touching a wall
PHOTO: xandtor

Back in May of 2016, when artificial intelligence was coming into its own, I looked at how it was starting to reshape the business world. As a business enabler, artificial intelligence (AI) was still relatively immature and only the most elementary of use cases were being introduced at that time. Well, fast-forward a couple of years and here we are — still talking a lot about AI and its subcomponents of machine learning, cognitive computing and deep learning.

Related Article: How AI Is Reshaping the Business World

AI and the Brand-to-Consumer Experience

As a marketer, I think about the brand-to-consumer experience daily. Over the past couple of years, we have seen a lot of automation driven by artificial intelligence with regard to the digital experiences that consumers have when interacting with brands. Business processes such as customer service, support and even sales have been successfully automated for efficiency in certain areas. But there have been failures as well — mainly related to cognitive and voice systems — with the most extreme examples being systems that ended up being trained to convey racist or sexist sentiments. 

There’s no doubt that the road ahead may be a little bumpy, but the ride will become smoother as more employees and organizations become comfortable with a set of AI technologies that have only been in the marketplace a few years.

What have been the early success and failures for artificial intelligence? What does the road ahead look like for AI as it relates specifically to the digital experience? Let’s take a look.

Related Article: How Machine Learning Improves CX (and Increases Revenue)

Early Successes With Deep Learning

The introduction of AI has certainly led to improvements in the digital experience from both an operational and executional perspective. In the operational area, AI has automated and improved the handling of quick, repetitive backend tasks. Processes such as updating service tickets, querying technical support and even providing product enhancement recommendations are now faster and more efficient. Why? Because machine learning algorithms can perform distributed and parallel computations against increasingly large data volumes with ease. That, combined with improved automation and delivery down into channel, serves customers well.

From a cognitive perspective, we are all aware of the successes of voice assistants like Alexa, Siri, Cortana and Google Assistant. While they have had some mishaps, overall they are enabling smart home automation and the completion of simple tasks.

We have seen early successes with deep learning. Using models with multiple trees allows for more complex decisions (more than one step) to be made with regard to serving people who interact with a brand. The early results of deep learning include higher utilization rates on the part of both employees and customers, and an increase in the volume of customer interactions that can be handled in a given period of time.

Still Room for Improvement With Machine Learning

While we have seen the launches of many exciting artificial intelligence initiatives, I do believe we can still make considerable improvements in the way we use AI. These improvements have to do largely with the use of data. From a machine learning perspective, brands use data and algorithms to inform what is called next best action or offer. These product and service recommendations, while originally triggered by a business rule of some sort, are now more carefully thought out by machine learning algorithms and models.

Delivering the result of a next best action or offer, in my mind, is the best chance a brand has to show that it really “knows you.” However, there are many everyday brand interactions that can be further improved with machine learning. Let’s consider a few.

Product recommendations

Product recommendations from the biggest brands are still lacking. This is due to the limited use or complete lack of robust data profiles and, of course, machine learning. It’s time we moved beyond just using historical data to inform future product recommendations. I think we will see the concept of collaborative machine learning continue to replace the low-level collaborative filtering that is currently being used for product recommendations.

Loyalty and rewards, coupons and discounts

Machine learning and predictive analytics could play a huge role in better personalizing loyalty rewards from airlines, hotels and even retailers. Combining yield management techniques with better machine learning algorithms and customer preference data would allow brands to offer loyalty-friendly consumers better loyalty deals and rewards. From a couponing and discount perspective, compared with today’s mass-market “spray and pray” approach, using price elasticity techniques and machine learning to better personalize coupons would lead to increases in revenue and conversions for brands.

Text and search

From a cognitive perspective, both text search and voice aren’t where I expected them to be by now. As we continue to see text analytics applied to both messaging and voice classification capabilities, these capabilities will improve. As people speak on the phone less and text more, the brands that can figure out text and voice classification and routing the fastest and best will be the winners, and they will have the most success from a sales and advertising perspective.

Digital advertising

Brand advertising has struggled to keep the attention of digital consumers, but artificial intelligence could help. Using cross-device targeting based on machine learning, ad platforms that leverage machine and deep learning can deliver better personalized ads to people. The analytical marketers that have the best understanding of consumer preferences and are able to time their initiatives so that they deliver those ads at the ideal moment will move from advertising failure to success.

Related Article: Machine Learning Ushers In a New Era of Customer Experience

AI Will Only Get Better With Age

As we move into the future with artificial intelligence, I believe we will see complexity and critical thinking capabilities of AI programs continue to improve. Machine learning particularly will improve and become more pervasive, moving from predicting what you want based on past actions to actually prescribing what you will need in the future. Cognitive computing capabilities will improve as more voice and text data is collected, and as more linguistic training of these systems takes place. However, I believe the most exciting advances will be in deep learning, where programs will be able to process and move data through complex decision trees and neural networks to provide better, more detailed and more accurate answers to consumers. This will move us beyond the conditional logic that is used by most customer service systems today.

The brand-to-consumer digital experience will always need the application and understanding of the human creative process as its main input. However, the automation and accuracy that artificial intelligence ultimately will be able to provide will give brands and their employees the time and support they need to truly delight their customers.