What’s driving machine learning in marketing? Part – 1

Machine learning will drive the new age marketing

Machine learning will drive the new age marketing


I’ll devote a lot of this series on delving into how machine learning is disrupting the business of marketing. But before we embark on that path, let us take note of something important.  And that is the field of marketing itself is witnessing technology-driven landscape level transformation. Marketing automation is gradually engulfing its wings on everything considered digital marketing and beyond. A lot of other tech interventions are also occurring under the umbrella of MarTech. In these contexts, more than ever before, machine learning as a concept is lending credibility and purpose to the claim of better outcome from technological interventions. Arguably, AI can help with managing marketing activities at unparalleled scale. It can also aid intelligent decision making on high scale data, something that is considered far beyond the realms of human possibilities.

Broadly, the premise of machine learning for marketeers is to help them understand their audience better and drive engagement. So, the realm encompasses everything from creating content that is more engaging to advertising that is highly personalised. Add to this everything in between that optimises the probability of user engagement. Hinging on that hope, everyone is professing that machine learning will dominate the new age marketing. But is that optimism borne because more and more people are going online and there is data and computational availability for analysis and decisioning? Not completely. The underlying reasons are manifold. In fact, exactly five factors are primarily driving us into this future. And they are –

  1. New data feeds availability – With new tools and tracking mechanism, user’s online footprint expanse today is larger than ever before. And marketers can leverage that data in deriving actionable insights in ways never thought before
  2. New consumer products – Voice searches and its mass evolution into virtual assistants will herald the way consumers will interact with media. And with evolution of IoT and connected devices, the touch-points will also further broaden in spectrum. Technology will increasingly enable each consumer action to be two-way communicable
  3. New content and ad formats – Voice content, new image formats, vertical videos, 3D, AR, VR and the list goes on. Every new quarter, we usher into a new paradigm where something breathtaking is shaping the way we are consuming content and ads. Half of it is uninhibited tech disruption. The remainder half, is driven by the consumer experience focussed businesses that are forced to innovate to keep the format seamless and unobtrusive, yet money making
  4. Intelligent inventory buying and targeting – Inventory buying and targeting is going programmatic and it benefits everyone. While in the short-term, some might complain that the returns aren’t as great as expected, its not before long that everyone concurs that it is the future
  5. Only AI can handle processing at scale – This is undeniable that once the scale of any data stream reaches disproportionate levels, it is beyond the capabilities of human teams to analyse and decision. Also, the value of decision being taken real-time is all the more important than before.

But how does all of that translates into focused actionable activities for marketeers? Let us look at the layman’s view of the marketeer’s job. She should –

  1. Get more eyeballs to content
  2. Do everything to increase probability of user engagement
  3. Increase goal conversion probability
  4. Reduce churn
  5. Build a self-sustainable and high growth marketing ecosystem

All sounds promising, right? In the next few posts, we’ll look at each of these separately and see how excitingly AI is disrupting the arena. In this series, today we begin with leveraging machine learning for getting more eyeballs.

Getting more eyeballs

Content Generation
One way to get more eyeballs is to create more engagement-worthy content. Machine learning tools these days can convert data into sensible text. It is time saving and good enough not to be deciphered as a AI-written content. A prediction by  Gartner says that by 2020, 20 percent online content will be authored by machines. While we aren’t expecting AI to churn out literary masterpieces anytime soon, but a majority of content generation tasks such as news, snippets and marketing collaterals can be helped by the all encompassing learnings of AI tools.

To begin with, AI NLP (Natural Language Processing) algorithms can summarise content. It has to be noted that summarising of content can be either “extractive” or “abstractive”. The former is something that is already used. A tool would pick up the first few lines from each paragraph and keywords. The latter is AI based and requires a thought process and intuition to work.

It can perform contextual search. It can help content editors and writers in AI assisted smart content writing. When I say smart content writing, it refers to creating narrative out of trained or observed datasets. This will be applicable in writing earning summaries for a company’s financials by looking at the datasets. It is also applicable to a marketeer writing a marketing copy and the AI tool auto-suggesting the language to be used for highest effectiveness. And in some cases, write the entire content. An example of one such startup is Narrativa.

Content Optimisation
AI can optimise content for a user, basis her engagement in the past or what the engine can learn about her – Whether the data collected is interest based, affinity based, demography based or past activity based, most web platforms these days rely on recommending content and products that are more relevant to the users, thereby increasing the probability of conversion and higher degree of engagement. An example of one such startup is MarketMuse.

Search Engine Optimisation (SEO) and Search Engine Marketing (SEM)
Both have taken the early path to incorporate machine learning. Search engine leader Google also last year made a public announcement on how it was going to increasingly rely on AI signals for SEO. Building PPC (Pay Per Clicks) intelligence is the way forward. Even tools such as Google AdWords are building more and more necessary automation to leverage ML and AI. AI can expand the plethora of features that can influence user behaviour, analyse their combinational impact without limitations and leverage reinforcement learning algorithms. A few years ago, Google made the tectonic shift towards AI in SEO when it brought the Rankbrain algorithm to augment Hummingbird, the contextual content AI.

Virtual Assistants
Virtual assistants are voice search driven. Also, strongly linked to it in the near term of marketing via the virtual assistant route. But essential to that game play is the decoding the voice search paradigm and associated SEO and marketing strategies for the future. An optimum step to gain foothold in that market is to be voice-search ready. With time, it is inevitable that consumers will converse with the voice assistant for seeking local business information, making bookings, interacting with connected devices, streaming content, etc. From that perspective, it is imperative that marketeers realise the emerging prominence of virtual assistants as the preferred tool of communication with the internet.

Programmatic Media Buying
While most consider programmatic advertising (or media buying) as a winning method to burn remnant inventory, the method also takes a lot of headache from targeted advertising PoV. Targeting ads to the most relevant customer via the machine learning route is the core principle of this method. Propensity models are at the core of this activity.

Media monitoring
With the advent of photo-sharing apps and viral content, an outdoor billboard today can potentially have far more impact than ever imagined before. And with the demand for being able to track that impact increasing by the day, analytics and AI are making giant strides in helping companies track and measure brand exposure during commercials, events shown on TV or social media. Companies are applying computer vision to pictures, videos and social media feeds to give a complete picture of traditional media outreach. An example of one such startup is Signal-AI.

In the next post, we’ll talk about the AI potential in increasing probability of user engagement.


Anubhav, a data scientist, writes about new developments and future trends in the machine learning and data analytics domain.
He can be reached at anubhav@thinkbigdata.in
Follow him on Twitter at: https://twitter.com/think_bigdata

Leave a Reply