To thrive in today’s data centric environment, many organisations are turning to artificial intelligence (AI) systems to maximise efficiencies and increase return on investment. In fact, according to recent research from IDC, AI spending in Australia is predicted to reach $3.6 billion by 2025.
Let’s start by taking a quick look at what we mean by the term AI. AI is a broad umbrella term for pretty much any technique where a machine can mimic human behavior. Under the AI umbrella there is machine learning and deep learning. Machine learning uses previous outcomes to improve the accuracy of future results, while deep learning uses algorithms that enable software to train itself in specific tasks like speech and image recognition. In other words, deep learning mimics human thought processes.
As digitalisation continues to rapidly evolve, advances in AI are proving to be invaluable. And this is especially the case within the realm of email marketing where personalisation can make the difference between an effective campaign, and one which fails to engage subscribers. In a world of inflation and rising cost of living, personalisation is seen by almost all marketers as the key to success. Moving forward, it will be crucial to see how marketers utilise AI with personalisation in mind.
Benefits of AI
Why is AI driven personalisation seen as the key to success?
Firstly, it’s worth noting that personalisation has rapidly evolved over the last few years from simply using a person’s name in a message in the earlier days, to today where consumers expect brands to identify their needs and interests, and tailor communications accordingly.
The ongoing pandemic and increase in cost of living has seen many consumers become more cautious with their finances. With many households now having tighter budgets, brands must focus on creativity to connect with target audiences and maintain and build loyalty. This is where AI can make a difference and assist marketers to distribute relevant and engaging material.
AI can follow the human brain’s cognitive process and learn to recognise patterns, make predictions about behaviours, and draw conclusions. For example, AI can be used for predictive eye-tracking to help marketers determine which aspects of their email draws the attention of the recipient’s eye most, and in what order. In fact, research from the Nielsen Norman Group shows less than 50% of email recipients scroll further than the initial page of an email. Therefore deciphering and clearly displaying the offer which is most likely to interest the customer is critical for engagement and a potential sale.
AI is also ideal for offer optimisation and is commonly used by marketers for ‘next best offer’ selection. This is especially helpful for emails that contain multiple promotions within. In this case, AI will analyse customer behaviour to determine preferences. Senders can then use this information to order offers that best match the customer’s interests and maximise probability of engagement. We’re also seeing examples of senders who are using AI to predict when their customers will shop next and send them an email ‘nudge’ if they fail to do so. Based on this approach, some marketers are even seeing an increase in average order value and customer retention.
Although AI may seem like the perfect solution for email marketers, it’s important to understand where else this technology is harnessed — like mailbox filtering and major email security solutions. In fact, it’s AI that prevents emails from being delivered through its ability to analyse human behaviour which is why positive subscriber engagement is crucial for deliverability. By understanding how and where AI operates more broadly, marketers can better recognise how to successfully apply it within their own campaigns.
Another potential point of concern is how bias can derail AI solutions. AI bias comes from the data that is collected and used to train machine learning models, and can reflect the cultural and personal biases of the people building the machine. This can produce unexpected results. A famous example of AI bias is the program created by Amazon, which was designed to analyse candidates’ resumes and search for future talent. The hiring tool used AI to give each candidate a score, which was based on applications submitted in the past. But it was later revealed the system was gender biased. Due to the male dominance in the tech industry, most applicants in the past were men, which resulted in the system teaching itself that males were the preferable candidates. This is just one of many real-life examples of how bias can be introduced into AI models.
Natural Language Generation and AI
Pre-AI, teams looking to launch a marketing campaign would brainstorm and draft content based on what they thought would best grab the customers eye. But today, multiple platforms exist, such as Phrasee, which combine Natural Language Generation (NLG) and deep learning to generate optimised text and boost customer engagement.
There are plenty of different ways to phrase a call to action like ‘buy now’ or ‘read more’. Platforms which use NLG technology allow users to browse the multiple phrase options, which are based on the language customers are historically most responsive to. While this is a helpful tool, machine learning needs to be used alongside it to rank phrases from most effective to least effective. An advanced dynamic optimisation system is then necessary to sift through the ranking to test them with real-time optimisation—refining the language to land on the most effective option.
The future is bright
Platforms powered by AI have revolutionised customer experience by creating a tailor-made digital journey for customers across all channels. AI’s power in email marketing is undeniable—it is allowing organisations to competitively connect with customers like never before.
We’re still in the early years of its development, and for marketers, this is an exciting time. In the field of email marketing, which is quick to evolve, there is plenty of potential for further innovation. I’m excited to see what possibilities AI will reveal next.
Guy Hanson, VP of customer engagement at Validity