The ABC's of Artificial Intelligence, Part Two
We tend to think of artificial intelligence (AI) in terms of tangible technologies, like robots or even Watson. But most AI-powered technology actually exists in a much more abstract form, in the form of algorithms. And that technology is incredibly important for marketers who wish to stay at the forefront of the industry.
Last month we looked at several key AI concepts for marketers, and here we dive into many more. These technologies have already changed the face of modern marketing and will undoubtedly drive its evolution in the coming years.
Internet of Things (IoT)
The Internet of Things (IoT) refers to the connection of everyday devices to the internet. It has already had momentous impact on the global supply chain because it allows machines and other devices to “talk” to each other. For marketers, IoT promises to provide much more context for customers’ habits, needs, desires and product usage, along with purchase intent.
This year, the number of IoT devices surpassed 7 billion, according to IoT Analytics. This is a boon to marketers, who stand to gain incredible insight into their customers at every stage of the buyer’s journey. But it also presents a significant challenge: all these connected devices already produce mountains and mountains of data, and as the number of devices grows, so will the quantity of data. Savvy marketers will enlist AI-powered technology to sift through everything and draw meaningful conclusions.
In the past, robots and computers could be used to automate tasks that fit within a specific set of rules. The device would be programmed to operate within the rules--nothing more, nothing less. But machine learning is exactly what it sounds like: machines can learn from the data they analyze, making predictions and forecasts using many more data points than humans could on their own. For example, while humans are able to analyze data from one source, say the website, machine learning means that you can glean insights from patterns in usage of the website, social media, email and social media, along with customers' responses to previous campaigns.
Machine learning has multiple promising applications for marketers. It's the ideal technology for sophisticated customer segmentation, allowing you to identify small groups of customers who share preferences and tend to have similar behavior patterns. Machine learning can also be used for predicting customer churn, which allows marketers to proactively reach high-risk customers and decrease the likelihood of their attrition.
Natural Language Processing
Machines were once only able to understand code. But that has changed thanks to Natural Language Processing (NLP), which refers to a machine's ability to understand human language through machine learning and AI. NLP has evolved in recent years, and in some cases this technology can actually understand the emotion or connotation behind the human language. NLP is often paired with Natural Language Generation (NLG), where the machine processes the human language and then generates a human-language response. Voice-recognition systems like Amazon's Alexa and Apple's Siri work this way.
Many marketers already rely on some form of NLP for sentiment analysis. But it's also promising for personalization. Most organizations engage an audience that is geographically--and therefore linguistically--diverse. Those language differences may be subtle (such as regional variations in American English), more obvious (such as differences between American English and the Queen's English), or even involve different languages altogether. NLP can help you personalize communications based on these differences, to deliver messages that are more likely to engage each recipient.
A neural network is essentially designed to mimic the human brain. The network is comprised of a series of layers. Each layer includes many nodes, or individual processing units, which are connected to other nodes in that layer along with nodes in the layers above and below. The lowest layer of the network takes in data and relays it to the next one. While the purpose of a neural network is to simulate a brain, the network will not function exactly as a human mind would. They are best used for pattern recognition, forecasting and trend prediction; and even generalization.
Microsoft used BrainMaker, a neural network software system, to optimize its direct mailing campaigns. The software identified the most important variables in the success of direct mail campaigns, so that future campaigns could be optimized. According to a company spokesperson, using BrainMaker increased the response rate from about 4.9% to 8.2%, a considerable improvement. This use of neural networks for campaign optimization holds great promise for marketers.
A branch of advanced analytics, predictive analytics are used to make forecasts about future events using techniques like data mining, statistical modeling and machine learning. To maximize the benefits of predictive analytics, you must start with an extensive set of historical data. Rigorous data collection in marketing is relatively new--think how the role of analytics has changed in the last decade alone--so now predictive analytics has only recently emerged as a widely applicable technology in the last few years.
Amazon and other online retailers use predictive analytics extensively to model likely customer behavior. It can also be used for predictive lead scoring, so that the marketing and sales team can tailor campaigns and offerings to leads and customers who are most likely to convert. Meanwhile, applying predictive analytics to data about your existing high-value customers can help you develop strategies to reach more new customers just like them.
Be sure to check out The ABCs of Artificial Intelligence, Part One for more information. Have a question about how AI can jumpstart your inbound marketing? Contact us.