MACHINE LEARNING: the next big thing in advertising

The data flow in 2016 is so fast that the total accumulation of the past two years (a zettabyte) dwarfs all prior record of the human civilization. There is a big data revolution underway. However, the quantity of data is not what is revolutionary, it is the fact that all this data is now usable. Statistical and computational methods of analysing big data are the real heroes in the story. The doubling of computing power every 18 months (Moore’s Law) is nothing compared to a big algorithm. Formally, algorithms may be defined as a set of rules that can be used to solve a problem thousand times faster than traditional computing methods. If broken down, it is nothing more than a new way of visualising and linking datasets to find solutions. Humans have always been far better than computers at seeing patterns. This is exactly why algorithms are a technological innovation. Humans have created algorithms that can now do their job of identifying and working with patterns to create knowledge. The new, ‘created’ knowledge is what we call insights. Algorithms are currently being used to make big data measurable, transparent and understandable at a faster rate. Analysis has become significantly more accurate, which in turn has replaced gut and intuition in the decision making process. Last but not the least, algorithms have provided the ability to create precisely tailored products or services for customers which can be improved with each generation.

Today, there are a number of different kinds of algorithms being used to analyse big data for different purposes. In this piece I will discuss machine learning (ML) specifically and its implications on advertising practices for several decades. Simply stated, machine learning is a collection of artificial intelligence based algorithms and techniques used to design systems that learn from the data that they are fed. ML algorithms are given a teaching set of data and then programmed to answer questions, diagnose problems, predict actions, control variables and outcomes based on that data. It then continues to add to its teaching set with every correct or incorrect answer thereby becoming smarter and better at completing tasks over time.

Marketers and advertisers have collected and analysed consumer behaviour data for a long time. Only it wasn’t big data as we know it today and their tools of analysis were not as efficient. As the amount of data they have access to has grown exponentially, their analytical capabilities haven’t. Therefore, ML is a powerful tool in the business to customise consumer targeting, anticipating behaviour and personalising advertising messages. Having the ability to accurately predict how much a customer is willing to spend on Christmas gifts by analysing their mobile internet activity is a game changer. ML can find these patterns in the data and deliver those results to advertisers in real time. They in turn can have programs ready to act on the information at the right time for the right customer to create maximum influence. It will become much cheaper for advertisers to reach consumers more efficiently over a longer period of time. It will eliminate irrelevant ads and helps consumers find what they need and want even before they look for it. Setting in motion the move towards cognitive, anticipative advertising and marketing.

Several service industries have adopted ML in different forms, Google’s search engine is a popular application of the program. Google observes each user’s search activity to deliver more accurate results with each run, the company also programmatically sells online ad spaces to advertisers. Facebook serves customised ads to its users based on look-alike targeting. Recommendations on Amazon, Spotify and Netflix function on ML programming based on their users’ past selections. Google Assistant and Siri are other popular uses of ML combined with natural language processing. The more Apple and Android users talk to them through their phones, the better they get at understanding accents, requests and delivering results.

IDC’s (International Data Corporation) research manager Gerry Murray predicts that one year from now hundreds of applications in use will be ML and AI powered. The effort to build and deploy ML programs will be easy and very user friendly. Their systems will be hi-tech enough to take signals and draw conclusions like a human brain. Cognitive marketing will soon be the norm with ML based programs making independent decisions. They will deal with tasks like setting segmentation rules that are customised to each user of the service or viewer of an ad without supervision. However, this is still the near future. Technology is taking us to a point where ML’s role will not just be to select the right message and design to deliver to the right person at the right time through the right channel. The program will be able to handle mass delivery of hyper personalised content and service. Furthermore, it will actually be able to have a persuasive two-way conversation in real time.

ML’s independent decision making function will soon become advanced enough to enable the machines to deliver convincing messages, replies and retorts to a wide variety of audiences. While the machines take over many human functions, marketers and advertisers will be skilled to manage these machines, establish parameters to ensure they conduct conversations with consumers in a way that is unique to the company or brand. Marketers and advertising executives will solely handle the governance. Their time will be dedicated to strategy building, crafting sets of creative executions, setting campaign goals (generating awareness, new revenue streams, driving affinity or loyalty) and finding ways to achieve them. ML programs will also predict ad performance before it is rolled out for marketers. Analysis will be based on historical campaign/brand ad performance records or track records of similar advertisers. Good brands already focus on sustaining a strong personality and image through graphics or style of advertising. ML programs will further help brands to exude their personalities through the quality or tone of actual conversations they have with consumers. Their machines will be designed to reflect a specific personality in order to build effective and personalised customer relations on a large scale.

What we are seeing in 2016 is just the early impacts of the merging of marketing and technology. Application of cognitive intelligence to marketing is changing the narrative of advertising to consumers. Its taking the focus away from targeting an audience or a collection of people with similar attributes and bringing it to each individual. It is identifying and responding to their unique personalities, character, media consumption, habit, desires and passions in real time. Advertising executives call this future, marketing to a market of one in order to drive a deep, custom made and personalised brand experience.

“My automated home system had already connected with Google self-drive and ordered me a car… On the way home later I received several invitations along the way to stop or to order dinner for home delivery all based on known preferences, what I ate yesterday, my bio read for today with rating from within my social sphere” - Yoram Wind, and Catharine Hays in Beyond Advertising: Creating Value through All Customer Touchpoints

As advertising executives it is critical that we remain conscious of this heavily intertwined ecosystem of media, advertising and technology. Creating catchy taglines or attractive ads is not enough. Long lasting brands will have to consistently deliver situationally relevant ads using interactive mediums to allow immediate action from consumer. Advertising is increasingly moving towards being a long term multichannel brand experience. In order to remain on top of this wave, advertisers should be fearless in exploiting technologies like machine learning because these advancements come with no limits. Their job should be to find the possible applications of technological innovations in advertising practices. Coalition with digital legs of tech giants like IBM, Accenture or Delloite is a good idea. A combination of resources will help advertisers become better at creatively using technology to create ads instead of tailoring their ads to suit technology. Spotify’s latest outdoor ad campaign is an ideal example of this with headlines that read:

“Dear person who played ‘Sorry’ 42 times on Valentine’s Day, what did you do?”

“Dear person in the Theater District who listened to the Hamilton Soundtrack 5,376 times this year, can you get us tickets?”

“Dear 3,749 people who streamed ‘It’s the End of the World as We Know It’ the day of the Brexit vote, hang in there.”

The years to come will be an exciting time for the advertising industry but access to big data and its application through ML and AI comes with plausible issues. Privacy will be one of the biggest. Through the ad above, Spotify has demonstrated its creative genius however simultaneously it has also made a bold statement about the specificity of its consumer data. The access to data that advertisers have will not be taken away therefore their sense of responsibility must be strong. The line between using and exploiting can be very thin. Additionally, with unlimited access to any consumer’s personal space (especially through digital channels) brands should respectfully give them a choice to opt out.