Design That Sticks: Applying Neuroscience to Predict Attention

Welcome back to the digital marketing podcast brought to you by targetinternet.com. In this episode, we have a fascinating conversation about applying neuroscience to predict attention. So basically, how can we use neuroscience to help you create higher performing digital products and assets?

So in this episode, I am joined by Kramer Reeves, who's CEO of iQuant, and Professor Peter Korik, who's one of iQuant's founders and a world-leading expert in neuroscience.

Now, if you're not familiar with iQuant, they take the concept of an eye tracker study. That's where we get loads of people to look at the same content and then try and create a heat map of where they actually look at that content to work out what's cutting through, but then using AI to really effectively predict where people look instead.

So the idea is you don't need to spend hours, days or even weeks collecting all that data, getting people to look at something, but actually can predict to a really high level of accuracy where are people going to look at something, being a web page, a social media post, a design or anything else.

There's some really brilliant insights in here from Peter about how our brains work and how actually we're really poorly adapted to interpreting stuff that we see on web pages.

But what's brilliant as well is they've set up a two-week exclusive trial for listeners of the Digital Marketing Podcast. So the theory is brilliant. There's some really great insights, but then you can go away and you can test this out in practice. And I've played with it and it is phenomenal.

So if you go over to targetinternet.com forward slash podcast, you will find it there in the show notes. But without further ado, over to the interview.

Okay, so I am here with Peter and Kramer. And let's start off with why do you think so many well-designed webpages still fail to convert and to deliver?

I think we have to step back and consider what we are built for. And my take is we are concerning our brain and visual system hunter-gatherers. We are optimized running around in the woods. You always know where you are, whether you are in a hut with your tribe in the woods hunting a wild boar or looking for an eagle.

And when you click on a website, bang, something new is in front of you completely. So you are out of context and you have to orient yourself. And this is very, very difficult, essentially.

And to top that, the designer knows this website very well. And I believe these are great websites, but you look at your website, which you know very well, with very different eyes than the end customer, for which everything is surprising.

And so far, different mechanisms in our brains kick in, and the different mechanism kicks in in your brain designing it, compared to the one who is looking at it the very first time. That's an interesting thing. This out of context idea is really interesting. So when someone lands on a page, what's really happening in those first few seconds then?

You try to orient yourself, which means that you look at the page and you are input driven. Let's say it was surprising to me when I learned about it. Our brain is mostly talking to itself. It is not sitting, idling, waiting for the input finally to arrive.

To give you some numbers, there is roughly 1 million fibers from each eye to the brain. But the two hemispheres we have are connected by 200 million. and connectivity inside each hemisphere is even more intense. So there are several orders of magnitude difference between how much information we get from the outside compared to what is going on inside.

You might compare it with a huge discussion room, 100 people in the Congress, and one... external is joining the party and tries to communicate a message. This is really a hard task.

And to guide this process, what to attend to, are typically very simple stuff. So, for example, traffic signs are bright red and white in order to catch your attention. And you have to use such methods in order to establish a context, guide the viewer what it is all about, and guiding along the story you want to tell.

Kramer, give us some kind of business context for how iQuant is kind of helping with this then, because I think it's really interesting. I've kind of introduced what iQuant is, but I think it's good for people to see how it relates to this in a business context.

Sure. We go back to this key point that many of the designs are built based on aesthetics, but the brain, as we're talking about, sometimes just can't actually convert that or put the context together to drive the user to the action that the designer wants. So what we've done at iQuant is to model out what the eye actually sees. We've created models so that we can predict...

what the user will see and likely do, and we've put that into our platform. So you can test your user experience, your creative, your ad, whatever it is you want that person to see, you can test that in advance. The first thing that every designer needs to do with their creative or deliverable is we need to make sure somebody sees it. What do they see and how do they see it?

And that's what iQuant does. Before the actual end user sees it, we can predict what they will see and how they will see it. And then we provide all kinds of data about that particular prediction at analysis so the designer can adjust and optimize before they put the creative or deliverable into market.

It leads really nicely in as well, actually, because, Peter, you often talk about behavior happening below conscious awareness. What does that really mean in a digital context?

In our research field, consciousness is the big question. Everybody's buzzing about it. It's an enigma. But when we are honest, we do a lot. And consciousness is like on the passenger seat.

nodding, saying this was correct, this was correct, or saying, oh, what did I do just now? So it is like after the fact and from our evolutionary origins, we have to act quickly.

And this does not allow for huge deliberation. So there's a concept of System 1 and System 2. System 1 is doing the stuff. And System 2 is on the passenger seat, can do in-depth reasoning, deliberation, and all that stuff.

But most of the action is done by System 1 very quickly. And so far, when we are in discussions, how do we do it? So a system two is overrepresented. It feels super important.

It always wants to be in the spotlight. But sorry, system one is a real one. The soccer player who really scores the goal and shoots it is present at the moment when it's important.

The other thing that I've kind of come across a lot and you talk about is saliency. And I think that it's not something a lot of people necessarily be familiar with. But why does it matter so much for design as well?

Wonderful question. Saliency is a property of the image to attract attention, to be selected for focused attention and to be acted apart. which means, as it stands out, there is not a single color which is salient per se. So if you could see the video right now, I have wonderful neon green glasses in front of a pale kin and a pale office environment.

They stick out. If I would be, for example, an alien with a green hat, the green glasses would, well, yeah, there's a strange green alien. And it has green glasses, but the glasses would not be that remarkable. So it's a property of the image which captures your quick attention.

And there you start your analysis. So it is quickly guiding and thereby often predetermining what you will do at the end. So for system two, to change what's going on with this quick reaction is much harder than when you are already on the correct road towards the correct goal. Therefore, salience, properties of the image, of your visuals.

which determine how the uninitiated viewer is processing the visual is super important. And this is what the model is doing. It is, so to say, predicting which parts of a visual are attracting how much salience.

And, for example... A typical sequence which we see is you make a great ad, which is nice, aesthetic, and your boss comes in and says, yeah, but there is Black Friday. We have to add 50% reduction here. And then he comes in two hours later and says, yes, but we have too much stockpile of product.

Why? We have to advertise that to get rid of it. And so you overload your ad with a lot of stuff, and this dilutes the salience. And you leave it to chance, where the viewer looks first, and essentially you leave it to chance what's happening.

And so far, this is something we often see, and an objective model exists. which, so to say, models the viewer and is ignorant of, well, your plethora of wishes, everything you want to communicate in parallel, is a real good base check and leads to much better, much more focused advertisements.

That's a great example. So what are the common mistakes that teams are quite often making when they're trying to guide attention? They are overloading, to say, too much.

And understandable because you as a designer, you know it all. You want to blurb it out in parallel, but you know your edge. When you are accustomed to it, you don't find it that much loaded. But the other one who is looking at it for the very first time finds it super crowded.

And our model tells you that. Then there is the clarity and the storyline, where you start. That is the first fixation from where you move on. So if you, let's say, the click here button, of course it's important.

You want to have a nice salient click here button. But if you start with that, and I see a webpage where the main message is, click me, click here, I swipe left and go somewhere else. And then so far, you also have to think about the storyline.

What is viewed when? And adjusted accordingly. So with that in mind, what kind of role do color, contrast, visual hierarchy play in what users actually notice? Because obviously I can see that's going to be highly connected.

Yeah, so here you are asking for a simple recipe. Let's say, again, how we are built, we would like to interact with what is close. So if you have the visual effect that something is close,

Most primitive is if it's large. So if something is large, probably it's close. We pay more attention to it because everything which is far in the distance, well, we have some time to adjust our reaction to that.

And then, of course, I mentioned color. Also here, color contrasts are more important. So that my green glasses, they stick out of my face because I'm not an alien from Mars. They are super salient and you will start looking there and seeing, well, this is a nice guy.

Do I trust him? Maybe. And from there, the story unfolds. And similarly, motion cues are very strong. But you have to think about you cannot give first priority to five items.

Right. Would be nice, but doesn't work that way. So that leads us quite nicely because you've mentioned that visual confusion can reduce conversion. What does that kind of look like in reality?

So we have the measured clarity, which is the opposite of visual confusion and determines how clearly, how unique the storyline, the full key of attention and the path is.

I want to... shortly discuss another issue. We are in modern AI times. So we have models, and originally people thought, well, as you are asking which features are important, and I mentioned colored and so on.

And we scientists feel proud if we can mention that. We have to admit end-to-end training models. surpass the performance. Which means we take human behavior, we take all the visuals, we train the model to make optimal predictions, and the models are even better than the experts.

Which is great. We use such models end-to-end trained for superior performance. The downside we pay is there are black box models. If you have a visual, it tells you all the clarity and the saliency and where the focus is, but it does not tell you why.

Therefore, we need, again, the second type of model as well, where we have identifiable features and can say, yeah, it is crowded because... And say, okay, something here, something there, and here on addition. This you have to change in order to help you.

This is like in school a teacher who says only, well, this was not a good answer. That doesn't help you to give a better answer. You need both. You do not need only the genius teacher.

You also need the teacher who is helping you to improve. It's fascinating. So for teams listening, what's something kind of practical, one thing they can go off and do to improve performance potentially?

To get an objective measures. We started with the designer looks at his own head with very different eyes because the brain is talking mostly to itself.

He sees his mental context in the visual context. But the customer at the end does not have this mental state. It looks at it for the first time very, very different.

And that's the second issue. The designer is accustomed to it. So take a well-known web page like eBay or something. If you see it the very first time, show it to your parents again.

They say, oh God, what's going on here? And they use it 200 times. Okay, then they know what to click where. But for us, in principle, for the uninitiated user, it's not really well designed, you must say.

And for here, objective measures, models which are not... trained to use the specific sites but behave like you see it the very first time are super helpful because they give you the reality check how your visuals actually look like.

So it's really interesting to me because so much of the work we do is fairly subjective and we talk about best practice and those kind of things. How do you think the role of neuroscience is evolving in design and marketing?

So neuroscience and science in general loves data. And so far when we are modeling, we have our measures, how good the models are, variants explained and all that stuff in order to get to the best models so that we know what we are doing. That's super important.

The knowledge about how the brain works can guide us. For example, I've repeatedly said the brain is largely busy with itself. The brain has mechanisms which are fast, unconscious, able to deal with sudden threats and new information. And we have systems which do deliberation.

But these are not completely separate, but because the first one feeds the information to the second one. And so far, you cannot only say, well, wait till the user really thinks about it and understands everything, because the fast system is selecting what is fed to the slow system.

And so far, neuroscience, or let's say cognitive science, helps dramatically in order to streamline these communication channels. What I explained with the brain is also an opportunity, because when you create a context, you can fit in what the brain typically considers as important in different contexts. For example, movement. If something is moving, well, there's action.

This is important. And we found that if you can have visuals only which are static because it's print press and there's just an image and you cannot make it move on paper,

the effect is still of moving images, of moving objects, which are photographed 50% of the real motion, which is quite a lot. So imagine you have a photo, a static photo of a car, compared to a video which shows this car moving towards you,

The static photograph communicates still 50% of the saliency effect. Interesting. Which is great. You get it for free without any video technique. And similarly, the contrasts, which are important, color, luminance, texture, and so on, guide the eye movement.

And this can help you to make better visuals. And so far, cognitive science is helping here a lot. Fantastic. So I'm really interested, Kramer, we kind of understand the theory, look at some really good practical tips of how we might apply this.

Talk to us about iQuant and what it does, who it's for and kind of where the product's going as well. Because I think since I was first aware of the product a number of years ago, there's been a huge evolution and things have really developed. So talk to us about that.

Okay, sure. So as we're talking about the core problems that we're trying to solve, the biggest problem is our clients and the designers out there have to go faster. They have to do it more effectively. So in a cheaper fashion, less expensive fashion, and they have to be more accurate.

So fast and cheap, accurate. That's what we're shooting for with our technology. Can we help our designers get the objective feedback earlier in the design cycle so that they can optimize and improve and put into market what's more important?

In addition to this, one of the key things, one of the key things we hear back from our clients is they are measured on outcome, on performance. How do we start with the end goal in mind? How do we start with the objective in mind? So we've incorporated into iQuant intent-based outcome inputs.

So the inputs start, what is your objective? Natural language, just write it and speak it. Tell our platform what you're trying to accomplish. And our model will then look at your creative, look at your design, look at your deliverable, whatever you want to test, and tie the analysis and recommendations to your intended outcome.

And that is important because if we don't know the end objective, we really can't test for the full outcome. point of what you're trying to build. We can model out what the eyes will see and how they will see it. But we've taken it a step further.

To answer your question, how has iQuant evolved? Over the last year, we've taken it a step further. And that's because of the evolution of AI and the accessibility through things like MCP, model context protocol, server integrations. We can bring in

the intent and tie that to what our analysis and our recommendations are providing so that's one of the most important changes that's been made over the last year it's really interesting because we had we did an interview of adobe literally a few days ago and they were talking about look we've got this data now we can bring it into one place and you can take things out there it way more quickly than you could ever before but the problem is if you spent months developing a campaign and you're very wedded to your designs You know, going back to what Peter was saying, actually, we need to be able to test those before we even get them out there in the first place and actually understand how people are going to react to this content.

So I think in this very busy, noisy environment where everyone's generating stuff using AI as well, actually making sure it's going to be effective is going to be so key to that. Exactly. If listeners want to go off and try and play around with this technology or test it out, what's the best thing for them to do?

Well, the listeners of your podcast are in good shape because we normally offer a free trial. We are going to extend for this group only to two weeks, a free trial of iQuant. It's super easy. We'll pop the link in.

You guys can get access to it and take it for a test run. And it's a great way to see if this is something you can include into your workflow. And this is a key point. You designers, we've been following in the design thinking process over the last 10, 15 years has evolved into this infinite loop.

We understand that you want to slot verification, pre-testing this type of system into your process, into your workflow. So whether or not you're in Figma, it's just a press of a button to test. Adobe, other tools. And so the best way to learn about this is to take it for a test drive.

Within about 15 minutes, you'll know, is this something that can help me? That's how easy it is and how quickly you'll deduce, do I get the insights I need to hit my business objectives? That's fantastic. We'll put that into the show notes.

So targetinternet.com forward slash podcast and you'll be able to find that special offer. I'll just say this with a real practical point of view. We don't have paid placements on the podcast. Whenever we have software on, people are a little bit suspicious of like, have they paid to be on here?

We absolutely don't. And I would say that iQuant, I've worked with an agency recently that we're using it. And it's made a huge difference to things because there's so much subjectivity. If you've got three or four expert designers, they all have different opinions and different aesthetic styles and things like that.

And actually what they are able to do is just take a bit of a step back and say, look, actually, this is what's catching people's eye. This is where actually they're looking in the first place. If we reduce down the amount of noise in this design, it's having a bigger impact. So very much coming back to a lot of things that Peter spoke about, but very much in practice happening really quickly as well.

Okay, Peter and Kramer, thank you so much for joining us on the Digital Marketing Podcast. Thanks for having us, Daniel. Thanks a lot. It was a pleasure.

For more episodes, resources, to leave a review or to get in contact, go to targetinternet.com forward slash podcast.