Ben Aflalo, Product Director at Dynamic Yield, delves into the evolving role of AI in marketing and what marketers and brands should anticipate next as AI continues to reshape the industry.
Throughout history, the basic purpose of marketing hasn’t changed. As in generations past, the role of the marketer today is to encourage consumer engagement in hopes of driving purchases and building brand loyalty. But while the function largely remains the same, achieving the ultimate customer experience has dramatically evolved with the arrival of personalization and artificial intelligence (AI)-based technologies, equipping practitioners with an entirely new and sophisticated set of tools to best connect with their target audiences.
These advancements allow for the automatic delivery of relevant, tailored customer experiences in a much simpler, faster, and more efficient way than was ever possible before.
In fact, according to a Demandbase survey, by 2019, 84 percent of marketing and sales professionals were either already using AI as part of their business operations or were in various stages of planning and implementing their AI strategies.
But what has the journey looked like for brands as it relates to these intelligent systems, how have they influenced the marketing workflow, and where do we see AI continuing to reshape the industry?
Let’s trace its recent history through the lens of Dynamic Yield’s growth within the space and evaluate the lessons learned as well as our predictions for what the future has to hold.
In the late 2000s, AI, and more specifically, machine learning algorithms, were mostly found within theoretical discussions in academia or left to the large tech companies such as Google and Amazon.
It wasn’t until around 2009 that machine learning gained its legs and initiatives like the Netflix Prize, a content recommendations algorithm contest, catapulted the industry’s know-how forward and demonstrated its potential commercial applications within the martech landscape and beyond.
Soon after, new disruptive algorithms dubbed ‘Contextual Bandits,’ ‘Collaborative Filtering,’ and more offered marketers greater accuracy, efficiency, and scale in how they delivered, analyzed, and optimized experiences across the customer journey.
These would eventually become the industry standard, but despite the many benefits and positive impacts on performance, early adopters wanted to better understand how the technology worked.
After all, if AI and machine learning were going to replace the deep product knowledge and domain expertise gained by marketers throughout the years, it only made sense that they would want to know exactly what went into the calculations and why, for example, a certain user was served a particular piece of content over another.
Simply put, AI could not function as a black box.
To meet the marketers’ growing demand for greater control over and understanding of the outputs, the algorithms implemented and technology had to adapt.
For instance, at Dynamic Yield, we introduced additional algorithms that were easy to explain, understand, and predict. This led to high adoption, and subsequently, many improved experiences which generated significant business impact.
We also allowed marketers to A/B test algorithms against the existing control or other algorithms, with the ultimate decision over which strategy to apply left to the marketer (based on the business results yielded from each algorithm).
In one AI application at Dynamic Yield, rather than automatically applying what the algorithm should recommend, our Predictive Targeting solution was designed to “detect” personalization opportunities, i.e., data-backed suggestions that teams could “click to apply” for additional projected revenue gains.
By 2017, more and more brands began to see AI not as the sole decision-maker, but as a vital tool for augmenting the decision-making process, gaining trust from marketers.
And ironically, as AI started to expand into many areas of everyday life in the form of personal voice assistants, smart home devices, web search answers, self-driving vehicles, as well as greater content and product recommendations, a shift in mindset occurred yet again.
Widely adopted across industries, AI was suddenly no longer a novelty, increasingly considered a must-have from marketers who now expected smarter decision-making.
And as machine learning technology improved – with the rise of deep learning-based recommendations that made AI even more intelligent – marketers came to trust (and adopt) AI at even greater rates.
Equally important, the deployment of cloud infrastructure made AI much more affordable and scalable. This confluence of factors led some brands to put such faith in AI that they treated the technology as a black box, suitable for making decisions with little human involvement.
Today, these advanced algorithms indeed show great results, and as we hoped, are enabling a quantum leap in the quality of experiences they enable.
However, many marketers require that AI still be coupled with mechanics to see the business value, for without the ability to understand the value it yields, there can be no full acceptance, and without acceptance, further improvements to the algorithms cannot be made to yield higher performance.
Looking to the future, we anticipate a more hybrid approach, with greater investments being made in algorithms that take more responsibility in decision-making while also providing marketers with a greater level of control.
Brands should, therefore, evaluate their AI-based tools according to their distinct needs and preferences, whether that means trusting the algorithms entirely, A/B testing every step of the way, or only applying machine learning if fully educated how it works.
Whatever the case may be, I expect AI will continue to play a major role in how marketers refine the customer experience and generate meaningful results.
