The Fast, Accurate, and Cost-Effective Pack Design Solution

Matthew Salem:

Hi everyone. I’m your host, Matt Salem. And you’ve tuned into another episode of Our Best Behavior, brought to you by Behaviorally. Behaviorally, the global market leader in evaluating shopper marketing helps brands define and diagnose the behaviors that drive shopper growth. Each month we share industry insights on trending topics designed to help clients make better shopper marketing decisions.

Today we’re joined by Nisha Yadav, SVP Omni Shopper Lead at Behaviorally, who will talk to us about Flash.AI, a new and revolutionary way to test pack designs. Nisha, great to have you join us today.

 

 

Nisha Yadav:

Hi Matt, glad to be here. Thanks for having me.

 

Matthew Salem:

My pleasure for sure. And we’re going to talk about Flash.AI today.

So, let’s just get that conversation going. It’s been almost a year since the launch of Flash.AI. And just curious to hear from you. What’s that year been like? What’s it been coming to this point? And what are you seeing for the future? And I know that’s a lot, so we could just kind of break it down into how that past year has been first and foremost.

 

Nisha Yadav:

Yeah, Flash.AI. Isn’t it amazing? It’s one of our most revolutionary new products. Uh, and it’s a way to test pack design. Leveraging the power of AI and image recognition and Behaviorally’s unrivaled database of consumer behavior data on pack designs. Uh, it’s something we came up with about a year ago, and clients are absolutely loving it.

We have tested hundreds and hundreds of different packs across many different categories and across all of the different offices that Behaviorally operates in. And in each time, we’re able to do this in a matter of minutes. It’s, it’s been really great, and our clients are really appreciating all of the benefits that Flash.AI has, which are namely: accuracy.

So, 85% of the time, uh Flash.AI technology is able to give you results that are similar to results you would get if you were doing a survey with consumers, and it’s able to do this very quickly and very cost-effectively.

 

Matthew Salem:

Very cool. And I like that it’s also been executed globally. It sounds like when you say ‘across all of the offices that Behaviorally operates in,’ that means around the world. Correct?

 

Nisha Yadav:

Yes, that’s right. We have offices across Europe and across Asia. In fact, we just opened one in Australia. Very recently.

 

Matthew Salem:

Very cool. I’ll be hopping on over there for all my Australian friends if you, uh, catch me between the lines. So, when we think about, when we think about the global aspect of it and the database that it’s linked to, can you tell me a bit more about how Flash.AI works? How it links to the database? And is that database being used in a way that we can really get down to global or regional differences? Or is it more at a category level and leveraging all of the data by category?

How does Flash.AI work? Let us peek under the hood a little if you will.

 

Nisha Yadav:

Yeah. So Behaviorally has the world’s largest database of pack designs. We’ve been in this business for the past 50 years, and over these years, we’ve collected more than 18 million consumers’ impressions of pack designs.

And that’s all stored in our database. So, what we did is we engaged AI to take a look at that data. So, take a look at the images themselves, as well as the data related to the images. We broke down the images pixel by pixel into data. So, all images at the end of the day, when a computer looks at them, it is data. It’s ones and zeros.

So, we broke it down pixel by pixel into data and then found relationships between pixels of data and data that we got from the surveys. And we use this information to create a network. And within that network, we are then able to test a new image and see where it fits within that network. And so that’s why we’re able to do this globally. Across all of the different categories in which our clients operate and look for differences, uh, within a category as well as within different regions.

 

Matthew Salem:

Perhaps said another way and asking you to confirm that I’m correct. If I had a package that I put into Flash.AI, and this package had 50% of the space taken up by its brandmark, 25% of the space taken up by its main image, and then the use of different colors in the background, let’s say. Would this technology see that and essentially say, here are other packs that are roughly 50% brandmark, 25% image, and similar colors. And how do those perform? So, we can predict how this design will perform?

 

Nisha Yadav:

Yeah, and it doesn’t just look at colors and logos and the design on its own. It looks at them collectively. You know, don’t think of it as being something that is restrictive. That just because it is AI, therefore you need to follow a formula. Because in fact, because it is AI, it is able to look at various combinations of color and logo and design and come up with multiple different ways in which this could have played out in terms of response that consumers may have to it and tell you how best your design will perform within those different parameters of the network.

 

Matthew Salem:

Very cool. So, what are the ways that you’re using the technology and applying it? What are some of the use cases with clients at Behaviorally?

 

Nisha Yadav:

Yeah, so we designed this primarily to help our clients get some data in early stages of a design sprint. So oftentimes, when clients are in their early stages, we find that clients are, you know, making decisions based on gut feeling. They’ve been doing this for many, many years, and they have a lot of experience in it. So, there is some, you know, instinct going in there and instinct is really good, but sometimes, you know, you do want data to back up your instinct, especially if you want to move in a particular direction and there could be certain biases across a team of stakeholders.

So, you might want to get some data to help inform your decision and move forward. You know, we would hate for our clients to reject an idea that is potentially brilliant, but because they don’t have data or because there are certain biases, they kind of leave that on the cutting room floor.

And similarly, we wouldn’t want them to proceed with something that is really terrible. And so, we encourage our clients to do some early-stage testing with very rough designs and then pick the ones that make the most sense and advance with those. So that’s what it was designed for.

And what we’re finding is that while that is a really good application, we also have other applications like competitive intelligence and benchmarking, where it’s proving to be very useful.

 

Matthew Salem:

Interesting. I love the initial idea of how to use it because oftentimes, in my experience with clients over the years, there are these meetings where they all get together and decide, okay, from the 10 or 15, 20 different designs that we have from our agency, including iterations of larger systems, et cetera, we’re going to take these four or these five.

And it’s very much a decision made at that round table, without any data without any support. So, the idea of bringing data into the game is phenomenal. What are some of the other ways that you’ve seen it kind of develop over time? Because it seems that that was the initial use case for it, but that others have naturally case to be?

 

Nisha Yadav:

Yeah. You know, uh, that image that you just painted of the, of the conference room, table and clients, um, making decisions based off of their instinct, that’s really what it was designed for. And oftentimes, you know, it was a little bit prohibitive to test things when they were at that stage when they were, you know, in a very early stage, because uh, would it make sense at that early stage to spend time, money and budget that could be spent somewhere else? Does it make sense to do that at an early stage, or should we narrow it down? And so, you know, we’ve come up with something that is efficient, it’s effective, and it is time-efficient, you know?

And, and so we’re able to give them a solution for something early stage, but we are also finding that for competitive intelligence, for instance, this is a very useful tool. So oftentimes, competitive intelligence doesn’t get the level of budget it perhaps could get because who wants to spend money on learning about your competition when you could be spending that money learning about your own brand?

And so, we have a tool that is very cost-effective to learn how your competitors’ backs are doing compared to your own, to understand where we stack up compared to the competition.

 

Matthew Salem:

Yeah. I love that idea, and you’re absolutely right. Clients are going to be interested first and foremost and understanding their own brand, their own designs, and putting money, time, and resources toward those evaluations. So, for us to be able to offer our clients an ability to take a peek at competition in a quick investment-friendly way and understand across KPIs how the competition is performing, it really can provide extra insight net value to any given client initiative.

 

Nisha Yadav:

Yes, absolutely. And another way to think about this is benchmarking. So again, at the beginning of a design sprint, things are moving very quickly, and you may want to know where your design is stacking up compared to the competition, right? And then provide direction to your agency on areas that you want the design to improve on.

And so, benchmarking is another area where we’re seeing a lot of clients use our technology because we can very quickly, sometimes in a matter of minutes, just benchmark your pack design against the competition and give you direction on where to take it for the design sprint.

 

Matthew Salem:

Interesting. Curious what about when you think about early on usage, kind of going back to the initial use case and you think about concepts and concept development and literally the accompaniment of a packaging visual on a concept statement page.

Is there a case for it to be used in that application? In other words, when you have your concept statement, clearly, the verbiage is doing a lot of the driving, if you will, but there is the idea of that complimentary visual, which does have some sort of impact. So, is there opportunity for organizations to better their concepts upfront their concept statements by leveraging Flash.AI for the accompanying visuals?

 

Nisha Yadav:

Yeah, for the accompanying visuals. Definitely. So, if there is a visual aspect of the concept that you would like to test, Flash.AI is an excellent technology to give you that information in a matter of minutes.

Now, if you want to test the concept statement, which means reading those words, Flash.AI is not the tool for that. There may be other tools that are also AI-based natural language processing tools that might give you some information on that.

Flash.AI is designed for images. It’s designed to look at images and how they play out within the context of a network of thousands and thousands of images that Behaviorally has in its database.

And to give you a sense of where that image would benchmark, it’s not designed to look at words and how those words are processed. And that idea is processed by people. There will be different tools that do that.

 

Matthew Salem:

So, in any of these given use cases, whether it’s early-stage screening, competitive intelligence, benchmarking, what does that deliverable ultimately look like?

What are our clients going to see when we deliver Flash.AI results in a matter of minutes at a very cost-efficient investment?

 

Nisha Yadav:

We would provide a visual dashboard of how that pack design compares to the network or to the database of the different pack designs that we are comparing it against. So, let’s say we are doing a comparison of a pack design to, um, the, uh, hot beverages category.

We will take that pack design, and we will load it onto our database. And within a matter of minutes, literally, the AI will read that image, and it will tell you how that image performs on a number of different metrics that we have been gathering over time, including personal relevance, what that beverage might taste like, how it will play out in terms of being found on a shelf and how it will play a play out in terms of a person wanting to purchase it. So, a number of key metrics that our clients have used as their KPIs for many, many years are included within this database.

 

Matthew Salem:

And does Behaviorally offer any additional perspective, even if it’s anecdotal in nature, to compliment them?

 

Nisha Yadav:

Yes, of course. I mean, um, data at the end of the day is data, and sometimes it’s helpful to have our consultants provide advice on how to read that information and, you know, what to do with that information. And so, it definitely comes with an overview of the learnings, what they mean and how to interpret that information and how to move forward.

 

Matthew Salem:

Well, it sounds awesome. It sounds extremely cutting edge. I love the use of AI technology, and I think AI is just going to become that much more important day-to-day in all of our lives and our business lives, and our personal lives. So, it’s great to see Behaviorally on the cutting-edge using AI technology and its tremendous database.

So really excited to see what this tool continues to bring to our clients. How can clients find out more? What should they do, Nisha?

 

Nisha Yadav:

Yeah. So please visit our website solutions page and look at our latest blogs and contact us today. And you can contact me personally, or you could reach out and have your teams have a capabilities presentation done for you. We would be happy to do that.

 

Matthew Salem:

Very cool. Nisha, I’m going to put you on the spot now. If you had the capability to leverage AI for your own personal use in your day-to-day life, what would the application be? How would you use AI to help you?

 

Nisha Yadav:

I think it’s amazing that we have a technology that is so similar to the way AI is being used already.

I don’t have a Tesla, personally. One day, hopefully, I will have one, but the way we use AI at Behaviorally is very much the same way. Elon Musk’s uses AI to build his amazing Tesla cars. And so, I’m just excited that our company is able to play in the same playing field as some of the biggest tech players out there.

 

Matthew Salem:

Very cool. So, you would use it in some way, shape, or form to help you drive, is what I’m hearing.

 

Nisha Yadav:

Yeah.

 

Matthew Salem:

Okay. All right. I would probably use it to help me pick out the perfect cigar, but I digress.

 

Nisha Yadav:

That sounds, that sounds like a good application as well.

 

Matthew Salem:

Nisha. Thanks a lot for coming on today. Really appreciate it. It was great to have you. Thanks for talking about Flash.AI. I’d also like to thank our audience for tuning into Our Best Behavior brought to you by Behaviorally, Nisha. Thanks again. And we’ll catch you next time.