Unlocking the Currency of Attention - Dentsu's Attention Economy Program Phase 2 Research
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Introduction The fact that the media landscape today is vastly • As an agency, can we bring clients and media more fragmented, complex, and cluttered than in the owners on board to participate, so that we can past, is unquestionable. The reality that the industry influence both the buy and supply side to drive the has not adequately adapted to this, is uncomfortable. industry in this direction? Herbert Simon, the 1970s Nobel Prize winning Three years later, we’ve met these objectives. The economist, coined the term “Attention Economy” cost has been an incredible amount of time and rigor saying: “What information consumes is the attention in conducting and acquiring the highest volume of of its recipients. Hence a wealth of information eye-tracking attention studies to date, working with creates a poverty of attention”. Fast forward for the most cutting edge and forward-thinking partners today, and this statement could not be more relevant in the space. of the media landscape. Yet the fundamentals of how the industry approaches media planning and buying We have just completed Phase 2 of our research has not evolved. Current value systems still consider program in partnership with Lumen Research, TVision, all impressions equally, regardless of how much and Amplified Intelligence. As a result, we now have attention they receive. Furthermore, there’s been a a comprehensive data set which has enabled us to proliferation of device-centric metrics that are easily develop the attention model we set out to create. measurable and proxies for ‘attention’, but they do This model will allow us to put attention into practice not represent a meaningful exposure. and deliver more effective solutions for our clients. It is with this context in mind, that three years ago, we set ourselves the ambitious challenge of transforming the way we plan and buy media. We established the Attention Economy program with Clive Record a goal of researching advertising attention and Head of Global Partnerships developing a value system more indicative of a real opportunity to communicate with a consumer. Most importantly in all of this, we needed to make the research tangible in practice. In many ways, this is where the devil is in the details. The concepts, research, and insights into human behavior is illuminating; finding a way to translate it and adopt it into practice across channels and platforms is an enormous challenge. As such, we had several issues to address: • Can we be confident that attention is a valuable media currency that predicts a real (versus optimistic) opportunity to communicate with a link to brand outcomes? • Can we develop an attention model that can be operationalized in a way that is useful to compare across channels, and captures the value and nuances of attention and human behavior within each? 01
The Speed Read In the latest phase of attention research with Lumen Research, TVision, and Amplified Intelligence, dentsu set up large scale eye-tracking panels to build our proprietary attention model. This model captures the drivers of attention, as well as the effectiveness of attention across platforms, channels, and formats. The output of this model are a massive database of attention metrics that can be used for cross-channel planning. That results shed light into several interesting insights such as: 1. Viewability and other device measures can be quite arbitrary proxies for real attention. In many cases, the % of viewable impressions reported are higher than the % of ads that get attention. In some cases, viewability standards leave value on the table, specifically when it comes to mobile or in-feed formats. Many of those ads can be “noticed” without being considered viewable and can drive some uplift in recall and choice. Therefore it is critical to evolve beyond viewability and measure attention better to get to a more genuine exposure vs. an optimistic one. 2. Creative is the biggest driver of how hard attention works; the difference between strong and poor performing creative drives the largest shift in outcomes compared to other factors. After that, it is followed by volition (whether attention is earned or forced), sound, and duration. 3. Duration is correlated with higher levels of attention, but shorter video ads can pack a lot of punch, they can be more “attentionally efficient” by delivering outcomes in a shorter amount of time. 4. Sound is nuanced; the impact of sound can be dependent on whether it is expected by the consumer to be part of the platform experience. 5. The new metrics for planning drive increased value. For one client, re-optimizing towards attention garnered 3.7 years of extra attention in one month at the same budget vs original plan that was optimized towards Reach and Frequency. 02
Attention is more valuable than our current value system At dentsu, we are committed to delivering our clients Can we rely on our current system’s metrics like with media exposures that lead to genuine outcomes for viewability to equally represent real behaviors their brands. Therefore, to advocate attention as a media that lead to brand outcomes? Not really. We see in currency, we needed to test its link to brand outcomes. our research that there is a significant difference between ads that meet MRC viewability standards Across Phase 1 and Phase 2 research, we conducted and ads that are genuinely viewed. Conversely, many robust multi-brand studies in the US, UK, and ads, especially those that tend to be in-feed and Australia combining eye-tracking technology with mobile can be noticed without being considered choice and recall measures to understand the drivers viewable (see figure 3). of attention and its link to impact. This enormous undertaking has left us in no doubt that the link exists Fig. 3 % Viewable % Viewed and insight into what drives this link. 98% 97% 97% 99% 85% 77% 79% 79% 76% The research shows that higher dwell times are 70% 72% 59% 65% associated with a greater likelihood to choose a 49% 36% 32% 83% 92% 92% 31% 77% 83% 6% 77% 76% 4% 7 brand’s advertising, and we see clear (and statistically 55% 55% 69% 69% 47% 43% 4 56% 7% 56% 74% 7 39% 43% 35% 39% 35% 43% 36% 36% 43% significant) uplifts vs the control group at even low 17% 2 17% 0% 25% 20% 25% levels of attention (see figure 1 and figure 2 below). A A B C E A D C B F B C D E F G H TV Fig. 1 Eyes on dwell time and prompted recall There is a similar pattern between the average 80% viewable time and the average number of seconds 70% someone looks at an ad (dwell time). Ads are 60% typically seen for much shorter durations than they appear on screen (see figure 4). 50% Recall 40% Fig. 4 Viewable time Av dwell time 30% 34.5 23.6 20% 19.4 21.2 21.1 10.0 12.3 11.4 12.2 9.4 10.4 12.9 7.0 6.9 8.7 4.6 8.5 7.4 6.4 6.0 9.3 10% 1.5 1.6 2.1 1.6 1.5 2.4 1.9 3.1 1.9 2.2 2.6 1.8 3.4 A A B C E A D C B F B C D E F G H 0% 05 10 15 20 25 TV Dwell time (s) Fig. 2 Eyes on dwell time and choice uplift What this chart reflects is that device measures like viewability and viewable time are rather arbitrary 18% proxies for real attention. Sticking solely to them is 16% 14% not a guaranteed precursor to improved advertising ROI. In the past, viewability has served a purpose Choice uplift 12% 10% 8% in reducing ad wastage, but today it is simply not 6% enough; attention gets us much closer to an effective 4% brand exposure. 2% 0% 05 10 15 20 25 Dwell time (s) 03
Not all attention is equal We’ve validated how important attention is, we have Fig. 6 Prompted Recall Choice Uplift the data and models to plan and buy it (as we discuss 20% Short (up to 10") further below), and the industry is following. But 80% Short (up to 10") Choice uplift 15% 12% they do not have sight of the data we have linking 60% 51% 10% % recall 5% 5% 42% 4% 5% attention to outcomes, and the knowledge that 40% 25% 29% 0% recognizing and valuing appropriately the nuances 20% 20% Mid (11 - 20) 80% Mid (11 - 20) 15% across seconds of attention is critical to the equation. 65% Choice uplift 15% 11% 60% 50% % recall 10% As with other forms of communication, how 43% 5% 6% 6% 36% 40% 5% 1% 25% 27% different advertising messages work is a function 20% 0% of a combination of circumstances – the message 80% Long (20+) 63% 20% Long (20+) 13% Choice uplift 15% 11% 50% and creative, the context, the device, etc. Focusing % recall 60% 9% 45% 7% 10% 33% 40% 27% 3% 23% 5% solely on how much attention an ad gets is an overly 0% 20% 0% 00 -2 2 - 55 - 101 0 - 15 15+ 00 -2 2 - 55 - 101 0 - 15 15+ simplistic approach that does not consider the Dwell time (s) Dwell time (s) impact of these factors. We also see that sound is an area with important Fig. 5 nuance. Having “sound on” generally lifts choice and Creative recall, but the strength of that impact is dependent Volition Sound Duration Audience /Brand on other factors, such as whether the audience is or 30% is not forced to view the ad, if sound is expected as 25% attention on recall Voluntary part of the advertising experience on the platform, 20% On In-Market Effect of 3.2* Short 17%* 8% 5% 4% or creative factors such as whether sound is primarily 15% 6% Mid Long Not in-Market music or a voiceover. The impact of sound to a “forced 10% Forced Off vs. voluntary” exposure is reflected in the figure 7 below. 5% 0% Fig. 7 Voluntary: Sound off Voluntary: Sound on Voluntary: Sound off Voluntary: Sound on As part of our study, we analyzed the key drivers 60% Forced: Sound off Forced: Sound on 12% Forced: Sound off Forced: Sound on of the value of attention. Figure 5 showcases the 55% 10% impact to recall each one has on a ‘typical’ 3.2 50% Choice uplift (%pts) 8% seconds of attention. Creative is by far the biggest 45% % Recall 40% 6% driver of how hard attention works. The difference 35% 4% between good creative and poor creative can impact 30% 2% 25% recall by 17%. After creative, the other key drivers 20% 02 46 81 0 0% 02 46 81 0 are volition, sound, duration, and audience (note that Dwell time Dwell time these factors interrelate, which is not captured here). Overall, the contribution of all of these variables Other analyses in the study further demonstrate and nuances will be reflected in each platform and how different the value of each attentive second format’s ability to drive recall or choice. As such, it can be. For example, we see that shorter ads can be is important that any value system used in planning more “attentionally efficient” compared to longer for attention considers the ability of each second to ones. This intuitively makes sense, ads that were translate into a brand outcome. designed to work in a short period of time are likely able to land their message quicker and more up front compared to longer ads. 04
Developing an approach for planning So how does all of this come together? The robust The beauty in this approach sits with its flexibility and comprehensive dataset that we have amassed such that we can tailor it to different brands, through our eye-tracking studies enables us to have categories, markets, and business objectives. Through attention norms that can predict attentive seconds our Attention Economy dataset, we have robust and the value of that attention. averages and norms across placements that will already deliver a significant improvement in planning compared to standard impression-based approaches. Combining factors like attention, These norms can be swapped to be category and dwell time, and the outcome model brand specific if our clients choose to conduct their with cost allows us to create a CPM own tailored attention measurement. They can develop their own benchmarks for dwell time and based on an “effective attentive recall (or other KPI) to optimize towards. second” specific to various media All this data will be made readily available across circumstances. our planning tools so that teams are able to easily leverage this for their clients. It is important to note that although we have norms for both recall and choice, we recommend leveraging recall as the benchmark for planning, as it is more robust and has less sensitivity to the specific brand/ creative. We also see that recall has a strong Gained 3.7 years correlation to brand choice. of attention in one month The graph below is an example of how this data can be put into practice, comparing what the format allocation would look like if optimized towards a standard CPM plan vs. a CPM based on an “effective attentive second”. The resulting optimization showcases quite a different mix when accounting for attention. We were able to maintain reach and cost for the plan but were able to gain 3.7 years of attention in one month. Fig. 8 Est CPM CPM EAS 12 8 7.5 5.94 5.55 55 5.12 2.5 2.5 1.55 0.29 Format 1 Format 2 Format 3 Format 4 Format 5 Format 6 05
Activation and measurement Beyond research, data acquisition, and planning, the We are also developing strategic partnerships with next step is to bring this data into life via campaign ad tech companies and scaled platforms on solutions activations. The attention model we’ve built through for activating attention data on their platforms. This our research can then be applied to impression attention model can also be used for measurement level data to predict the amount of attention those so that brands can have a more dynamic indicator of impressions will receive. how their campaigns are performing. We are also pushing the envelope within the TV As such, our “Attention Algorithm” marketplace by testing models with bold clients and can put theory into practice and advocating for fresh methodologies on how to utilize enable optimization towards high TVision’s attention data as a currency. We have been leading the charge on collaborating and rallying attention inventory in display and media owners and industry bodies behind this. video in real time. Delivering meaningful progress in the Attention Economy The Attention Economy program, the largest in scope increased collaboration needed within the industry and scale of its kind, is an important contribution to ecosystem to accomplish any robust change. This improving advertisers’ communication effectiveness. involves lots of research, lots of testing and strategic The research has further validated the value of partnerships. attention as a media metric, and most importantly, how we can apply this in practice. We are working with At dentsu, we are continuing to rally our clients and innovative clients and media partners to ensure we the media owners we work with to continue to move build attention metrics into their media frameworks the needle and to solve for the sizable gap between to drive success. current metrics and real human behavior. However, we know we are at a turning point and are willing to do As with the adoption of any new currency or forward- the collaboration and hard work to help build a fairer, thinking model, there will be some risk-taking and more robust, and effective advertising landscape. 06
About the Attention Economy Program In one of the largest studies of its kind, dentsu’s Attention Economy initiative has significantly invested in the creation and data acquisition across the US, UK, and Australia. We work with the leaders in the space such as Lumen Research, TVision, and Amplified intelligence in robust and independent studies on attention. Since March 2018, we have been measuring attention to advertising through eye-tracking. Additionally, we are collaborating with forward thinking partners such as Facebook, Spotify, Snap, Teads, Yahoo and Broadcasters and Tech platforms. Mobile eye-tracking panels of 3400 respondents were built with Amplified Intelligence, further eye- tracking panels of 6,000 were built with Lumen and large-scale TVision data was purchased based on a panel of 5500 households in the US. Data was collected both “in the wild” audience behavior across multiple platforms and ad formats as well as conducting extensive controlled experiments for brand ads in 20 plus sectors across numerous platforms. Contacts Joanne Leong, VP, Director, Global Partnerships Alaina Pollock, Director, Global Partnerships 07
You can also read