Using Machine Learning to Build Brand Trust

When the Luddites burned the machines that threatened their jobs in the 19th century, they expressed a popular sentiment at the time – the distrust of new technology. But two centuries into the future, humans are building machines that are becoming increasingly intelligent. Artificial intelligence (AI) has entered the mainstream, and more sophisticated machines are now being used.

Nonetheless, skepticism remains in the creative industries. After all, many people think of brand trust as something subjective and impossible to quantify. Brand trust is all about maintaining a level of intimacy between the brand and the consumer. This intimacy comes with human interaction and not with heartless machines. 

Machine learning is changing this notion. In a forum on advertising, CA Technologies Senior Vice President Anna Griffin shared that AI and machine learning help in forming “a tight, quick connection with relevant content and become a beautiful handoff to a human being to create better context and to better serve someone and ultimately be more trusted.”  

This assumed connection between ‘better context’ and ‘better service’ is based on the related concept of ‘hyper-relevance’. To build brand trust, Information Age notes that companies can use a wide array of information to craft personalized and hyper-relevant customer experiences.

These kind of technology is already seen in content aggregation service providers such as YouTube, Spotify and Netflix, where users face a staggering variety of content. Thus, the services aim to curate relevant and substantial content for the user. In order to predict what type of content a user wants, AI can be used to track user behavior, preferences, and other factors. And because user preferences evolve, the service should also be equipped with learning capabilities that enable it to keep up with users. Maintaining ‘hyper-relevance’ is therefore key to keeping the user engaged with the service, paving the way for brand trust.

But while hyper-relevance is clearly important in building brand trust, it is not enough. In fact, according to PageFair, ad blocker usage increased by 30% in 2016, and it doesn’t seem that it will get any lower. 30% of users cite security as the number one reason for installing ad blockers, while interruption is the second most cited at 29%. PageFair’s head of ecosystem Johny Ryan surmised that “privacy and data leakage in ad tech” as well as “the industry’s lack of an approach or interest in privacy” contributed to this apparent lack of trust from users. 

Machine learning is poised to address this issue, as it also aims to incorporate macro elements, most important of which is proper timing. Users do not want to be on the receiving end of a relentless barrage of ads. Sophisticated machine learning models are now taking into consideration macro patterns that can influence how a user may perceive brand interaction. 

This may mean using more specific permutations of machine learning that demand a certain level of human intervention. Ayima’s data scientist Simon Löfwander details how marketers can use ‘supervised learning’, a specialized marketing approach that leverages machine learning. It can be used to project how a business like e-commerce will perform in a set period of time. The process incorporates data such as market trends, seasonality, and media spending to assess how well the brand can acquire and/or retain customers. Supervised learning’s defining feature is how the training set is monitored in such a way that the test input is applied systematically. This allows you to determine exactly which aspects of the business model works and should be retained, rather than just shoving all sorts of data into the model and let it do the job.

Of course, getting data from users is the tricky part, as not all users want to share their browsing preferences, shopping behavior data, and other sensitive information. This is, in fact, where brand trust and marketing reach an impasse, as consumers will only share data with brands they trust. Ironically, getting this trust in the first place is precisely the problem.

This is where creativity and machine learning join hands again, as brand trust is more than just about giving consumers what they want and expect. It is, more than anything, about building a relationship that accepts the consumers’ complexity. As discussed in a previous post here on Bulldog Drummond, data is not the customer. Humans are creative beings, and to gain their trust, you need to tap into their creativity and complexity. This extra mile is where the hard work starts.

Lastly, an important variable to keep in mind is the provider’s responsibility for proper and ethical data management. More than relatable campaigns and personalized experiences, customers expect their favorite brands to respect their data. If we put the power that is in our hands to good use, we make ourselves worthy of our customers’ trust.