Can you test how an ad will perform?

How advertisers access the subconscious part of our brains is largely to do with making us respond emotionally. There are businesses out there that can put metrics to these mannerisms. One of these ad testing agencies is UK based Kantar. Their clients come to them, looking for help at all stages of the development of their marketing campaigns and they're trying to use methods based on science to assess how an ad may perform. Vera Sidlova, Kantar creative director explained a few of the tools in their arsenal to James Tytko before testing it on Julia Ravey...

Vera - We use a tool called 'link AI', which is a survey based tool. We have a set of questions that measure the strength of association between the ad and the brands. We measure how much you enjoy watching the ad, whether it stood out of the crowd, whether it would really capture attention. We also capture how it changes your opinion and predisposition towards the brand. But in addition to asking survey questions, we also use a neuroscientific method called facial coding, which is a software that we use using your webcams. We ask you if you'd agree to be facial coded and if you say yes, then using your webcam we monitor your face while you're watching the ad. We're not actually watching your video physically, the machine just records the movement of the smallest muscles on your face to decode whether you're frowning, or smiling, or maybe experiencing a little bit of disgust, or confusion, or maybe a bit of a smirk. It does that analysis second by second, as you watch the ad. It is a combination of that neuroscientific method and your answers to survey questions that we use to determine quantitatively how an ad is doing.

James - How do your clients usually act on your analysis? I presume they come to you looking for data driven reassurance; that their campaigns are going to be successful. How do you convince them or reassure them that what they're going to put out there will be received well?

Vera - We have metrics that are validated to sales. Our clients know that if an ad scores high in link, it is going to deliver on ROI.

James - How do you convince them of that?

Vera - We've done studies to show that ads that have scored high on link then had higher responses in sales. We've got some big metadata to show and to prove that. But in terms of a single ad we'd be looking at, we have the strengths of the database, we've got the validation work, but then usually what we do is we look at the data together and we're trying to figure out the why behind that data. For example, we could have an ad that we see isn't resonating with people as much as the client had hoped. It's not doing terribly, but it's not doing great either and so the conversation would be, "what can we do between now and airing to make the ad more enjoyable?" If the ad was, for example, supposed to be enjoyable because of a funny joke, we would study using facial coding, whether the joke was actually being received, whether all groups found it funny, whether there's a comprehension problem.

James - Does it tell you things different from what you can get from the survey? Is it offering alternative data?

Vera - Yes. Our faces often show things that we ourselves are unconscious of. What the survey gives you is what the respondent themselves remembers and is also willing to disclose in a survey. But what the face tells you is a very second by second view of how they felt at a particular second of watching the ad and those might be things they aren't even conscious of themselves.

James - Vera sent me a demo of the software they use at Kantar to record the micro-expressions which help them analyse ads. I wanted someone who knew next to nothing about what the software was looking for to run the test and I also knew I wanted someone who I thought might produce some fairly emotive responses to get some detailed results. This left just one candidate.

Julia - James has sent me a link to an ad testing service I've been presented with the 'Coca-Cola' Christmas advert, an advert for 'Plenty', which is kitchen roll. My camera's just come on - loading the emotion detector. Welcome. We are about to play a video for you. The graph below will plot your engagement with the video over time using various metrics. Coca-Cola Christmas.

[Coca-Cola Christmas advert plays]

Julia - So cute. Oh, I'm crying. Oh my God. I'm crying. I hadn't seen that before. Oh no. It was a dad who was trekking across to give Santa a note from his daughter. The letter just says, Santa, I want my daddy over Christmas. Well that emotion was off the charts. My valence, which I'm guessing is my engagement, I was really engaged with that other, so that to me was a very emotional one. Okay. Next video. I can't imagine I'm gonna be emotionally engaged with kitchen rolls, but you never know.

[Plenty kitchen towel advert plays]

Julia - Ooh. Are you joking me? That one was really funny. It was essentially a Christmas day and everything was going wrong. There was a cat that pooped on the floor. There was a baby who was sick all over someone. In every instance where there was a mess, the kitchen roll was there to clean it up. Yeah. My disgust was pretty high, but I was extremely expressive. My expressiveness was on the ceiling throughout all of it because I think I was laughing. Right. Well, I'll pass back over to James now to learn a bit more.

James - Wow. If I'm honest, that went pretty much as expected. Julia's emotional engagement broke the scale with those two ads, but it's not just obvious expressions, which the software can detect.

Vera - They could have frowned a little bit when a certain element appeared or that could have showed disgust maybe when there was a food shot. There are a lot of things that happened in the course of even a 30 second ad that your brain would never be able to recall because you wouldn't be conscious of even making that small muscle movement on your face.

James - I want to pick your brain a bit and ask your opinion on what you see as the role of science in this industry in what you do?

Vera - I personally see the role of science as helping us understand consumers and the people who are consuming our advertising. I think advertising is part art and part science. We've got the great creativity that agencies bring, but how it's going to land and what effect it's going to have is something we may not know initially. That is why research is important, both at the individual advertising piece level, but also broader. How are people perceiving advertising? How does their brain work when they watch content in general? What makes them smile? What makes them laugh? What are they sensitive to? I really think there is a big role for science to play in this and to really help all of us understand just what makes us tick, because I think it's no secret that none of us watch TV to watch advertising. We don't go on our phones to watch a digital banner; it has to do its work. It has to capture our attention. It has to tell us something that's meaningful to us and the 'how' of that and 'how we do that' is quite a mystery and research can help unveil that.