从新奇转向必要：机器思维时代市场营销的应用与预测（英文版）.pdf

2018 iProspect | All Rights ReservedASIA PACIFIC Iain Addy Mark Byrne Joanna Catalano Dan Kalinski Rohan Philips Nate Shurilla Bowan Spanbroek EUROPE, MIDDLE EAST AND AFRICA Scott Abbott Morten Bruhn Hjsgaard Aurelien Loyer Jack Swayne THE AMERICAS Jake Hewlett Jeremy Hull Sam Huston Misty Locke Nate Nicely Varun Pramanik Catalina Salazar Ron Smith National Head of Paid Media and 48% believe it will allow them to automate manual tasks so they can focus on strategic goals. Fewer than 2% think Machine Learning will have minimal to no impact on their business in 2018. With a better understanding of the technology , clearly defined business goals and the right data, brands view of Machine Learning will shift from a futuristic idea to an actionable solution that integrates data, technology , and real-time activations for superior campaign performance. ROHAN PHILIPS Global Chief Product Officer, iProspectMachine Learning wont fundamentally change the role that marketers play; instead it will enhance what marketers can do by helping us optimize actions, extract hidden structures from within the data, summarize data in real time into more concise views and descriptions. It will help us move our attention from the tactical to the strategic components of marketing, augmenting human experiences and enriching human competencies. CHRISTI OLSON Head of Evangelism for Search, MicrosoftiProspect/FROM NOVELTY TO NECESSITY/20EIGHTEEN Brands are in need of transformational business performance. Marginal gains are no longer enough to define in the digital economy. The advancement of technology has given rise to the expectant consumer who demands brand interactions to be hyper- personalised and hyper-relevant. For half a century, computer scientists have been experimenting with Machine Learning and reveling in the possibilities it presented. While their theories and projections have been compelling, it was hard to imagine a day when Machine Learning would actually solve a business challenge or have an impact on our daily lives. Today , however, we clearly find ourselves at the intersection of available technology, the mass accumulation of consumer data, and the human ability to apply data science to solve commercial challenges and realise profitable business results. The implications of this intersection will change the face of marketing forever. Every time Google autocompletes a search, Amazon recommends a product, Spotify queues up a new song, or a device recognizes a voice command, people are interacting with algorithms and smart machines. Consumers are more attuned to this, and they see the benefits. It no longer feels like science fiction. In the midst of all this progress, marketers and consumers alike can get lost in the industry hype and terminology . The term “ Artificial Intelligence” often calls up thoughts of dystopian futures, and the term “Machine Learning” sounds interesting in theory , but what is it really? For the purposes of this discussion, here are iProspect s definitions: MACHINE LEARNING ma.chine learn.ing /me SHen lerniNG/ noun The customised computer process that algorithmically trains a program to independently recognise trends in data to deliver increasingly accurate and efficient results. Machine Learning makes it possible to respond quickly to evolving consumer expectations and behavior at a pace and scale that is not humanly possible, and it provides insights into real-time patterns and trends that were previously indiscernible for humans, and therefore largely untapped. ARTIFICIAL INTELLIGENCE artificial intelligence /rde fiSHel in telejens/ noun An intelligent and powerful programme that analyses and categorises information without human assistance with an unprecedented level of accuracy and speed. A.I. is a breakthrough technology that has been made possible by the innovation of Machine Learning. The applications of A.I. in marketing include deep learning, virtual agents, natural language processing such as voice search, and biometrics enabling facial recognition. 0809The story of digital is really about consumer empowerment and individualism. Consequently, understanding consumer behaviour such as motivations, attitudes, purchasing patterns, triggers and turn-offs, has become the make or break of todays brands. The success of Amazon or Alibaba is not accidental. Both businesses staked their future on cracking the code of consumer engagement through smart use of customer data. Machine Learning, which leverages algorithms to detect patterns in vast volumes of consumer data and actually learns from it, gives businesses opportunity to gain competitive advantage, provide better service, stronger business performance and faster operations. Were now at a point where were moving from understanding to predicting behavioural patterns. This is an exciting time, and there is even more disruption on the horizon. DAN KALINSKI CEO, iProspect ANZMachine Learning should be seen as an enabler of a type of customer data platform. It can add more colour and context to datasets that are manually unmanageable due to their size, and provide marketers with the much needed insight they require to have one-to-one conversations with consumers around the world. We recommend organizations embrace these new possibilities by tearing down data silos across the business, bringing datasets together into a unified view, and investing in training people on how to interact with data in scalable ways via Python, R, and SQL. Finally, it is important to cultivate a culture of experimentation while leveraging the cloud to build, measure and learn from data experiments in search of the next great insight. RAHUL PARMAR AdTech Lead, Google Cloud 12Structured data is helpful, but it demands extensive resources such as hardware, software and people to create and maintain. On the other hand, unstructured data is full of irregularities and ambiguities that make it difficult to understand and trust. Consequently , the first step towards making sense of unstructured data is the time consuming investment of hours of labeling and orga- nizing on the part of the marketing team. As ever-increasing amounts of data become collectable, Machine Learning algorithms are perfect for sorting, ordering, classifying, and enriching that data. Intelligent automation engines can even generate audience groups from historical performance, allocate budgets, and build a bidding strategy . This ability to tackle unstructured data and organize it into something meaningful is a big breakthrough. For marketers, it opens up amazing possibilities. Machines get smarter as they are fed massive amounts of data continuously generated by users interacting with the system. While Machine Learning empowers companies to be more dynamic and exploratory , it s important to note that Machine Learning will never supercede the need for humans in marketing. There are crucial things that machines arent able to do, like develop a business strategy, apply intuition, interpret results and have meaningful relationships. Machine Learning is less about marketing automation and more about the marketer s ability to solve previously unsolvable problems using technology . Marketers are well aware that they need to amass mountains of data. Making sense of millions of data points and then applying the learning to a business goal is another thing entirely . 13 iProspect/FROM NOVELTY TO NECESSITY/20EIGHTEENMachine learning will open up the “great unknown” for every marketer whether that is an unidentified customer segment, an anomaly in your website performance or a different view of campaign performance. This is where Machine Learning is changing the game for marketers enabling them to find patterns and opportunities across massive amounts of data. It allows marketers to achieve new efficiencies by automating some time consuming and repetitive tasks, giving focus to more strategic execution within their roles and enabling brands to deliver more personalized customer experiences at scale. At Adobe, we are passionate about making Machine Learning accessible to marketers, and lowering barriers to innovation for brands with tools like Adobe Sensei, the companys AI and Machine Learning framework. Marketers should understand that AI and Machine Learning are tools intended to amplify human creativity and intelligence not replace it. TATIANA MEJIA Group Product Marketing Manager, AdobeA recent Forrester Research study cited, “Marketers who use Machine Learning are almost three times as likely to report revenue growth rates higher than their industry average. They are also more than twice as likely to occupy a commanding leadership position in the product/service markets they serve.” Until now, brands refined online messages by designing a few different versions, running a campaign, analysing the results, and finally , opting for the version that worked best. Thanks to Machine Learning, 2018 will be the year that personalization at scale becomes truly possible.Media and messages will be better aligned, resulting in a more relevant experience for consumers. At iProspect we believe there are four performance marketing areas where Machine Learning can truly put the consumer at the heart of the media plan. Real-time is now possible at scale. No prior solution has been able to deliver the level of responsiveness that Machine Learning provides. Consumers see offers change by the minute, based on the data they leave in their wake, and this behavior generates millions of data points for machines to use to become smarter; that in turn lets marketers achieve greater accuracy and therefore spend less to convert more. 16 iProspect/FROM NOVELTY TO NECESSITY/20EIGHTEEN“We were able to put Machine Learning front and center with the work we did with iProspect. We needed a smarter way of using our data, and iProspect worked with us to build a proprietary Machine Learning system that analysed more than 12 million individual user IDs and selected the best 400,000. This was then able to produce messaging based on user behaviour and changed advertising in real-time. The impact of Machine Learning was immediately evident, we were able to drive continuous performance improvement, updated the cluster adding 50% of new IDs every week and reduced bounce rate across all our properties by 17%. Its clear that Machine Learning can have a transformative impact on our business and Im looking forward to utilising the possibilities in the months ahead. To succeed in this ever-changing digital economy, it is imperative to be driven by a pioneering and forward-looking mindset: iProspect has been and still is the right partner to compete and win thanks to their innovative approach and solutions.” GIAN LUCA DE SARIO Head of Media Planning from being able to access better segmentation, to applying more accurate performance attribution, to leveraging huge data sets in performance marketing. 18 iProspect/FROM NOVELTY TO NECESSITY/20EIGHTEENEurostar is a high-speed railway service connecting London with destinations like Brussels and Paris. Eurostar asked iProspect to tackle their constant busi- ness challenge: how to sell the unsold seats on their trains, compete with the airlines, and ensure that Eurostar appears on top of the search results when customers query by destination. iProspect had to devise a strategy to answer it in a smarter way, and return better results. iProspect provided the in-channel and customer experience and D2D provided the technical, algorithmic and statistical knowledge. The custom Machine Learning media buying system for Eurostar, powered by iProspect CORE, was able to automate the media buying process and optimize the results at the fraction of a price. The team made smarter portfolio buying decisions, improved search account structures, and automated search query report processing. In just five weeks, CORE reduced CPC by 12%, increased conversion rate by 26%, and drove a CPA decrease of 30%. EUROSTAR SAYS BONJOUR TO MACHINE LEARNING 19