AI: The new minds in marketing

Bryan Dollery

by Bryan Dollery

Last month, we took a look at the bots on our mobile devices. Now, we turn our attention to how the AI tech behind them can improve marketing strategies today, and how things could pan out in the future.

The Exhibition of the Society of Model Engineers at London’s Royal Horticultural Hall. On the stage sits a metal, human figure. The machine stands, bows, and opens the exhibition with a four-minute speech. The mechanical man is lifelike, making hand gestures as he speaks, looking left and right to address the entire audience. After his speech, he returns to his seat.



You’ve just read a true story from 1928.

One of the world’s first humanoid robots, Eric was such a great success that he was taken on a world tour, inspiring people with this new and exciting technology. robots were a symbol for what couldbe achieved with machinery. A mechanical man that could wave at an audience has limited practical uses, but the meaning of that development was imperative; people looked at the first robots and asked themselves – what next? If those same hands can be built to equip tools, perhaps they could assist in manual labor, perhaps they could facilitate mass production.

What the model engineers of the early 20th century probably couldn’t envision was the more commonplace bots of today; they don’t move in metal shells but instead serve from within computers. Now, as “machine learning” gains traction in marketing technologies, the same question arises – what next? What can be done by us, as people, with computers that can think? We thought about these very possibilities – hopefully, we didn’t run away with our imaginations too much.


Structure the unstructured

We’ve been banging on about this for a while, but bear with us. The amount of data that is produced daily is incredibly vast. The majority of that data never gets effectively utilized since 80% of it is unstructured. For example, a database sets up a structure, and the data it retains then adheres to that structure. However, for the fragments of data floating around the articles, reports and social media posts on the web, there is no structure to unify them. AI is capable of applying a structure to that scattered data, allowing us to sort and utilize it.

What this means

Current automated content creation tools like Wordsmith are dependent on structured data, being fed in the right format. AI will allow us to generate content based on unstructured data. Imagine that before launching a campaign, you could run an automated report on what your key demographic has been engaging with most in the past week, on what channels and from what locations. Not only would you be able to produce a current, insightful report in moments, but the AI technology behind this could even identify patterns and links that would otherwise go undetected.

Predictive analysis uses data on what has worked before, but cognitive marketing technology can utilize artificial neural networks and algorithms to make predictions based on patterns that we would have otherwise missed. This is invaluable for CMOs needing to process large amounts of data to get a concise picture of their customers’ interactions.



A quick predictive timeline of how the use of AI will continue to develop:

  • By 2017, hundreds of applications will utilize cognitive computing (1)
  • By 2018, 20% of business content will be authored by machines (1)
  • By 2020, 50% of all companies will use cognitive computing in their marketing and sales efforts (2)


Cognitive computing


Understanding the customer

IBM have shifted their focus toward AI technology. Their technology platform, Watson, is available in a few different forms, each suited for different purpose or field. One such offering is Watson Engagement Advisor, a service built to interact directly with your customers.

What this means

Natural language processing (NPL) has come a long way, so much so that Watson can even detect the customer’s tone of voice and relate that to their query. With the pattern-spotting and predictive capabilities we’ve already mentioned, can we expect to see AI tools learn from the queries they receive? Through machine learning, perhaps we could identify which problems are causing the customers the most frustration and detect patterns in the cause of those problems.


For now, human intervention is still needed to ensure the quality and reliability of information that a learning machine ingests. Watson, for example, is fed a corpus of knowledge that is built and curated by humans. If it is already possible for AI technologies to structure unstructured data, perhaps it will be possible in the future for an AI system to build and curate its own knowledge base. Only time will tell.

With Eric being rebuilt at London’s Science Museum, who knows – maybe he and Watson will shake hands one day.


Want to know how your marketing strategy could benefit by implementing AI?