The Behaviorally Blog

Marrying AI and a Robust Database to Innovate Testing of Early Pack Designs > Behaviorally

Written by behaviorally | Apr 14, 2021 8:00:00 AM

Yesterday, Behaviorally (formerly PRS) launched Flash.AI™, a new innovation designed to provide rapid iterative metrics for pack design performance early in the design process. Flash.AI joins the other tools in our Define suite of research solutions (PackFlash® and ShopperFlash™) which identify and validate the elements that influence consumers along the physical and digital path-to-purchase.

We prioritized the development of Flash.AI as part of our innovation pipeline because clients kept telling us about a genuine unmet need. They had shared their concerns that pack design alignment between brand managers and their design agencies, particularly in the early stages of design development, often lacks the voice of the consumer, resulting in packs that fail in the validation process.

Developing Flash.AI has been very important for us because it really marries two things to which Behaviorally is deeply committed in our strategy and new brand promise: a digital-first approach to uncovering shopper insights and leveraging our legacy of experience in packaging and shopper marketing in order to drive growth for our clients.

That legacy of category expertise is more than just a body of institutional knowledge expertise shared intellectually by our staff of researchers. In the almost 50 years since we were founded (as Perception Research Services), we have observed over 18 million shopper behaviors and decisions. From that, we distilled a database that includes KPIs for predicting performance derived from what consumers saw as they evaluated packs on shelves in simulated shopper environments.

Parsing a database that big at scale requires the kind of robust analytics only available using AI and machine learning. The interesting twist we have added in Flash.AI is that through application of computer vision, we are able to achieve image recognition of new packs we test.  Literally we are training machines to “see” the pack designs as humans do and compare them in split seconds to all the packs we have tested as they were “seen” by human respondents! And because the AI analytics are also parsing the text of the KPI metrics, we can rapidly score packs against multiple criteria for performance. Flash.AI can evaluate packs for new products, restages and even competitors’ products with category specific recommendations for optimization.

Now here is where it gets interesting. Because we can do this unbelievably quickly (more on that later), we now have an innovation that can serve that previously unmet need for brand managers and their design agencies at the early stages of development. In the rapid-fire moments of iterating and refining during the pack design development process, there has never been time to test and predict which elements will drive findability, convey efficacy or ladder up to support brand equity. Brand managers and their design agency partners relied on gut instinct rather than voice of the customer, often finding out too late in the validation process of a “final” design that they didn’t have a winner. Having a cost-effective, rapid testing solution that is backed by a reliable database of metrics and can fit seamlessly into the early design development process will revolutionize the odds of creating winning pack designs with greater predictability!

So back to this notion of “fast”. Often in the ideation sessions for developing a pack design, teams might work on multiple iterations, refining and modifying designs as they solicit feedback from stakeholders across the organization. There might be multiple iterations in the course of one day.  So, in order to even be useful, tests for performance predictability needed to also accommodate speed and iteration.

When we originally looked at how fast we could turn around results, we framed this internally as “in less than a day” or “within hours”. Our CEO challenged us that in order to seamlessly integrate into the design development process, we had to go back to the drawing board and do better.  We torture tested the solution and we can happily report we can deliver results in a dashboard within 120 minutes. All without launching a single new survey!

What Flash.AI represents to us is really three things:

  • Innovation: leveraging digital cutting-edge AI technology and machine learning to uncover predictive analytics in early-stage pack design.
  • Ingenuity: inventing a way to access the power of our expertise and database of pack and shelf designs to create a truly agile insights tool.
  • Client First Orientation: by focusing relentlessly on the challenges, our clients face making shopper marketing more effective, we can build new solutions that genuinely help them drive growth in their businesses.

We are excited to share results from our early adopters and welcome inquiries on using Flash.AI to drive performance in early pack design development.

THE AUTHOR
Ruben Nazario is the Vice President, Project Lead on the innovation team at Behaviorally (formerly PRS). He has been responsible for bringing the OmniPath® and Flash.AI™ solutions to market to help clients decode the complexity of consumer behavior.
If you would like to talk with Ruben about how behavioral design can help your product perform optimally digitally as well as in brick-and-mortar retail, you can reach him on Twitter @RubenDNazario or connect with him on LinkedIn.