Why Christmas Adverts Work – A Data-Driven Approach
The Christmas season is upon us once more, and with that comes a desire to spend more time with what truly matters to is, to find meaning, peace, and solace. For many this will mean joining family, friends, and loved ones – caring for others, and feeling cared for. But to bring joy to iMotions, we need only one thing – human behavior research.
As Christmas comes around, there is at least one thing that is as reliable as the over-consumption of eggnog: Christmas adverts. By now you’ve likely been inundated with broadcasts of commercial cheer, each clamoring to connect you to the spirit of the holiday season, while also connecting you to the brand.
The premium Christmas advert has now become a staple feature of the season, giving companies the chance to capitalize on the festive cheer, and be discussed about at office (and home!) party conversations all through the month of December.
Clearly, there is great incentive to form a lasting and positive impression around this time, and the high cost of making such an advertisement reflects the competition to get there (some of which are rumored to have budgets of up to $12 million). But what makes them work?
Do They Know It’s Christmas?
It’s clear that each video clip makes the most of a story, and makes the most of emotional moments – everyone likes a feelgood story around Christmas time. An emotional story arc is a powerful psychological tool to make a brand memorable and salient, and the advertisements really make the most of that.
There are even more subtle parts of the films though, that also change how we feel. One of the ways in which emotions are shaped is through the displayed facial expressions – we respond strongly to others, and sometimes we really feel what they feel (or at least, what they appear to be feeling).
Clearly in tune with how Buster is feeling – that’ll be the mirror neurons.
This can result from the firing of mirror neurons that can respond when watching someone (or something) carry out an action, and when performing the action ourselves. In this way, we really are in tune with the way in which a person behaves.
Through an emotional plotline – from one feeling to another – the story can be guided by both content and facial expressions, as the audience reacts to both. From this, the trajectory of the advert can be tightly controlled.
That might be part of what makes them work, but we ultimately want to know – which one works the best – and why?
Three Kings / Three Adverts
My first step in finding this out was to analyze the facial expressions shown in each advert, and to watch how each story develops. I wanted to see how the Christmas adverts used facial expressions, and if they could be tied to the emotions of the participants. Ultimately though, I wanted to know if we could predict the level of enjoyment and engagement of the Christmas adverts from bionsensor data.
I chose three adverts from this season, although of course there are – roughly – another 25 octodecillion available (this year), so this might not be the perfect list to you. But these represent some huge brands that clearly spent startling amounts of money for a 2-4 minute advert.
Advert 1. First up, Wes Anderson’s trope-filled and train-based H&M advert is a slightly gloomy, and rather snowy reminder to “Come Together” at Christmas time.
Advert 1 – H&M
To begin with analyzing facial expressions, I carried out post-processing in iMotions, which gives an automatic analysis of the facial expressions within the video.
The Affectiva AFFDEX analysis reveals how Adrien Brody (who stars as the train conductor) and others, show mainly contempt with bits of surprise and sadness (the disappointment of the delayed train), which finally gives way to joy, as Christmas on a train turns out to be pretty fun (with a “chocolate flavored hot beverage with whipped topping”, how could anyone complain?).
Adrien Brody defying train safety rules in H&M’s advert.
Advert 2. Next on the list is the advert from one of the biggest companies in the World – Apple – who use a Halloweenish approach to the Yuletide season, starring a saddened and singing Frankenstein’s monster as the star. The advert ends (sorry to ruin the shock ending) with everyone merrily singing, and we are reminded to “open your heart to everyone”.
The narrative arc is similar to H&M’s in some ways, and the facial expression analysis reveals this. First, the main displayed emotions are disgust, sadness, and contempt (as Frankie is judged, and sad to be judged), before giving way to joy and surprise, as the audience quickly gets over their fear of the undead, and has a grand singalong.
Advert 2 – Apple
Advert 3. Third up is an advert from John Lewis and is, at the time of writing, the most watched Christmas advert on YouTube of 2016. It features Buster the Boxer, a dog with a curious jealousy for a medley of creatures that take over his family’s trampoline. His envious rage is only shadowed by his cuteness. He waits and waits for Christmas day, surprising and shocking his owners by showing off his intrepid bouncing skills, and we are given the phrase “gifts that everyone will love” (but especially Buster).
The story is characterized by the facial expressions, showing contempt and disgust (getting the trampoline up is hard work), giving way to joy and surprise, as Buster is happy to start his career as an Olympic trampolinist, and his family is surprised that he harbored such ambitions.
Advert 3 – John Lewis
The automatic facial expression analysis adds weight to what we can see from the story – the range of emotions that are expressed as the narrative develops. It also gives us some idea about how it will make people feel – as we react to the displayed emotions of others, we would expect them to feel the same as the onscreen characters. But which mixture works the best?
The three adverts competing in the most important competition of the year.
Biosensors and Mirror Neurons
Next up came the humans in the form of iMotions employees. Eye tracking, galvanic skin response (GSR) measurements, facial expression analysis, and a quick survey would be used to gauge their reactions to each film clip.
Thankfully, no-one chose option 1.
These combined methods can let us know where a person is looking, their level of physiological arousal, their emotional valence, and how much they consciously liked each advert. Tying this together allows us to appraise the enjoyability of each advert, and which moments are critical to its success.
Yes, that’s a Christmas tree in the background.
We can also see how well the facial expressions sync between the adverts and participant. An advert that elicits the same expressions that are shown on screen would likely be engaging material, as it triggers matching emotions (and potentially mirror neurons too).
What the Data Shows
Looking at the data, we can see that H&M’s advert produced the highest number of GSR peaks on average. This suggests that despite all the vigor and zeal from Buster, that delayed trains are really the thing to get our pulses racing. The higher physiological arousal suggests that the emotions that were felt throughout the advert were at least felt strongly.
A portion of the GSR peaks that were elicited by H&M’s Christmas advert.
As for those emotions – facial expression analysis revealed that the three adverts elicited similar, but ultimately distinct patterns of sensations. The John Lewis (Buster the Boxer) advert is matched in a simpler trajectory to the on screen characters – while the advert itself shows a range of negative emotions that become positive (with surprise), our off screen participants displayed contempt, mixed with joy.
It’s a pretty good match, but not exact, as the negative emotions appear intertwined with the positivity. Buster did well, but to really impress our participants, he might need to learn some new tricks for next year.
Apple on the other hand, were less successful – poor Frankie. The advert itself shows almost all negative facial expressions (fear, disgust, sadness, contempt), followed by joy and surprise, while the participants firstly displayed joy, a dashing of contempt, and a final hurrah for joy.
A brooding “Frankie”.
This might seem like a counter-intuitive result – why would the participants first show joy, when the on screen characters appear so glum? Now, if we were limited to conventional methods (just surveys or focus groups), we wouldn’t even be able to pose this question. Fortunately, with everything at hand, we can not only ask this, but also answer it.
The eye tracking shows us the way, wherein the main dissonance is revealed – along with Frankie’s face. With all eyes on his closeup, we can see a part of the early negative on screen emotions (his resting face isn’t exactly one of jubilant joy). It appears that his abnormal appearance is a crowd-pleaser though, which is certainly something to consider – negativity can be beneficial to show in some cases.
Tracking “Frankie’s” facial expressions – it even works on the undead.
Putting it Together
As for H&M’s train-tale, the comparisons are closer. Our digital compatriots show contempt, surprise, and sadness which arrives (full steam ahead) at joy – while the real-world participants show contempt, followed by joy and surprise. This is a slightly closer fit than the two other previous clips, suggesting a greater (although not perfect) emotional matchup.
It also appears that this emotional alignment may be more closely related to the ranking than the story per se: the Buster epic also ended with a surprising and joyful ending, yet wasn’t ranked as highly, and all adverts began with negativity that becomes positivity.
So what does this mean? With the GSR data, and the corresponding facial expression data, it appears that (drum roll, please)… H&M’s Christmas advert was the most engaging – pulling in the participants and making them feel strongly about the characters.
All eyes on the eyes.
And the Survey Says…
The adverts were watched in a randomized order, with the survey directly after each, to avoid any confounding biases from the order of presentation. The adverts in general were well liked, which is certainly a thumbs up for the Christmas advert industry (well done guys).
Overall though, the winner from the survey was “Come Together” by H&M (and Wes Anderson), with an average score of 4/5. While Buster the Boxer’s efforts were well acknowledged, they ultimately fell short, with an average of 3.75/5. Apple’s “Frankie’s Holiday” was awarded a decent, but middling 3.5/5, on average.
This corroborates the findings from the biosensor-based approach, wherein the most engaging advert also appears to be the most liked.
The survey results.
So what can we take from this? It appears that strong emotional mirroring can be important for an advert’s perceived enjoyability, as measured by GSR and facial expression analysis. The surveys allow the participants to speak their mind to an extent, but the biosensor measurements allow us to reach a little bit further into what they’re thinking. Combining these methods can give us a clearer picture of how an advert is experienced.
Of course, this isn’t the complete picture, and this isn’t the perfect study – the Buster the Boxer advert has been an outright success, with a huge number of YouTube views – this may have to do with marketing of the advertisement itself, rather than the content, or it could be due to the sample bias (and sample number) that will unavoidably occur when recruiting participants in-house.
The methods within though do give us a clearer picture of the dynamics that play out on our screens, and from our bodies. Using these together can provide insights that have previously been unreachable.
The winners celebrating.
Let It Snow
This has been our festive investigation, and I hope you’ve enjoyed reading about the interplay of emotions that we can observe and measure, for Christmas adverts (and beyond). If you’d like to learn more, or at least want some reading material for the holiday season, then download our guide for experimental design – it’s a Christmas gift from us to you. And from all of us here at iMotions – Happy Christmas.