27 TECH FEED Audience Data
content and infrastructure online. Sky’s decision to migrate its TV services to IP, and ultimately remove the need for a satellite dish, is an example. Among other things this will help Sky to deeper user personalisation. Sky’s addressable TV service, Sky AdSmart is already enabling advertisers to target households based on factors such as age, location and life stage from a combination of Sky’s own customer data as well as info from consumer profile experts like Experian. FUELLING PERSONALISATION And, according to Sky, it works. Switching channels during a targeted advert is reduced by 48%. Virgin Media has integrated AdSmart into its set-top boxes too, giving both companies combined access to 30 million viewers across the UK and Ireland, and crucially more scale to compete with their true rivals, the social media networks. Fuelling personalisation like this requires massive amounts of user data. Digital-first companies such as Netflix are experts in leveraging every swipe, point and click from their viewers to deliver sticky consumer experiences. However, traditional broadcasters haven’t been in a position to capitalise on the information at their fingertips and have lost valuable market share as a result. “This is beginning to change,” says Downey. “Video service providers, who have either made or are in the process of making the shift to OTT, have implemented user-experience platforms that allow them to harness the power of data to deliver truly personalised experiences. Whether this is the ability to identify a segment of users likely to churn and tailor promotions to prevent this, or increase customer loyalty through the delivery of highly targeted content, success will come to those who put the experience first.” Virgin Media owner, Liberty Global, is taking its data mining to another level. It is harvesting consumer information aggregated from its 24 million customers, accessed over 14 million devices, and uniting it with third-party data to offer insights on advertising and programming to granular targeting goes hand in hand with technologies to automate the process of insertion in individual streams. According to IAB Europe/IHS Markit, more than half of European display ad revenue is now traded programmatically – that is, automatically. The industry makes use of an intricate web of ad agencies, exchanges, networks, broadcasters with which it partners. The effectiveness of increasingly
EVEN A 0.1% ACCURACY IMPROVEMENT IN OUR PRODUCTION WOULD YIELD HUNDREDS OF MILLIONS OF DOLLARS IN ADDITIONAL EARNINGS
demand-side platforms and supply-side platforms to manage the delivery of those ads around the clock. Data is a hot topic since the Cambridge Analytica scandal smashed the illusion of Facebook as a neutral, free service. Google, Amazon, Twitter et al have not been subject to the same data regulation as broadcasters, but in the EU that changes fromMay (see boxout). EXTRACTING YOUR INTERESTS But digital first companies, as usual, are one step ahead. Google chief Sundar Pichai has already alerted us to the company’s shift to an “AI first” world. Applying an AI is the difference between making multi-millions in ad dollars and multi-billions. In a Google white paper (https://arxiv. org/pdf/1708.05123.pdf), the company explains that an online publisher’s revenue relies heavily on the ability to predict click-through rate (CTR) and calls this “a large-scale problem that is essential to the multi-billion dollar online advertising industry”. Identifying frequently predictive features, and at the same time exploring unseen or rare cross features, is the key to making good predictions, the paper states. Since current systems are inadequate, Google’s team posits the use of a “novel neural network that explicitly applies feature crossing in an automatic fashion”. Unsurprisingly, Microsoft is at it too. A recent research paper (www.microsoft. com/en-us/research/publication/model- ensemble-click-prediction-bing-search- ads) from its Bing search unit notes that “even a 0.1% accuracy improvement in our production would yield hundreds of millions of dollars in additional earnings.” Alibaba is on a similar wavelength. To better “extract users’ interest by exploiting rich historical behaviour data”, crucial for building a CTR prediction model, its data scientist proposed (https://arxiv. org/abs/1706.06978) an industrial scale Deep Interest Network to be developed and deployed within Alibaba’s display advertising system.
According to US academic, and former advisor to the Obama White House and Facebook, Dipayan Ghosh, AI will increase the speed of ad mediation, inundating users with content tuned to their personal desires. In an op-ed in the New York Times , he predicts: “It will abet the seamless and accurate development of ‘look alike’ audiences enabling advertisers to upload their customer lists and automatically send ads to like-minded people that they do not already know. And it will enable automated contingency-based marketing, allowing clients to programmatically trigger certain kinds of content to be shared in the moments after real world events transpire.” The fallout from the Facebook scandal has everyone scrutinising exactly what data tech giants hold. Google basically knows everything you’ve ever done on its platform – every email you’ve ever sent or was sent to you via Gmail, every Google search you’ve made – and deleted – and every Google Ad you’ve ever viewed or clicked on. It’s worth remembering that Google also owns YouTube. Facebook has a similar database on all its users but bowed to pressure and (blaming a bug) is now deleting all videos filmed using the company’s desktop web camera tool, which most users thought had been deleted anyway. The company is also requiring advertisers to actively certify that they have permission to upload contact lists for targeting customers on Facebook, in order to better preserve user privacy. You don’t have to be Elon Musk to feel nervous about this. Even erstwhile Trump chief strategist Steve Bannon pledged his allegiance to restore “digital sovereignty” at a recent Financial Times event. He defined this along the lines of reclaiming our own personal intellectual property from big tech firms albeit as part of his wider nationalist polemic. “As the industry integrates AI into digital advertising, disinformation operations and legitimate political communications will gradually become concerted, automatic and seamless,” Ghosh suggests.
Powered by FlippingBook