Early this decade in Minneapolis, an angry man walked into a Target store and demanded to talk to the manager. The man was furious that Target had sent a mailer with coupons for baby clothes and cribs to his daughter who was still in school. “Are you trying to encourage her to get pregnant?” he furiously asked the manager who had no clue about what he was talking about. The manager found that the mailer was indeed addressed to the man’s daughter and it contained marketing materials for maternity clothing and furniture. The manager apologized. He later called the man over the phone to apologize again. This time the father sounded embarrassed. He had talked to his daughter and she was due to have her baby in August. The family was unaware of it. The man apologized to the Target manager on the call.
How did Target know about a teenager’s pregnancy that even her family didn’t know about? The answer lay in Target’s ability to sift through customer data with the help of machine learning. Target assigns every customer a Guest ID number and stores a history of all their purchases and demographic information. By using historical buying data of all the ladies who had signed up for Target baby registries, Target could identify 25 products that, when analyzed together, helped assign each of them a “pregnancy prediction” score. More importantly, Target could also predict their due date to within a narrow band of time to send mailers and coupons aimed at each stage of pregnancy!
Humongous data volumes make the use of machine learning inevitable
Machine learning is described as the ability of systems to learn or improve automatically with experience. Or, the system improves its ability to do its assigned task without any new programming. The use of machine learning in marketing becomes inevitable because of two factors: Data volumes and time. According to EMC , by 2020, the volume of data in the digital universe will grow to 40,000 exabytes or 40 trillion gigabytes. The human memory can hold only the equivalent of one gigabyte of data. Marketers often have to laboriously dig through very high volumes of data that they risk missing the big picture. Machine learning enables marketers to sift through large data volumes productively and save their time for more creative and critical work.
Scale and speed: That’s what machine learning help content marketers achieve
With the increasing demand for personalization, marketers must deploy, reuse, and remix content in new ways to engage individual audiences and drive conversion. Machine learning can be used to include the most relevant keywords in content. It can combine relevant copy, images, and video for a target audience and smartly suggest what to offer next after your audience has consumed a piece of content. Optimizing content for different channels is something that machine learning can also do. For example: shortening visual content or pruning copy for different channels. As the machines learn with experience, they can tell you which content is working and help you amplify the type of content that works at a faster clip, so you can adapt content almost instantaneously. In essence, machine learning enables marketers to achieve speed and scale to maintain the required content velocity without creating a lot of content. For B2B marketers, the development of knowledge graph technology helps them connect with consumers by providing recommendations based on insights into their interactions, emotions, and interests.
How will the future of content marketing evolve with machine learning?
In addition to specifying the subject matter of the content that you should create, machine learning might also specify the right medium and the tone for telling brand stories. By building basic machine learning components into interactive content, you can personalize content even more. Using a combination of in-house data and user feedback you can customize real-time content on a one-to-one basis for an immersive and personalized experience. Eventually, machine learning will help marketers essentially test the innumerable directions that users might take using content in their journey through multiple channels. This will help marketers ensure that each step is in sync with the preceding as well as the succeeding one. In short, it will save immense human effort, time, and minimize the potential of errors.
No. Machines won’t replace humans
The subject of machine learning always draws ethical questions. Externally, it can raise questions about privacy as the Target example shows. Internally, it raises the question whether machine learning will eliminate marketing jobs. Machine learning does not eliminate human effort. First of all, it’s difficult or nearly impossible to teach machines creativity – at least as of now. Companies will require data scientists to build the technology and infrastructure to exploit data. More importantly, extraction of knowledge from data will require human input. Last but not least, it's worth remembering that only humans can produce great content that evokes emotions in other humans. It still takes talented marketers to encourage users to interact and create value from the insights derived from data. Where machine learning makes a difference is in improving marketing efficiency and outcomes.