Mastering the Machine: 4 Practical Ways Marketers Can Use AI

An era where robots bring us Deliveroo, cyborg electric cars drive us throughout post-Brexit Britain and an army of sentient kitchen appliances automate every chore - no matter how small - throughout our now smart homes.

The trouble is, for many of us marketers, this a false dawn we’ve heard many times before; a new technology that’s going to fundamentally change the way we live, how we engage with audiences and how we sell. Please, we’re all still holding out for Google Glass or Minority Report style Augmented Reality, let alone waiting for all those hours we invested in Second Life to pay off.

AI is surrounded by a particular murky uncertainty and that’s because it’s more hyped than everything else, ever. Just think, this is an idea that’s always been top of our collective hive mind, one that’s been living with us in the zeitgeist of popular culture for the last 40 odd years.

Who doesn’t know and love The Terminator? RoboCop? Or the Matrix?!

That in itself has created a problem, a problem of fiction over fact, as we immediately think of the hype, the Hollywood Blockbuster, instead of the practical application. And the truth is there are legitimate uses, real applications that can be applied right now to help us enhance our advertising, marketing plans and even product development.  

We need to look over this and take a leap of faith with AI, as the truth is, it will bring about massive changes to our industry. And why do we know this? How can we be so sure? Well, just look at what’s available right now, at the beginning of the shift. There are legitimate AI or machine learning firms who can, even right now, offer us real tools, services and algorithms to help us do things easier, do things smarter and even do things cheaper in a variety of helpful ways.

The challenge is thus: how do we separate the science fiction from the fact?

Here’s a list of four practical ways marketers can use AI to make your life easier (though we're not wholly convinced by number four...yet).

1. Use an open source deep learning tool

Using machines to learn about customer behaviour, actions and interests has been around for sometime - just think of Amazon or Netflix and their recommendation engines - but, did you know, you can leverage those same tools to expose the full range of your products to your audiences? In 2016, Amazon made DSSTNE (pronounced Destiny), its own deep learning platform, open source thereby allowing anyone to use it to build machine learning tools to automate, well anything data related, including the process of understanding what products a customer is most likely to buy, based on a variety of factors, and then serving up that product at the most relevant moment within the purchasing journey. There are also other similar open source deep learning tools available too - including TensorFlow or Theano - which have their own extremely practical uses, from analysing behaviour in real time to recognising imagery then serving up relevant responses. All of these can be used to enhance the way your own eCommerce, content or brand interaction activity interacts with customers.

2. Automate chat

It’s weird to think that people would actually prefer to talk to computers over people but, hey go figure, it’s happening and a lot of brands are already jumping on the bandwagon. We can blame Siri, Alexa or (whatever you want to call) Google’s Assistant for this bot explosion, each based on intelligent systems that not only recognises your speech but the questions you ask them, paved the way for people to want an easy, fast, and alternative way to get answers to questions quickly. Now, at a time when 1.4 billion people regularly use messaging apps, there’s a huge opportunity to use chatbots and automate the provision of information to your audiences quickly, cheaply and easily. In 2017, these chat based tools have spread like a virus across a variety of platforms to help people in a variety of ways. Cleo, for example, is a free intelligent banking assistant that analyses your spending habits and talks to you via Facebook to provide real (and handy insights) on how to spend (or not spend) your hard earned cash. Similarly EMS.ai, currently in open beta, looks to do something similarly but with property. You simply chat to the AI telling it the type of property you’re looking to rent, and it does the hard work, sending you relevant listings and even making the viewings. If you’re still finding it hard to get your head around why people would come round to this kind of behaviour then but put it this way - who would you prefer to talk to, a chatbot or an Estate Agent?

3. Know your customer

Attribution, value and propensity modeling is a big thing right now but it’s basically a fancy way of saying ‘understand the value of what people do, when they’re likely to do it and market to them at the moment’. If it sounds a bit like being a stalker, you’re right, but knowing the answers to these questions can help brands exponentially. Imagine being able to prompt a sale by being proactive and calling a customer when they start dragging their feet? Or knowing which, based on previous behaviour, day is the best to promote your product as it’s more likely to be sold? Disclaimer, it can be complicated but the principles are actually simple on paper - it’s just marketing 101, know your audience. And the fact is, with the right partners and machine learning tools, it’s easier in application than you’d think. The first step is making sure you identify your customers as individuals, then map the points where customers are most likely to purchase or be further along the point of purchase. The second is harder as you’ll need to layer those steps in real time, which is impossible without AI as you’ll need to work through big data in real time. The machine learning tools used by the likes of Ignition AI can do this, literally, in seconds. And that’s exactly what it did when it for the British Army. In 2016, the tool was used to sieve through millions and millions of records to identify not only the type of person who’d be right for a military career but the exact moment when it’s best to contact them. The results were scary; the team acheived a 139% increase in successful Officer applications to the Armed Forces, just by knowing when to serve up an advert at the most relevant moment.  

4. And finally, automate your creatives

For the risk takers and early adopters out there, why not ditch the Creative Director and hire a robot? Creativity is often regarded as too human an element to replace but turns out some people are willing to put the modern day Don Draper and Peggy Olson to the test. Advertising agencies in Japan used robot creatives to produce TV spots and then pitted them against their biological counterparts - check it out, it’s pretty crazy - while thegrid.io uses an AI called Molly to constantly review user interact to design websites on the fly. Still, take all this with a pinch of salt. In the former example, the actual production had to still done by real people while the latter, well, there’s not really a choice on how your site looks, which isn’t great for brand control. Our humble opinion? The tech isn’t there and, in both these cases, it’s eing used as a marketing gimmick rather than an actual tool. With that in mind, we feel that it’s better to play it safe and stick with a proven creative like our own Design Director, Anna Jehan.

The truth that hides amongst all the noise around AI is that there is opportunity for marketers to shift towards a new era, one where we value a scientific approach, based on numbers and fact, over opinion, hearsay or attitudes like ‘it’s what we’ve always done.’ AI, machine learning and automation are all tools that will help us get there - allowing us to create leaner, tighter and smarter marketing briefs.

Written by Mikey Kelk, Account Director at Matter Of Form