Link for Video

Evaluating the Power of Video Across Contexts Using Neuroscience, Behavioural and Survey Measures


Video continues to be an effective and emotionally engaging way for brands to tell their stories. The growth of the accessibility of portable devices and improving technical capabilities means consumers are able to' watch what I want, when I want' – giving 'video' a broader remit.

  • Programme watching has increasingly expanded beyond linear TV viewing. Catch up/on demand and an increasing number of programmes are recorded for people to watch at a convenient time. While TV continues to generate significant advertising investment, it's clear that TV watching is evolving beyond scheduled viewing and beyond linear TV – particularly in developed markets.
  • Increased touchpoint opportunities and of growth of online investment/accessibility. Multiple screens are being accessed and this is made possible by the portability and popularity of mobile devices. There has been a continued and significant rise in media investment in digital and a drive to better understand how to best measure effectiveness in this complex media environment.

While these evolutions present marketers with the opportunity to connect with consumers through increased 'screen time', they also present challenges. Greater control delivered by catch up and digital viewing means viewers can fast forward through ads, skip content and avoid ads altogether. Furthermore predicting the impact of online video content is not straightforward because of the different ways exposures are paid for and presented within digital.

This paper introduces a new Kantar creative development solution that supports optimisation and measures the power of a video developed for both TV in its evolving form and digital contexts. Link for Video is a direct response to the need for an efficient way of reviewing how well an asset will perform in these different environments.

The objectives of the Link for Video solution:

  • Provide a robust evaluation of the power of the creative and diagnosis to optimise context-specific placement
    • Recognise the way consumers engage with video content within core digital contexts and the evolution of linear TV, and differentiate between what makes impactful copy for each of the contexts and how to measure this
  • Recognise the importance of emotional and intuitive (system 1) response in evaluating the reaction and associations that consumers will take away from the copy.
  • Emphasise the importance of behavioural measures within digital contexts as the only way to capture the natural in-context consumer response which is essential to support optimisation, and evaluating retention and the impact of partial views.

Executive summary

This paper summarises our key learnings from extensive piloting and R&D work focused on what makes effective video and how best to capture and measure these components for a given context. We look at:

  • The universal truths around telling an emotional brand centric story and the importance of System 1 neuroscience approaches in capturing the most honest appraisal of emotional intensity and message take away.
  • The way in which ad presentation, mindset, and context change the creative qualities needed to drive impact and determine what copy development needs to focus on to support optimisation at a context-specific level.
  • The benefits of in-context ad research to provide the most precise account of whether the context will enhance or suppress the video's performance.
  • Interpreting behavioural response in a way that supports optimisation and determines the relative performance of the video in driving viewer retention, interactions, and avoidance in light of ad length bias and key considerations for the YouTube and Facebook contexts.
  • How to diagnose the impact of partial views in the digital contexts by gauging the extent to which take away will be compromised.

Regardless of context, emotional resonance, brand connection and the intuitive response, are consistently important

Creating meaningful brand encounters through story telling

Video provides an ideal opportunity to tell a brand story and what better way to engage with consumers than through a story.

Videos that tell stories are proven to achieve better results

Stories generate more expressions whilst people watch the ads, and lead to greater involvement with the advertising ensuring people engage with the brand rather than tuning out.

The truth is that people don't think long and hard about ads and what they mean. Therefore they will not remember the whole story but take away the bits that were most engaging for them – and most of the time this is driven by scenes in the ad that hold the most emotional content. It is these scenes that will be magnified in their minds and remembered. If the memorable scenes are intrinsically connected to the brand and these scenes convey the intended brand associations, you have a powerful creative vehicle.

Emotional response and brand associations are best measured intuitively

The more emotionally engaging an ad is, the more memorable the ad will be. More traditional survey measures allow us to understand someone's thoughts about an ad following reflection but it is much more revealing to see the intuitive response. Facial coding, our System 1 neuroscience approach, gauges the emotional reaction to a video directly from facial expressions. Facial coding not only measures feelings towards the video as an entire story, it's also a sensitive diagnostic of emotional intensity, scene by scene as the story unfolds. There is a clear relationship between emotional intensity and sales. We have therefore included facial coding as a core part of how we diagnose the viewer's journey through the ad as well as including this in our impact measure for TV. The impact score is our validated predictive measure of in market response. It reflects the extent to which the creative emotionally engages consumers and therefore the potential it has to positively shift brand associations.

Emotional response, measured by facial coding, has been validated to sales

There is an underlying relationship between 'Expressiveness' from Facial Coding and sales uplift

The relationship is yet stronger for medium/big established brands where emotional engagement has an even stronger influence on advertising success

Source: Kantar
Sales shift: 4 weeks post versus 8 weeks pre. Based on 2nd exposure expressiveness means score. Excluding ads with strong Persuasion (>75 Persuasion Percentile). Base : 17 2 cases (Global ) Base filtered on Medium /established brands (130)

Enjoyable advertising that resonates gets the attention of the consumer, which then opens up the opportunity to build associations about the brand. It's very easy to overestimate what a good piece of copy can deliver. Of course, if you prompt someone immediately after seeing an ad with a list of strategic and often complex messages you are likely to get strong recognition of these messages. This doesn't mean that these messages will be recalled when faced with the brand. What we need to understand is what intuitively comes to mind having seen the ad – and it's these associations that the advertising is most likely to support or build in reality.

Our 'Intuitive Associations' technique is a neuroscience system 1 survey approach. We measure the speed at which associations are endorsed. The speed of endorsement relates to how intuitive these messages are for the viewer. In the next example, we see that 'agility', which was one of the key intended messages of the ad, was not intuitive and therefore very unlikely to cut through above other messages which were much more prominent. In contrast it does come through in prompted communication. This shows that people will take this message out if they think harder about what they have seen and heard, but since people don't tend to think hard about ads it is highly unlikely to cut-through on air. Intuitive Associations (IA) is therefore the lead measure in Link for Video to evaluate message take out and we use spontaneous impressions to provide diagnostics by fleshing out how this intuitive response is articulated/interpreted by consumers in their own words, and captures post-rationalized takeaways.

Instinctive communication

Objectives around communications
  • Communicate the 'feel big. Drive small' proposition
  • Portray the 'fun, cool and agile' characteristics of smart car

Base: Total (225), TV/Neutral (125), Pre-roll (100), Social (100)
NB: Illustrative case study as not client commissioned, video objectives created by Kantar.

Emotional intensity alone does not drive brand memorability – the story must be brand centric

How well the associations created by a video will be remembered in relation to the brand comes down to how well the ad conveys an emotionally resonant 'brand' story as opposed to just an emotional story. Ultimately we need to be able to measure the power a given ad has in influencing brand associations by directing the viewer's attention to the brand itself, and leaving them with positive feelings towards it. Therefore, when you include the strength of brand linkage alongside measures of engagement, there is more discrimination in how the advertising works in relation to driving sales.

Source : Kantar Validation
Sales shift : 4 weeks post versus 8 weeks pre. Based on 2nd exposure expressiveness means score. Excluding ads with strong Persuasion (>75 Persuasion Percentile). Base : 17 2 cases (Global ). Low defined as bottom 30% based on impact score , high defined as top 30% based on impact score

Creative qualities that predict in market impact are different by context

Emotional resonance and brand connection are consistently important in determining how successful a video is likely to be. There are however, significant differences that occur in the mind-set of the consumer, based on context and the way the video is presented that mean that impact cannot be measured the same way across contexts.

Measuring how much attention a video will hold is key to predicting impact on TV

For TV, the viewer's state of mind is less focused on the context because consumption of TV context is more passive and the exposure is less direct. Therefore understanding the likely attention that will be paid to the ad while it is playing on TV is important in determining efficacy. Validation of in market impact based on TV advertising, shows that ads need to actively involve viewers in either a positive (interesting, unique, involving) or negative manner (irritating, unpleasant or disturbing). Therefore Link's active involvement survey question is a key indicator of the likely attention you will pay to the ad and directly relates to TV ad awareness.

This evaluation of likely 'attention' paid during screen time is unique to TV and is not needed to determine the impact of digital copy.. In the online environment, active involvement does not relate to how much of a video a consumer chooses to watch nor does it show a relationship to online ad awareness.

The below digital validation work shows the correlation between different impact measures and in-market response (measured by online ad awareness). It is the first research of its kind to focus solely on digital video. The analysis shows how the ads generating high impact relate to higher increased ad awareness in market, where impact is measured in different ways. When active involvement is removed from the impact score, the validation is strengthened and we see greater discrimination between low and medium impact ads.

The validation results showed ads that are in the top 25% of new impact score generating higher % increase in online ad awareness pre to post campaign and greater discrimination between bottom 25% of ads and middle 50%.

Source: Brand Lift Insights (video only) validation based on 66 cases
Markets in validation: UK /US /France/Germany /Spain/Italy/Belgium /Russia/Netherlands/Mexico

Online video is not one and the same – impact needs to be measured differently for pre roll and Facebook video

We see an important distinction in how TV copy and digital copy generate impact but there are important differences in the way digital ads are presented and perceived by the viewer, which mean that Facebook and pre roll should be treated differently when determining what drives copy impact.

Pre-roll precedes the video that viewers have set out to watch. In this situation, the consumer's mindset is mission based. Viewers are focused on the screen as they wait for their video to play.

Viewers in Facebook are most likely responding to information, being prompted to visit by updates from friends or browsing within a personal and social environment. There is a great deal of personal content to draw attention so there is a lot of distraction to overcome if a video is to standout. This becomes increasingly difficult for the auto play ads which will play in mute. In this context, therefore, it is important to include a measure of whether people will stop and look at the video, as a key indicator of attention the video is likely to receive while it is playing.

In summary: creative qualities, which drive a memorable exposure are different for each of our three video contexts. The level of distraction and 'direct' attention to screen means different attributes are needed to predict the power of the creative. For this reason we have different impact measures for all three video contexts in Link for Video.

Context can enhance or suppress ad performance based on perceived suitability of placement

We know that the creative qualities that drive in-market cut-through will differ by context and therefore the combination of survey measures required to predict impact are different for each context. What we wanted to understand beyond this, is whether response to a given survey measure (e.g. enjoyment) would differ by context or if you would get the same response to the same question regardless of context. Kantar conducted research on 13 ads, where the same video ad appeared in five contexts: Context free, TV, Click to Play, Auto Play and Mobile. Each context was researched as a monadic cell and the digital contexts were presented in a live environment, with ads embedded in YouTube and personal Facebook pages. Results were compared across ads to determine if survey response to the video (e.g. enjoyment, branding, and active involvement) changes as a result of contextualising the video in a specific way.

Ads often perform similarly in different contexts but this is not always true

Based on a full exposure to the video ad, R&D showed that survey response to the video was consistent across contexts in just over half of our cases. Most notably a 'good ad is a good ad' no matter the context. Purina's "Beggin" ad is an example that consistently sits in the top 10 % of the Kantar ad database for cut through.

However survey response did not align in all cases. Out of 13 ads tested, six saw notable differences across contexts. These differences were categorized as follows:

1. Enhanced performance when the video is clearly meant for specific context.

In this instance the ads performed better in the digital context as it was obvious from references to 'skipping' that this video was created for online and enhanced by the context.

2. Rejected in some contexts if the video is of a more sensitive nature

In this example the video was perceived to be a little risqué. Mobile and Click to play contexts therefore saw a better response to the video, where the context is more personal and where the exposure is controlled by the viewer.

3. Suppressed performance when the video is dependent on audio

Facebook is likely to see a different response to content when audio plays a vital role in delivering the story, mood or message because exposure is often in mute.

We can see that in just under 50% of cases we do see differences which shows that in certain situations the context can enhance or suppress the way someone feels towards a video.

The implications of the above learnings:

  1. There are instances where scores will vary by context and it is important to conduct research in context to uncover these instances. The core contexts should align with media investment and therefore TV and Pre Roll are core in Link for Video, with the option to add Face Book context.
  2. Creative quality is the dominant driver of variation in scores on the survey measures. Meaning it is reasonable to combine the data and review sub groups of interest without needing to commission extra boosts

Source : IAB

Survey of client and agency intentions, (not actual spend)
  • YouTube is representative of other key channels (video sites, original video sites, TV shows online)
  • social is smaller

Behavioural data is essential in contextualising response to digital ads

Impact is only part of the story, impact measures the potential the video has based on a full exposure but not everyone will get a full exposure.

The value in the way we conduct our digital ad development research is that exposure is done in a live environment and this 'natural' exposure means we can record behavioural response.

Kantar's Link behavioural measures have been validated against in market behavioural measures collected by YouTube as shown below:

Correlation between Link measures & YouTube behavioral measure

R: 57

Learnings on behaviour within a digital environment and how these relate to copy development

1. Ad length has an impact on retention

Longer ads have greater storytelling potential and digital gives far greater licence to run longer length film. Although the cost/efficiency tradeoff seen for longer TV ads does not apply to digital, there is a clear tradeoff in viewer retention. It becomes increasingly difficult to retain your audience beyond 90 seconds, and there is a strong retention bias towards shorter ads. Ad length is therefore an important consideration when contextualising viewer retention.

Source: Kantar (Digital Link)

2. Viewer retention is lower for Facebook compared to YouTube

We would expect to see a different response in Facebook compared with YouTube where exposure is forced for the first five seconds and therefore, by nature, will be higher. In contrast, people need to notice the ad in Facebook before they can watch it. Therefore contextualising response in the relevant context gives you a better understanding of the video's performance.

3. Interpretation of behavioural data is dependent on how exposure is being paid for

The complexity of the digital landscape and differences in the way views are paid for means there is no 'one' behavioural measure that can be universally applied to understand the efficacy of digital advertising. It is therefore vital to consider at what point the exposure is charged for to know which behavioural measure is most valuable. If you are only ever paying for a full view for example, the measure'% watched whole ad' will be more of a measure of efficiency of build in reach as opposed to efficiency of spend. Focus and interpretation will vary depending on how advertisers are paying for the exposures.

4. Impact & message take away do not have a linear relationship with % ad viewed

Behavioural data is used to diagnose whether take away from the ad is likely to have been compromised or not in the case of partial exposure. Impact does not have a linear relationship with % of the ad watched. The only way to diagnose take away is through scene-by-scene analysis, overlaid with emotional response to the ad up to the point where most viewers skip or stop watching. This allows advertisers to review if there is potential to optimise the narrative to improve retention and/or viewer takeaway. Taking the example below, this means that, if the emotional response peaks during the scenes that cue the desired brand take away, and this happens in the first 30 seconds of the ad, it is possible that the ads are equally impactful.

Source: Kantar Digital Link

5. Even the strongest ads could be compromised in the Facebook context if audio is a key element.

In Facebook, ads are presented in autoplay and will run in mute unless viewers actively unmute them. Our database shows that on average, fewer than 20% of people choose to watch an ad with audio enabled in the Facebook context in a natural exposure. This is important to consider if an ad is highly dependent on audio. As an example, Smartcar's 'Reverse Parking' launch ad demonstrates how an exceptional ad that performed in the top 25% of UK TV ads in terms of impact, and generated twice the expected levels of retention in a pre roll context, struggled in Facebook due to its dependence on audio.

Our case study showed that many people watched the ad in mute on Facebook. Lack of audio significantly compromises response to the ad, affecting both overall enjoyment and message take away. The 'news' is lost resulting in much lower relevance – suppressing the ads contribution to longer term equity.

6. Early brand reveal can support retention in Facebook and create a one-dimensional 'brand' encounter for those that skip in You Tube

There is an interesting difference in response within the Facebook and YouTube contexts. It is the 'brand' itself that has the strongest relationship with playback measures within the Facebook context.

Viewers in Facebook, which is a more personal context, are driven by a pre-existing relationship or interest in the brand itself that help attract & maintain attention. Therefore one of the reasons an ad is watched in Facebook can be how linked the brand is with the ad. The correlation between branding and playback measures shows how important brand posts can be in reaching consumers, and the importance for the brand to be easily recognised, in Facebook, while people are browsing their newsfeeds.

And in YouTube, for those that are not going to watch the whole ad and may only be exposed to five seconds, a brand presence at least gives incomplete exposures the opportunity to create a subconscious encounter – albeit superficial.


Link for Video is a direct response to the challenges marketers face when creating video today. It has been designed to evaluate the potential of a video across TV and digital platforms, and to help optimise content for each context.

Video continues to be an engaging and effective way for brands to tell their stories. We live in a world where viewers will no longer distinguish between different screens but will simply expect content to be readily available on whichever screen they want to use. Developing content that is fluid and effective across screens demands copy development tools that meet this need.

Different contexts can alter response to a video by enhancing or suppressing viewer experience and pose different levels of interaction, avoidance and distraction. Therefore Link for Video, designed as an in-context solution, provides feedback at a context level. It also has the ability to aggregate data where response is more aligned for deep dives on sub groups of interest.

As an in-context solution we collect and report behavioural response. Link behavioural measures are central in evaluating the efficiency of the copy in creating full views/longer views in relation to other ads of the same length, and identifying where people are most likely to skip. By understanding where consumers skip in an ad, and the proportion of people that skip, marketers can determine whether the paid for impressions support meeting objectives set for the ad.

We have put emphasis on neuroscience approaches as the lead measures of emotional resonance and message take away. We know consumers don't work hard at decoding advertising. Therefore the intuitive response is the most important measure to determine which scenes and messages are most prominent in a video, and most likely to be remembered.

Contact Us

Kantar ©2020