Your attention please: Using AI based solutions to help marketers hold people’s attention for longer

Click here to access the slides.

Click here to access the slides.

Transcript of recording with Hannah Mann and Mark Bainbridge – generated automatically by HappyScribe which means it will be about 80% accurate – if you spot confusing errors, please email The timestamps are included to help you jump directly to a point of interest.


[00:00:11.270] – Hannah Mann

Hello and welcome, everyone. Thank you for listening to our paper today titled your attention please: using AI based solutions to help marketers hold people’s attention for longer. We hope to bring some new ideas and inspiration about how AI and automation really can make an impact on your research right now. And it does not have to be complicated or difficult. So my name is Hannah Mann, and I’m the founding partner, or of one of the founding partners of Day One strategy, which is a research agency specialising in the health care sector.


[00:00:43.150] – Hannah Mann

And as we’re talking about attention today, we thought it fitting to start off by mentioning some of the things that we know are vying for our attention outside of work today. So for me, it’s definitely fashion brands, jewellery brands and dog rescue homes, which must say something about me. But presenting with me today is Mark Bainbridge. And I’m going to hand over to Mark now to introduce himself.


[00:01:04.390] – Mark Bainbridge

Thanks very much. My name is Mark Bainbridge on the co-founder of Dragonfly. We’re basically an artificial intelligence product, which simply put shows you what the human brain sees first content instance.


[00:01:18.220] – Hannah Mann

So what are we going to cover today? So first of all, the issues that the rise of the attention economy and what that is and why that has come about, then moving in to look at the tech that can help brands and researchers win this battle for attention, providing some live examples to help demonstrate just how easy it is to use. And then finally finishing up with a summary of why this all matters and the value that it can bring to the table in market research.


[00:01:44.770] – Hannah Mann

So starting with the rise of the attention economy and why this has come to be. Well, there’s been a definite transition over the last few years in terms of our ability to focus and concentrate on just one thing. In essence, our attention span is waning. So it’s getting harder and harder to hold customers attention and to keep it to demonstrate this point. A study by Microsoft conducted and concluded that human attention span has dropped from 12 to eight seconds in just a few years.


[00:02:15.130] – Hannah Mann

And yes, that is actually less than the attention span of a goldfish. So that really is something that we are bombarded by information all of the time. And our brains just cannot cannot cope with all this information. There is a limit to how much it can process on top of that with being reprogrammed to expect answers fast and easy. So we don’t expect to have to try too hard to understand our messages being directed at. And as just alluded to and might have noticed in your own lives as well, and in the modern world at large, it means that we are constantly bombarded by information, messages which is causing this problem.


[00:02:52.740] – Hannah Mann

So to demonstrate this point in numbers, there is over 200 million items of content produced every 60 seconds. And we supposedly exposed over ten thousand ads on a daily basis, which is staggering. And it takes only nought point four seconds for the brain to engage with the brand message in the precognitive site, which means we’re constantly trying to digest all that’s going on around us. So it’s no wonder we can’t because we’re experiencing cognitive overload on a daily basis.


[00:03:23.790] – Hannah Mann

All this means is that we’re in this new age of the attention economy and the various books and articles written about this, but in its simplest terms, when something is in short supply, prices for demand, it becomes a scarce commodity and its value goes up. So as a result, to get hold of that scarce commodity not only becomes harder, but it also becomes more costly. The why do we care? Well, I work in health care and being able to get a clear and often really complicated message about, say, a new drug through to a doctor is very, very difficult.


[00:03:56.260] – Hannah Mann

I don’t always care about the brand name or anything of that brand, really. And it’s definitely not the most important thing in my day. So in order to cross through, our clients need to work much harder across three key areas. So what I say, I say something interesting in you to get noticed how they say it doesn’t matter, that they may have five pages of data that they want to share. It is still huge. They still have roughly nought point four seconds to get their attention.


[00:04:22.620] – Hannah Mann

And you need to make the messages easy to digest and remain focussed on only the messages that really, really matter to drown out the noise. Don’t worry, there is a solution that can help with this dilemma in the form of predictive attention. I know now that Mark is going to explain exactly what this is.


[00:04:43.110] – Mark Bainbridge

Thanks, Hannah. So actually, our story in terms of retention prediction started back in 2010 with a piece of academic research from Queen Mary, University of London. They were looking initially to give robotic devices, human visual logic, interpretation abilities that they could contextually navigate more effectively around where they were delivering their robotic solutions. And it started a piece of research which ended up lasting about four years. In the end, only after the university were looking specifically at what we call System one biological human processes.


[00:05:15.240] – Mark Bainbridge

They’re trying to look at how the human brain works as a biological computer effectively. So we make a big separation between System one and System two, which is our emotional response and all the things that make us individual, unique and human. So what they establish is that the biological system, one state, the human brain, ingests content and processes it using five principle neural pathways. And it helps us then to prioritise what we’re looking at. Just give you a quick insight into the neural pathways themselves, our ability to discern between darkness, to discern between different textual surfaces, to look at contrast and edges which help to define shapes in space and time.


[00:05:51.900] – Mark Bainbridge

And ultimately then we can look at colour, shape and perspective. And the final fifth neural pathways are uncanny ability to spot anomalies and patterns and orientation lines. So if we click onto the next one, basically what Dragonfly does is just kind of turn this this mathematical formula that the university published to the academic world back in 2014. And we partner with them effectively to turn that into a software solution that allows us to use and repeat the use of that biological algorithm.


[00:06:24.540] – Mark Bainbridge

And what we’ve done is effectively created a software solution that ingests content, applies the algorithm, and then outputs that in terms of the series of heat maps, metrology layers and regional analysis capabilities all instantly available to us. So we can do this with pretty much any content and it’ll help us to understand what the human brain sees first in any instance. It was also tested by Mittie in the US and they compared us to human baseline sample and we scored just under 10 percent of human interpretation level.


[00:06:57.300] – Mark Bainbridge

So for artificial intelligence, that’s considered to be pretty, pretty spectacularly good. What it does, it indexes content and we allow we’re allowed to apply effectively a metrology layer of the index maps that we see. And I’ll show you some in a moment just to get a sense of what I’m talking about. But before we kind of reveal what it does, a few little kind of insights into how it does and what we should be looking out for. So our index ranges from zero, which is low salience typically shown in the heat map as the blue.


[00:07:27.420] – Mark Bainbridge

So the local temperatures up to 100, which is the highest. And these areas are really interesting to us because these are the areas of high, salient attention. It means we can look at that content and we can instantly see where the attention fixation is. It also helps us to understand how we digest content because we also established in the academic research the human brain can contend with between three and five items of content in that spectacular zero point one four seconds of initial attention.


[00:07:54.120] – Mark Bainbridge

So we can show basically what the human brain sees in that instant high school’s over 80 foot content. We want the consumer to see a really important. It means they’re getting the attention focussed. They need schools below 50 a typically a little bit more concerning if they’re significant because it means we’re not likely to see them. So if we move on to the next one, we can have a look at hopefully some content. So I randomly selected some advertising content here to have a look at it to try to help understand what we can do.


[00:08:23.370] – Mark Bainbridge

So what we’ve got on the left hand side here is a banner. Ads we can see is the grid that’s been placed over the colour mapping. So the colour heat mapping shows instantaneously where the attention focuses. Now, I don’t know what I did in this case is trying to convey or the sequence of message hierarchy. But what this tells us is where the human attention focuses. So we can see in the. The five or six red squares, their schools way up into the 90s, there’s a 100 school there, which is the high salience you can achieve.


[00:08:52.810] – Mark Bainbridge

So that’s where the focal attention point is. It might not be what they want the consumer to see, but it’s telling them in an unbiased, objective way is what’s happening with that piece of created the two numbers in the Green Square. They’re also quite interesting. So they call the digestibility index. They tell us about how comprehensible that piece of creative is. Now, on a scale of zero to one hundred, we’d expect good, clean, digestible content to be tracking somewhere between 25 and 35.


[00:09:18.130] – Mark Bainbridge

So in this instance, they may have a few issues they might want to iron out. You know, they’ve got quite high digestibility, which means the ad may be a bit complex for the human brain to absorb. The other is that we can see where the focal attention points are and three of them seem to be over the one leg of the runner, which may not be where they want the marketing payoff to be the central. And that’s just simply a view where we just use the basic heat mapping.


[00:09:41.170] – Mark Bainbridge

And on the right hand side, we have automated regions which we can apply to content. Well, that helps us to understand is how that content is performing. So we’re looking for here is obviously the descending sequence of numbers, which helps us to understand the way that the brain is effectively digesting what we’re looking at in that particular piece of content. The downside of this, again, for either us is that because all those numbers are scoring well below 50, the chances are we’ll probably disregard most of the content in that.


[00:10:09.440] – Mark Bainbridge

So the next example is just to take you into the ditch while we have a Google Chrome extension, which allows us to look at live digital assets or assets on station servers, in this instance, we can see, mom, I love it or hate it. There it is in all its glory. We can apply something called probability of perception, which is what percentage of viewers are going to see content. And then we have automated regions we can apply over this content.


[00:10:32.300] – Mark Bainbridge

So we’re beginning to understand is what percentage of consumers visiting this page are going to see these these content items. So some things are performing really well. So the product skews are doing pretty well and some of the product detail might be underperforming. So, again, we’ve got an instant view of how digital assets are working for us. And then the last example we wanted to show you is how we can apply Dragonfly to video motion content, which again, might be quite interesting.


[00:10:57.200] – Mark Bainbridge

Topical ads post pandemic. And here we can see how Dragonfly applied to visual content is beginning to help us to understand whether we’re actually delivering with significance the content that’s in view. Now, the downside of analysing a TDC Post publication or broadcast is that, you know, it’s too late to make any changes to it, but it’ll help us to understand in the edit process, you know, how to get the best out of content. So typically, if we’re looking to place products in video sequence, we can show that the products are getting the right level of visibility and attention.


[00:11:31.610] – Mark Bainbridge

So the nice thing about Dragonfly ideally is it’s an iterative to allows you to optimise your content for human attention before you publish it.


[00:11:41.870] – Hannah Mann

So what can this technology really do to help our clients that think about the value it potentially adds? But I think this quote says it all, it’s no, it’s not it’s not really good enough to guess if something is being noticed, a notice for the right reasons, we need a better and more definitive measure with predictive attention at which it can bring to the table. This technology helps us to truly understand what is most likely to get people’s attention, which immediately gives brands an advantage over the competition.


[00:12:15.000] – Hannah Mann

It doesn’t really matter how great your product is if no one takes any notice of what you’re saying about it and really it’s going to get overlooked. But in summary, there are two clear advantages. So this technology allows us to go beyond what people say that day or take notice of and gives us more concrete evidence of what they will actually see and then just in their life. And if we apply predictive attention to promotional materials or store layouts, we can add we can learn quickly how to optimise the layouts to increase the chances of them being noticed.


12:47.850] – Hannah Mann

And finally, we can use a set to make sure that the key messages that we know will have the greatest impact on behaviour are the ones that get noticed first. So in other words, this is a quick and efficient tool to enhance what we know as humans works, but to make our communications even better. Only a few things we’ve not mentioned already, but I think a really, really key yet is tech. It does improve the advice we can offer as researchers.


[00:13:14.180] – Hannah Mann

But on top of that, it’s really, really fast. It’s frictionless and it’s cost effective. And in fact, the tech it’s instant in terms of how it looks and the analysis can be done in just a few hours. So the question really is, why wouldn’t you use it rather than why would you? And we look forward to your questions. Thank you very much.