When many of us hear or read the phrase ‘artificial intelligence’ (AI), we instinctively think of it in its most grandiose sense, with sophisticated robots performing super-human tasks for the greater good of the world. Using our Data Management Platform, we work with clients on practical ways that AI can be used to make their marketing more effective – and here we’ll share four guiding principles for doing that.

What does AI mean for marketers? On a basic level, deep learning is one of the most powerful AI methodologies, and a class of Machine Learning Algorithm that can be used by businesses to drive, solve or perform specific tasks: deep learning allows technology to learn complex relationships between variables in large amounts of data. From a marketer’s perspective, this can help marketing departments in numerous ways, from personalising customer context to orchestrating the customer journey itself; ensuring those communications are received by customers at the most relevant times and locations. To realise the potential of AI, marketers must discover and embrace its ability to create detailed, hyper-personalised customer journeys – based on accuracy, clean data – quality data, and subsequently, effective outreach in the right context.

1. The right direction – accuracy through AI

While customer journeys are unique, manually creating a different pathway for every single customer would simply prove to be an inefficient – and often impossible – use of resources, especially for large, enterprise-sized businesses. However, thanks to AI’s ability to identify logical patterns among huge amounts of data, technology can learn from the behaviour of all individual  customers, and distil this into signals among the irrelevant noise to create precise journeys, depending on specific rules – regardless of whether that individual interacts on any on or offline channel.

However, if brands are truly determined to use AI effectively in their marketing efforts, they need to take the time to be smart about how it can fit into what they do. Ultimately, what you put in is what you will get out. Before simply diving in at the deep end and feeding your AI software as much data as possible in the hope of seeing instant improvements, businesses must build smart foundations for what they are trying to achieve. Whether it’s to grow their customer base or simply increase engagement among its existing one, the data sets which are fed into AI must contain ‘predictive and behavioural signal’: that is, the customer must indicate intent to take the next step in a journey. Which specific product, e.g. flight, did they click on? Which device or channel did they search from? How many times did they search, and at which moments?

2. Building a foundation – clean data is quality data

No matter what the end goal is, businesses must focus on data quality. This is arguably the most important part of implementing AI technology into your marketing efforts, as the quality of data being used plays a huge role in determining the relevancy of customer journeys delivered. If the data does nothing to better identify your customers as individuals with their own tastes, preferences and shopping habits, businesses cannot expect their AI to perform any magic tricks for them.

This is because despite its importance in determining the success of AI, many marketers will rely primarily on broader industry insights – i.e. 3rd party data –  to inform their marketing strategy. But this is nothing more than a blanket approach to marketing which fails to consider each customer as an individual. Instead, marketers are assuming that the customers of a certain ‘similar’ business act and behave in the same way as the customers of another business.

3. Start small – smartening standardised customer insights

The truth is, this is rarely the case. All you need to do is compare the customers of a budget domestic airline to those of a high-end, international airline to understand how much customers within the same industry can differ. The context for one airline and its unique business model, proposition and customer-base is very different from another, and it can be as inaccurate to relate one to the other as it is to identify a photo of two different animals as the same, simply because they contain the same backdrop.

Start with your own customers to build quality; then, businesses smarten these foundations by using broader industry algorithms – such as an airline using airport data, doing so with the intention of shaping and augmenting the organisation’s existing business logic and customer journeys.

4. Quality control – augmenting with broader industry insights

So yes, broader demographic and trusted partner insights – i.e. 2nd and 3rd party data – can be used to help inform AI-enabled machine learning, but ideally individual businesses should start with using their own standardised customer data to deliver truly effective results. Standardising your customer data should be done on your terms: for example, you decide that someone’s typing of ‘Heathrow’ or ‘Lon’ refers to their search intent for the destination you’ve classified as ‘London’ – not something else. Using this hyper-specific, self-collated data as a foundation for quality, AI can truly meet the expectations that many marketers hold.

On this note, it is always important to remember that there is no such thing as a cookie cutter approach to AI within marketing. Two different businesses might require two different data sets to achieve two different goals, and so every solution must be considered on a case-by-case basis. Also, because we as humans can’t possibly process all of the data being fed into AI applications, we must make sure the data used is relevant enough for us to place our trust in.


There is little doubt that AI will continue to shape the future of marketing – and is already doing so – but the benefits it can deliver to businesses will only ever be as great as the information that is being fed into these algorithms. AI technology cannot do all the legwork for us, and it relies on clean, standardised data – and human business logic – in order to deliver quality results.

This is why marketers need to get their approach and methodology right before kick-starting their machine learning – and the application of these learnings – with new technologies. After all, the real trailblazers in the industry will be those who have carefully considered how AI technology will fit into their marketing efforts, with their own customers, before letting it shape their customer journeys.