Alan Gourdie on when A.I. technology becomes commonplace and how NZ businesses should start preparing now

Over the last 30 years we have witnessed the very nature of marketing change. We have seen the focus shift from creative advertising, through digital marketing and mobile apps to the current focus on data-driven engagement and data-centric marketing. The future, however, will be machine learning, artificial intelligence and predictive marketing.

According to New York Times journalists, John Markoff and Steve Lohr, many of the tech industry’s biggest companies like Amazon, Google, IBM and Microsoft, are jockeying to become the go-to company for A.I. by 2020. The market for machine learning applications will reach $40 billion market research firm IDC estimates. And 60 percent of those applications will run on the platform software of just four companies — Amazon, Google, IBM and Microsoft.

In the very near future, intelligent software applications will become commonplace and machine learning will touch every industry. Today, only about one percent of all software apps have A.I. features, IDC estimates. By 2018, IDC predicts that at least 50 percent of developers will include A.I. features in what they create.

We are already seeing an epochal shift in how businesses operate and compete with the development of a new generation of companies using this type of technology and engage with customers one-to-one. These companies use data to personalise and predict what its customers are going to do next and then engage them with intelligent marketing programs.

The most disruptive companies in the world, for instance Uber, Amazon and Netflix, have one thing in common. And that is that they drive business growth and rapid global roll out using data-driven, personalised, predictive and responsive engagement. Those companies that can unlock the power and value of data are seeing higher engagement and viewership and higher revenues from customers. For example, the majority of Amazon’s and Netflix’s sales are driven off predictive marketing engines using machine learning technology.

The impact of these companies on the New Zealand marketplace is clear to see. Many New Zealand businesses are now in industries or verticals that are being disrupted or threatened by new competitors who have already adopted machine learning technology.

Machine learning technology delivers much deeper customer engagement. Typically, three to five percent is the average conversion rate from rules-based marketing. Machine learning-based marketing typically delivers 10 to 20 percent in the first year and then ramps up to 30 to 50 percent over the next three to five years.

Only the fittest will survive in this new generation of marketing. When a competitor is getting 25 to 50 percent conversion rates on its offers by using better technology, New Zealand businesses can’t afford to be getting conversion rates of three to five percent.

The reason for these lower conversion rates is that most New Zealand companies still use rules-based, segmented marketing, where a customer is put into a ‘persona bucket’ or into a predefined segment. But in this new generation of marketing that is not going to work anymore. To put it bluntly, rules-based technology is outmoded. Instead, New Zealand businesses need to build intelligent engagement programs that use a machine learning capability which will enable it unlock the power of data.

The problem with segment driven, marketing automation is not just low conversion rates. Rules-based marketing automation systems are actually fuelling consumer frustration and increasingly creating brand damage. And as more and more organisations put marketing automation into the business with rules that are dumb, it creates more and more clutter in peoples’ inboxes. In short, customers are disengaging from brands and hitting the ‘delete’, ‘dump’ and ‘unsubscribe’ buttons.

When you look at the statistics in major global studies investigating loyalty programs in the USA and the UK this becomes evident. Cap Gemini, for instance, found that 90 percent of commentary on social media on loyalty programs is negative. Why? Because, as the leading loyalty experts Colloquoy found in its studies, 50 percent of all respondents said the offers weren’t relevant. Within that, 49 percent of 18 to 35 year olds said they were actively starting to disengage from those brands that were effectively spamming them. And that’s how we get to the remarkable figure from Cap Gemini that 77 percent of all loyalty programs launched in the US and UK in the last two years have failed.

If this is the problem then how do we fix it? In order to compete, New Zealand businesses have to be able to send personalised offers and messages; send offers that predict a customer’s behaviour and be responsive so that if customers don’t like an offer or don’t respond to it, then they don’t get it again. And finally it has to be real time and contextualised, meaning that if a customer is in a particular place then show them offers that reflect where they are and make those offers easily and instantly redeemable.

So what do you need to be doing if you want to adopt machine learning in your business? Firstly, you have to be ready to start on a journey. Think about your level of programme and data maturity. What data have you got? What form is that data in? Have you got any data on past promotions and any feedback that you can use? If you don’t have the data, then you need to start putting together a data capture strategy now.

Each organisation will be at a different level and you need to be honest about what stage you are at. Some organisations do nothing with the data, while the most advanced will probably be using rules-based marketing. Once you have data, you can start training up the machine learning based on this historical data. Then the machine learning will go in and start to do pattern recognition on the data.

If you’re still not convinced, maybe this example will help. In recent years, the supermarket giant Tesco dropped the amount of offers it sent out by nearly two-thirds but massively increased its revenue from conversion and ROI by using this type of technology. Tesco achieved a 675 percent growth in its bottom line, as a result of its data-driven loyalty programme. By analysing customer data, Tesco found that 80 percent of the discounts and offers utilised by customers came from 20 percent of the offers generated. This prompted a reduction of offers from 750 to 300 a year, amounting to approximately $600 million savings in promotion expenses while increasing market share.

Machine learning creates enormous efficiencies because you no longer have teams of campaign managers generating and managing hundreds of offers – many of which are actually delivering negative ROI. So instead of sending out 100 offers, you can send out 30 offers but each one of those 30 offers will have a much higher conversion. And not only is that a much better use of a business’s time and energy but it will also surprise and delight your customers when they start to receive offers that actually matter to them.

What is machine learning marketing?

The best way to describe machine learning is to focus on the learning piece. Machine learning uses algorithms that interrogate large data sets to look for patterns of behaviour down to n=1. It then uses that manipulated data to predict what that customer is most likely to do in response to the engagement activity being generated. The critical piece is that machine learning learns from the data and the customer behaviour. And the beauty is that it’s all automated and the algorithms adapt to the behaviour of the customer.

Machine learning doesn’t work on a predefined target segment model. It works on an outcome model meaning that it looks for the customers with the highest conversion likelihood at the lowest possible cost. It then goes into the database to find those customers and ranks them and that is the list of customers you will get your offer served against.

Alan Gourdie is the former chief executive of Telecom Retail NZ and founder and managing director of Quantiful, which focuses on machine learning, data-driven personalisation and customer engagement. 

About Author

Comments are closed.