Much of the conversation at Connected Commerce Summit London 2026 was about winning the digital shelf: being found, ranked and recommended. It's the right priority. But every route to purchase ends at the same place, a shopper looking at a pack, and a question the industry still answers with surprisingly little agreement: will this pack get noticed?
Ask five people in a CPG business and you'll get five answers: instinct, a focus group, last year's sales, a creative director's eye, a hunch about what the competition will do. Each has a place. None is a shared, evidence-based way of knowing, and the cost of that is real: packs are signed off, printed and shipped on opinion, and the learning happens on the shelf, when it's most expensive.
It doesn't have to work this way, because attention isn't a matter of opinion. People respond to visual information in consistent, well-understood ways. From a distance, the eye is drawn to what stands out from its surroundings, before anyone has consciously decided to look. Closer in, attention shifts to navigating the range and working out which product is which. In the hand, it's about the physical thing itself. Different stages, different processes, but the same ones for almost everyone.
Two things follow from that. The first is that attention is never absolute. A pack doesn't compete in isolation, it competes with everything around it. A redesign can look excellent on its own and still change what stands out beside a rival, sometimes in the rival's favour. You only see that by assessing the pack in context, and it's almost impossible to spot by opinion alone.
The second is that packs are physical. Those of us who work in brands tend to think in concepts: the idea, the positioning, the artwork on a screen. Shoppers respond to what's in front of them, with its shape, its scale, and the way it sits next to its neighbours. A pack that works in a presentation can behave very differently on a shelf.
If the brain responds in consistent ways, the way we measure attention in pack design should be consistent too. Today it isn't. Methods differ, definitions differ and results rarely compare, which makes them hard to trust and harder to act on.
A shared standard would change that. It would need to be grounded in how people actually see, rather than how we'd like them to. It would need to assess a pack in context, against the competitors it will really sit beside. And it would need to be common enough that a brand, a retailer and a researcher can look at the same result and agree what it means.
That won't be built by one company, and it shouldn't be. It needs brands, retailers and researchers to shape it. We think it's time to start, and at Dragonfly AI we want to help get it moving. If you'd like to be part of it, get in touch.