From audience to activation, minimising data leakage

20/12/2017

Big Data delivers huge pools of unstructured information, but agencies and clients need to stitch together different types of data sets to close the gap between planning and activation without losing lots of data along the way. Cate Hardiman and Dan Hagen explain how applying in-depth motivational survey data or structured customer databases can offer a solution and create truly people-led advertising

Kirstie Holsworth Kirstie Holsworth New Business and Marketing Carat UK big data Admap
https://carat-cdn.azureedge.net/media/9272/big-data_square.jpg

Until the very recent past, the route from an audience to activation was simple. We used a range of different data sources to create targeting approaches and profile audiences, delving into their consumer journey, exploring emotional and behaviour triggers, and then campaigns were activated against broad demographic audiences. But the industry has been on a journey from one-size-fits-all to a more bespoke approach.The digital economy has created an explosion of data. More data will have been created in 2017 than in the previous 5,000 years of humanity. And this proliferation of data points contains pathways and signals that can be connected to individuals.

We can link these, with increasing confidence that we are creating a holistic view of a person, exponentially improving our ability to target, create unduplicated reach, and to deliver relevant communications. Moving from reaching unknown to known audiences with greater personalisation increases the probability of being noticed, reduces duplication, and gives brands greater opportunity to create clear meaning in people's lives.

THE OPPORTUNITY
There is huge potential in this. First, to satisfy the needs of brands - by making it possible to understand people more deeply and to connect that understanding to distribution channels better than ever before. And second, to satisfy the needs of consumers: people want more personalisation, in the products they buy, the services they receive, the experiences they seek out - even in the dialogues they have with brands, celebrities and influencers.

One of the myths of Big Data is that it exists as a pool of information that we can dive into and find pearls of insight, just as long as we have the right tools and approaches. However, the reality is messier than that.To create a useful resource, we need to stitch together lots of data from multiple sources, exploring the best route to gather, refresh and link to individuals.

But there is also a challenge in creating meaning from this data.The sandstorm of real-time data generated can often create more noise than understanding. Broad Big Data sets made up of petabytes of data are great resources but they are mostly unstructured and, in many instances, we still need to frame analysis, and direct the outcomes we want. This is where in-depth motivational survey data or structured customer databases can play an important role.

It's why agencies and clients are trying to get the best of both worlds by creating innovative assets by stitching together different types of data sets - audience measurement tools, quantitative surveys, digital media owner platforms, customer databases - to ultimately close the gap between planning and activation, once and for all, without losing lots of data along the way.

THE APPROACH
Over the past three years, we have been working to create a seamless link from insight to activation.We have been integrating Dentsu Aegis' own global quantitative panel, namely the Consumer Connection System (CCS), and data sets including client first-party data, with audience measurement systems and addressable platforms via different data providers.

The challenges are both theoretical and practical: finding the right statistical approach to integrate and map audiences; linking individuals in a consistent, meaningful way; ensuring links to activation. These all required a rigorous process of trial and error. Equally, different markets have different data challenges and finding the best approach to create a people-led data
infrastructure varies.

Through this process, we have identified a range of approaches which allow us to create a link from the creation of audiences through to activation. Our ambition is to do this with the greatest level of data fidelity. We look to use population-size data sets, creating unique IDs where possible, integrating different databases via a one-to-one match and connecting with publishers following a good, better, best hierarchy. The approach deployed is based on a combination of market and publisher requirements:

  • Good: Cookie matching between ID and a DSP. However, the IDs matched are not the same for each DSP as synching is dependent on the whitelist of sites within each DSP and on the audience they can cover
  • Better: One-way match between ID and a publisher. We match IDs with publisher data in a secure safe haven but do not receive campaign metrics reported on an individual ID level
  • Best: A Two-way match between ID and publisher, in a publicly addressable marketplace.We match IDs with publisher data in a secure safe haven and receive campaign metrics reported on an individual ID level.

THE SOLUTIONS

  1. A POPULATION LEVEL DATA SET INTEGRATED WITH MEDIA IN THE US
    In the US, Dentsu Aegis' M1 ™ platform has leveraged the data held on nearly 300 million individuals to create the largest addressable premium publishing marketplace: Publisher Addressable Marketplace (PAM). A one-to-one match with premium sites including AOL, CBS Interactive and Pandora has allowed us to create a truly people-based ID. Agency media teams can plan, buy, and measure media against real people, paying only for impressions served to known individuals.

    Case study: US regional bank 
    A US regional bank wanted to reach individuals located within a certain radius of its branches.The initial goal was to expand direct mail efforts to an online environment and increase results through digital advertising.This had been a challenge in terms of understanding and reaching who the bank's online customers were since they were reaching audiences based on cookies through the DoubleClick Bid Manager open display media exchange.

    Using M1, audiences were created from individual attributes of US consumers based on thousands of attribute data accessible within the M1 platform.These audiences were then matched and targeted at a one-to-one level across Publisher Addressable Marketplaces; addressable audiences, premium publisher inventory, and person ID-based reach only available through the M1 platform. By creating unique people-based audiences, M1 identified new, high-value customers. Each audience was served tailored, consistent creatives to best connect the brand experience and drive action.

    The M1 platform enabled the client to achieve immediate conversions through precise targeting. Campaigns resulted in an 18% higher dick-through rate and a 47% decrease in cost per lead.

  2. PARTNERING WITH ACXIOM TO CREATE CUSTOM AUDIENCE SEGMENTS FOR ACTIVATION IN THE UK
    In 2014, we integrated Dentsu Aegis Network's Consumer Connection System with Acxiom's Info Base, giving us the ability to create bespoke custom audiences for digital activation. Any audience that can be defined using our media, lifestyle and attitudes study run with 17,000 Great Britain consumers, can be extended and activated programmatically and through various digital platforms.

    Case study: Wickes
    Over the past couple of years, Wickes' media strategy has mainly focused on winning the consumer DIYer. In mid-2016, Wickes noticed that this focus was undermining their trade 'heartland' and they had started alienating their core and consistent spenders. They wanted to use communications to address this but had to ensure that our trade and consumer targeting approaches were distinct. 

    We needed to connect data from different sources, leveraging data partnerships and integrations, to create a bespoke audience segment for deployment. Within Facebook and Google, we were able to build audience extensions based on first-party data. But we knew that this audience voraciously consumes news, sports and entertainment content delivered via a range of digital platforms on mobile devices, particularly in break times and on lunch breaks - the perfect opportunity to drive relevance and salience.

    Within the standard third-party data segments, a 'trade' audience covers a large spectrum of occupations - from electricians to hairdressers. Wickes needed to target only traditional tradespeople - for example, electricians, plumbers, builders, carpenters, construction workers, gardeners. We needed to deliver our message programmatically within these environments, so required a smart way to leverage our data sources.

    Using occupation variables, we built a precise Wickes trade audience on CCS, which Acxiom matched to their lnfoBase using their proprietary recognition algorithms. But it was vital that we managed the creation of lookalikes closely, restricting the number of broader consumers who might be allocated to the audience.

    Acxiom used a golden blend of attitudes and other variables, and across the thousands of lifestyle, demographic and behavioural variables they have, the ones that best differentiate the trade audience from the UK average population were chosen and used to build a lookalike predictive model. Acxiom's marketable database of 45 million individuals was scored, creating a ranking where the most likely people to be in this niche Wickes trade audience were chosen. Top lookalike bands were then pushed through LiveRamp to Turn.

    We then created a whitelist of sites, such as Yahoo, Sky, eBay, to ensure relevant context, and optimised the campaign for delivery on mobile at the key times this audience would be looking for content.

    We were able to achieve the lowest cost-per-sign-up we have ever seen in our digital campaigns. This new data targeting, which we applied on Facebook, has performed better than Google AdWords, which tends to be the cheapest performance channel for Wickes. This performance was accompanied by significant improvements in brand equity among the trade market, compared to QI 2016.

  3. CREATION OF A POOL OF KNOWN CONSUMERS IN CHINA VIA TENCENT
    Due to the extremely stringent and rigorously observed restrictions on data in China, third-party data is limited or nonexistent, so we needed a different approach in this market. Data held within walled gardens are protected and maintained by the data owner. Technology platforms, therefore, presented a solution. We have formed a strategic partnership with Tencent, the largest online media company, encompassing WeChat along with other apps, giving us access to their vast data resources and facilitating the integration of our Consumer Connections System with their real-time behavioural data.

    Case study: Large snack company
    A large snack company wanted to reconnect with young Chinese consumers, reflecting their rapidly changing media behaviour, through a campaign to build brand awareness and drive relevance. 

    To gain a deeper understanding, we conducted extensive analysis using our proprietary tools, breaking down our youth audience into key tribes. This generation was born just after China had opened its doors to the outside world and was raised during its massive growth spurt. Culturally, they are distinct from those who have gone before them, being more global in mindset, but equally aware of the responsibility they have in guiding China's future as its population ages. They adopt new technology quickly and are voracious consumers of all media channels.

    The integration between the Dentsu Aegis Network and Tencent databases allowed us to match users and capture online behavioural data for each of these segments. By understanding similarities and differences in media behaviour and attitudes, we created a segmented campaign across paid, owned and earned channels.

    The fused data allowed us to uncover the detail of our audiences' behaviour on all Tencent's websites. This provided us with a clear direction on where to allocate our ads, in terms of both time and position. For example, with pre-rolls, we used the integration to identify the exact time and drama series preferred by our target audience. In addition, we created tailored mobile games for social platforms which were widely spread via key opinion leaders. The campaign saw an unprecedented level of reach, engagement and conversion, and won a number of industry awards.
     
    Of course, this is only the beginning, and as we are able to more directly put targeting strategies into action, the approaches taken will clearly be put under the spotlight. For the first time, we will know which 50% of our advertising investment is wasted. And we will be able to understand clearly how to create that balance between, on the one hand, mass reach that drives efficiency at the expense of effectiveness and, on the other hand, increased targeting that drives effectiveness but often at the expense of efficiency.

    It will also provide us with the opportunity once and for all to change the centre of gravity from which we plan - moving from looking at people through the lens of our brands, instead of looking at our brands through the lens which people view them, thus creating truly people-led advertising solutions.

This article appeared in Admap's December 2017 issue.

Kirstie Holsworth Kirstie Holsworth New Business and Marketing Carat UK big data Admap
^Back to Top