Recycling An Audience Is Bad For The Environment

DSP

For many marketers, digital media has become the perfect platform for a game of Buzzword Bingo. If you’ve got ‘programmatic’, ‘re-targeting’ and ‘Demand-side Platform’ (DSP) on your card, there’s a fair chance you may have a winning line. But beyond the geek-speak, how can these terms actually help marketers overcome their most common challenges – optimising their marketing budget and driving customer conversion? In the current climate for digital marketing, these are the key environmental issues.

But, as Al Gore demonstrated, when it comes to the environment, there’s always room for an Inconvenient Truth. So here it is: some of the solutions currently employed by marketers may be very effective but they aren’t particularly scaleable. Knowing the ROI from your RTB may help you deliver some marketing gains – but it won’t be long before the metaphorical ice caps begin to melt. To progress, digital marketers must face up to the ultimate dichotomy: they must push for climate change, and move away from advertising models that place a disproportionate weight on the principles of recycling. It’s bad for the environment.

Examination of the present climate reveals that digital display advertising continues to grow. IAB/PwC data on digital ad spend shows that in the first half of 2014 UK advertisers spent more than £1 billion on digital display alone, with the sector growing by 30% year-on-year. In an attempt to bring measurability to their media spend, brands are increasingly investing in ‘programmatic’ technologies that enable them to buy and publish digital display using automated systems – DSPs – and real-time bidding across multiple ad exchanges and inventory sources. In 2013, 28% of digital display advertising was traded programmatically. By 2017, it’s estimated that up to three quarters of all digital display adverts will be purchased and placed via programmatic technologies.

Recycling Trends

A major advantage of the traditional DSP approach is the ability to target recognisable prospects – and nurture them through the sales funnel to increase conversion. The automation enabled by DSPs has also brought increased efficiency to digital media buying processes that had historically involved a high number of manual steps and suffered inherent shortcomings in terms of reporting, metrics and administration.

Alongside this, the market shift from contextual selling to more audience-based targeting has proved a key driver of the programmatic approach. In an effort to identify John Wannamaker’s elusive 50% of audience – and reduce the amount of wasted media spend – marketers are using DSPs to help them focus on users based on patterns of behaviour and other variables. In the process, the targeting of ‘in market’ prospects has become more accurate – and customer conversion has invariably increased.

So what’s not to like? ‘Traditional’ DSPs , most of which usually offer a self-service, hands-on, approach to their licensing model, bring speed and efficiency to digital media buying and help drive conversion by improving targeting further down the funnel. Targeting users that are already demonstrably ‘in-market’ is a powerful approach – but I would argue that elements of it are less than optimal. The majority of DSPs are designed to facilitate two related methodologies; retargeting and predictive analytics. Both can deliver healthy uplifts in measurability and customer conversion – but used on their own they can present immediate challenges with scaling the audience you want to target. You guessed it, it’s another inconvenient truth.

Retargeting works by embedding tags and pixels on key pages of an advertiser’s website that, having placed a cookie on users as they visit the site, helps brands track user journeys through the funnel. If they drop-out or don’t convert, brands can ultimately use that cookie to find that user across media owners and through ad exchanges and retarget them with appropriate messages to persuade them to return and convert. Similarly, the use of ‘predictive analytics’ is currently de rigueur in digital marketing. This approach relies on using algorithms that enable DSPs to identify prospects that display the same behaviours and characteristics as known customers – and target them through the funnel. This is effectively ‘lookalike’ targeting – following a facsimile of the parent segment.

For in-market customers that have visited a brand’s website – which is an essential component – retargeting and predictive analytics are effective approaches that can have a strong influence on conversion. But in both cases, brands are not reaching a new audience – they are preaching to the choir. It’s a relatively straightforward form of digital recycling that may help convert recognisable prospects, but it doesn’t help marketers identify net new audiences. Brands should also be seeking ways to keep filling up the hopper with new customers – but without the cookies of in-market visitors, how can you identify potential new prospects? Ultimately, the sole reliance on recycling is unsustainable.

Moreover, the majority of ad tech solutions used to deliver these programmatic approaches are human-assisted – they require technically adept teams to identify the target audience based on market understanding and data-driven insights. And therein lies another problem; human-assisted optimisation isn’t truly scalable. The proliferation of digital media has led to a burgeoning volume of data points and an overwhelming level of digital inventory.

Defining a target audience or segment requires the analysis of an infinite number of potential variables. Using a human-assisted approach to constantly appraise performance, understand data trends and develop the best campaign response isn’t optimal. We refer to this challenge as ‘the curse of dimensionality’ (aka, there are too many options!). Programmatic, for many traditional DSPs, means automatic – but it doesn’t necessarily mean intelligent.

So faced with such a mass of digital noise and pollution, how can marketers ensure their ad spend is intelligently optimised? The answer is to turn to the next generation of DSP – which leverages machine learning capability. Programmatic machine learning enables brands to focus on new customer acquisition – true prospecting – by targeting users who are neither ‘in market’ nor ‘in funnel’. Machine learning is the automation of audience identification and campaign optimisation; it uses a velocity and volume of data mining to enable marketers to identify new audiences that standard retargeting and lookalike techniques can never achieve.

By using a machine rather than a human brain, marketers can optimise vast and multiple combinations of data to identify trends, define demographics and inform campaigns – all focusing on new customer acquisition, rather than recycling messages to the same known prospects. And the final icing on the polar ice cap is that machine learning solutions are not only sustainable, they’re always on – perpetually canvassing across the web to find, and target, new audiences that human-assisted technologies cannot achieve. Programmatic machine learning never stops – it’s always learning, adapting and evolving as data builds and user behaviours change.

Global Warning

The emergence of machine learning in the DSP market may seem like another contender for Buzzword Bingo, but in the plainest of English it promises to transform how marketers meet one of their biggest challenges: optimising marketing budget. Its capabilities reinforce an inconvenient truth that climate change must happen. It’s a global warning: marketers’ reliance on recycling known audiences is neither optimal nor scaleable. It’s bad for the marketer’s environment. Future generations of true prospecting will rely on machine learning.

Richard_Foster

Richard Foster holds an in-depth knowledge of the European digital advertising industry with over 15 years’ experience and a track record of success in building, scaling and managing businesses within these sectors. Prior to taking up his role as Chief Commercial Officer at Intelligent Optimisations, Richard worked in a number of managerial and commercial roles for organisations including Krux Digital, TRUSTe, Future and Audience Science. With expertise in ad exchanges, data trading and data management platforms, Richard’s proven success in defining and implementing digital strategies is drawn from market experience. It is this experience that he brings to bear in his role at IO; helping to deliver exceptional ROI for online advertisers through machine learning, discovering new audiences and providing business insights.