest on both designs, seen by ~90% of consumers.

Considering cereals belong to the food category, this might be exactly what you want. However, let’s bring the context of shifting consumer behavior to the picture to make the whole story a bit more interesting.

Have you heard of a growing trend called health span planning? In short, health span refers to the number of years a person lives in good health, and the term is becoming a major focus in consumer behavior, shifting the conversation from simply living longer to living better.

This mindset is driving demand for products that support long-term wellness, such as personalized supplements, functional foods, sleep and stress aids, wearable health tech, and even longevity-focused skincare. Products that can demonstrate real, measurable contributions to long-term health will stand out in this evolving wellness landscape. And in the packaging world, those kinds of benefits are usually communicated best with on-point claims that shoppers can see.

So keeping that in mind, let’s see what behavioral KPI outputs are telling us about the visibility and engagement of those two products’ benefits. Product benefits placed closer to the brand name and higher on the pack on Design 1 are more visible and engaging compared to smaller and lower-placed elements on Design 2.

Bonus tip: AI or not, behavioral KPIs work best in a mixed-methods approach  

At EyeSee, behavioral precision and the confidence we find in numbers are always fundamental to our research—but they’re just the starting point. We are deeply context-aware and consistently strive for nuanced insights and actionable recommendations. The same principle guides us when our behavioral insights are AI-powered.

When we combine our Predictive Eye Tracking methodology with MaxDiff (which reveals which pack commands the highest preference), Click Tracking (where respondents identify what they like and dislike), and a well-crafted traditional survey (did you convey the key messages, is your communication well received and understood?), we significantly expand the range and depth of our KPIs. This integrated approach goes beyond AI-based visibility, attention, and fixation order to include - preference rankings across designs, liked and disliked elements, purchase intent, likability, and relevance.

In conclusion

While AI-powered behavioral KPIs like Visibility, Attention, Scan Path, and Heat Map offer a fast and scalable way to assess design performance, their full potential is only unlocked through expert human interpretation. Numbers alone can’t account for context, brand strategy, or shifting consumer expectations. That’s why mixing AI with proven research methodologies - like MaxDiff, Click Tracking, and traditional surveys - with experts' category knowledge and shifting human behavior, is essential. This blended approach helps transform raw outputs into predictive, actionable insights that drive smarter decisions. At the end of the day, it’s not just about what consumers see (although what is not seen cannot be bought), but it’s about understanding why it matters and what to do next.

Eager for more? Check out Shaping pack design with AI Insights: Kellanova’s case study.