Turning the AI Hype into marketing reality

Recent step-change advances in Artificial Intelligence and cognitive computing are set to make a genuinely massive impact on marketing. So it’s time to sift the hype from the hope and identify what this will look like over the next couple of years.

Recent advances in AI underline both its potential and power. One such area is its ability to consume massive amounts of structured and unstructured data and drive incredible insights through ever complex algorithms and in some cases, automate decision making.

“I’m astounded about how AI is going to transform every industry,” said Leslie Fine, VP of Product at marketing-tech giant Salesforce. “Companies today face an imperative to integrate AI into their products and services, or risk becoming less competitive with companies who are applying AI to improve customer experiences and make more intelligent business decisions.”

So how exactly is AI doing this? How should we expect the marketing landscape to change in the next years? And what does this mean for the humble marketer?

Step by step

One thing that’s not likely to happen any time very soon is brands replacing their entire marketing stacks with AI-driven platforms. Rather, companies are experimenting and refining AI-driven solutions focusing on isolated tasks. I’ll take a look at some examples of these in a minute, but thought it might first be useful to look at how and why it is important right now before we get on to exactly what it looks like.

A thoughtful blog by Advertising Age asks what AI needs to do to impress the marketing cohort, and looks at how AI works – what are the human evaluations and decisions that AI has to turn into algorithms (and then to keep learning from and evaluating) in order to cut the mustard? It proposes a Five Step Programme.

1.    Understand why marketers do what they do

2.    Teach technology how to understand abstract information, such as creative

3.    Program it to consider all scenarios before each move

4.    Make individual building blocks work together as a holistic system

5.    Introduce checks and balances to make sure AI doesn’t go rogue

While the last of these raised a big smile from me (through Science-Fiction thoughts of a huge AI machine which once activated would come to run and replace reality à la Matrix) the others are very useful for considering just what a big job AI truly has on its hands. As human beings we instinctively know which images or combinations of words (for example) have more of an effect on us than others. How is a computer to make the same calls correctly? And that then needs to roll out into the string of decisions that need to be made by marketers:

What’s the best time to communicate with a given individual? If we replace a heading in one communication how do we evaluate its impact qualitatively rather than quantitively?

The imitation game

There are many examples of AI in marketing action. Sticking (somewhat arbitrarily) with the theme of five, here are five of them.

• Recommendations and content curation

We all remember the bad old days of recommendations which weren’t at all smart or sophisticated – “Customers who bought ‘Elton John’s Greatest Hits Volume I’ might like ‘Elton John’s Greatest Hits Volume II’”. Today, companies like Netflix pour serious money into getting their recommendations right as they know what’s at stake: recommend the right Box Set for someone and you’ve got them for another 8 to 80 hours…

• Identifying and preventing fraud

This is an interesting area, as it is – although not everyone likes to admit it – something of a double-edged sword: theoretically, AI has the potential to aid the fraudsters as well as to combat them. But the sheer power and scale of AI, combined with the machine-learning ability to continue learning, means AI will be able to spot patterns, identify non-human intruders, and uncover new scams, faster and more reliably. As companies move away from traditional name/password methods of verification, AI will also come into its own as a means of providing ‘left-field’ methods of individual identification.

• Speech recognition and language recognition

While we’ve probably all had some fun with the likes of Siri and Alexa mis-recognising words and questions, the technology driving these is unquestionably very impressive and improving exponentially – through learning, through constant automatic refining, all the time. Language recognition is of course the all-important corollary to speech recognition. Which brings me to Natural Language Processing, which is where things get very interesting around AI’s ability to accurately process unstructured data – for example you’ll be able to type in “We want to rent a family car which doesn’t have to have loads of legroom but does have to have big and versatile baggage space” or “We’re looking for a holiday no more than two hour’s flight which combines good beaches and good cliff/sea walks of low to medium ability” and the computer will generate relevant recommendations.

In fact the whole area of ‘recognition’ is extremely fertile ground for AI – we’ve not even got time and space here to touch on image and facial recognition.

• Sales forecasting

There are many ways Artificial Intelligence can improve forecasting but one which is set to hit the ground running is the capacity to revolutionise how companies view inbound communication. AI and machine learning can filter and analyse inbound emails and (based on past behaviours and conversions) suggest appropriate next actions.

• Bots and messengers

Like many others I see chatbots as probably the most interesting and immediate application for AI – so much so that I’ve written about it in much more detail in another blog which, if you’ve enjoyed this one, I encourage you to read. And if you’ve absolutely never heard of chatbots before, there’s every chance that you may have engaged with one online or via mobile text as these are, put simply, applications that enable a computer to ‘hold a conversation’ with a human. Such high profile brands as Facebook are putting their faith in a bot-dominated future. Salesforce’s LiveMessenger Chatbot is leading the market in helping companies across verticals take the plunge. Be advised: chatbots - in one form or other - are here to stay.

A brave new world

As Leslie Fine of Salesforce suggests, applying AI to marketing scenarios and marketing technology is a way of creating an advantage. The ability to run AI is now in the hands of everyone because the power and data to run it are within reach of everyone. It’s not reliant on a mainframe computer - it’s already present in ad-tech and marketing automation platforms.

AI means the role of the marketer is changing. On one level it’s natural that this might seem a bit scary, as it means “relinquishing control and trusting the technology to do what they've traditionally relied on complex technologies and teams to handle -- but at a far greater pace and scale.” But with all eyes on the prize, the potential rewards are great. AI is the key to rescuing marketers from the big data hole they’ve been dropped into and placing them back into the strategic, creative side of marketing.