AI and machine learning has been shaping our experience as consumers for some time now. From the ubiquitous Google search to Spotify and Netflix recommendations, or asking Siri or Alexa just about anything, we haven’t really resisted the idea of bots making our lives that little bit easier.
So why is it only now that AI is on every marketer’s lips?
“I think there’s now a democratisation of AI,” explained Marketo’s Customer Success Director Mike Handes at ADMA’s recent Town Hall, Marketing in an AI-first World.
“Technology vendors have embraced it, making it more accessible for marketers to create greater engagement or better outcomes for customers. Also, there’s now more trust in the machine. In the past, the idea of automating communications brought a sense of lost control.”
Most marketers are already experimenting with some form of AI-enabled marketing – even if it’s just a self-optimising campaign on Facebook. But if you’re ready to take the next step into more efficient, scalable and personalised strategies, where should you begin?
AI Town Hall, Sydney
1. Start with the outcome
There’s no point in experimenting with new technology for its own sake: you need to work out first what problem it could solve.
“AI can help us do things that humans just can’t do fast enough or smart enough. But if you’re not setting out to achieve a certain goal at the beginning, you’re not going to know what will come out of it,” explains Beyond Intent cofounder Aryeh Sternberg.
Here are just a few questions to consider:
- Can AI play a role in developing better products?
- Are there customer journey friction points it could remove?
- Can it improve customer experience?
- Could it make marketing operations more efficient?
- Are there communication campaigns we could automate?
- Can it help us personalise our most important marketing messages?
This all begins with a clear understanding of your customers, and what matters most to them.
“One of our big audience insights is about lifelong learning, so for me it’s about machines learning your behaviour and suggesting new things to try, like recipes, using voice activation. I don’t think that’s too far away,” said Brigitte Slattery, Group Head of Digital and Marketing, Lifestyle – Foxtel.
2. Validate your data sources
Machine learning will only be as good as what you teach it – and that means the quality and quantity of data you can feed into the machine.
"The customer has told one part of the business a thousand times what they want – but the marketers over here have no clue. Until we can break that down and give AI access to learn from everything, we're stuck."
Aryeh suggests looking across business channels, noting so much data exists in plain view. “For example, in retail, there’s tonnes of data in your CCTV feeds – hotspots, traffic flows, demographics… if AI can help you use it to solve a problem, that’s great.”
There's still a lot of work to do when it comes to connecting the dots across business siloes.
“Data exists, but it is separated,” he said. “The customer has told one part of the business a thousand times what they want – but the marketers over here have no clue. Until we can break that down and give AI access to learn from everything, we're stuck.”
3. Narrow down your options
Brigitte admits multiple agency, platforms, tools and data sources can make it feel a bit overwhelming.
“We use the DVF principle to assess our options: desirability, viability and feasibility.”
- Desirability – is it going to help you meet customer needs or solve their problems?
- Viability – will you get a commercial return from it?
- Feasibility – do you have the capability, skills, time and resources to implement and manage it effectively?
4. Be realistic about your resources
Feasibility is probably the deal-breaker because while AI promises to make marketing so much more efficient, it still requires a lot of human effort.
“To me, our biggest challenge as marketers is this: I can tell my clients the number one message they should be delivering, but unless you combine that with the right content it's not going to be engaging,” said Mike.
“I can’t at this stage envisage a machine having the human-like capacity to do that creative element,” admitted Aryeh.
Brigitte described the example of Foxtel’s ‘behind the scenes’ Facebook Messenger content for Wentworth. Although it was served up by a chatbot, “it took a lot of resources. They went in thinking it would be quite automated, but when you’re extending the story there is only so much automation can do.”
Keeping a close watch of your machine’s moral compass is also something only humans can do – for now. “Unfortunately there is still potential for data bias to take hold and manipulate outcomes,” noted Aryeh.
With all this in mind, it’s exciting times for marketing.
“AI holds a world of possibilities for us,” concluded Brigitte. “I think it will change the way we communicate with customers, and also the experiences we offer them.”