The marketing industry has been one of the first areas where digitization and automation have taken place. As internet usage exploded and the number of digital advertising channels has grown to large proportions, it became obvious that marketers needed complex and effective tools to help them manage things like media buying, user analytics, consumer journeys, and marketing campaign management.
Over the years, increased data availability and depth of historical records, paired with a slew of both startups and initiatives by more established players, has brought in a number of potential game changing products to help marketers automate their activities. More recently Machine Learning and Artificial Intelligence have provided a significant improvement to marketing technology. In this article, we will take a look at some of these innovations and discuss their benefits.
What is AI?
Before going into more specific details, it may be useful to go through a quick refresher on what AI is.
Artificial Intelligence is a field of study which focuses on creating intelligent machines capable of solving problems and making decisions typically addressed by humans. Traditional semantic algorithms allow systems to produce an output based on a known input. With Machine Learning, we can teach computers to produce an output without knowing exactly what the input will be. Systems paired with AI use this technology to make automated decisions, while also continuously learning and adapting to a different input.
For example, recognizing spam emails can be difficult if that logic is based on a fixed list of keywords or phrases. Machine Learning, on the other hand, constantly learns new and different combinations of words that could be a part of a spammy email. It is then, up to these intelligent systems to make a decision and label the email according to the predicted parameter.
Media Buying Automation
Most marketing spend is still going towards the two most prominent advertising platforms, Google and Facebook. For some time, automated programmatic buying has been the staple approach to display advertising. In a typical media buying strategy, advertisers set up most of the parameters in a Demand Side Processing (DSP) system, which then bids automatically on ad placements through an ad exchange. This is done based on provided preferences which must be manually managed based on constantly changing advertising strategy.
An AI augmented system has a significant advantage over this method as it allows for dynamic modification of provided parameters based on their performance and other signals. Let’s say a sports bottle retailer is running a media campaign. Their strategy includes a list of online publishers and a list of keywords that will be used to target their buying efforts. They may have also set up certain performance targets to make sure the underperforming ads do not waste their advertising spend. What happens when a relatively unknown publication releases an article on proper hydration regiment, which goes viral, but is not in the list of publishers being targeted by the system. Here is where AI can add incremental value. In addition to being able to dynamically change the existing parameters, it can also monitor multiple available channels and adjust targeting based on reader engagement.
- Albert.ai – is an automated marketing platform. It uses machine learning models and AI to dynamically adjust parameters including buying, scheduling, and targeting.
- DeepIntent – is a contextual marketing automation platform. It uses consumer journeys and sentiment analysis to place bids only on ads which would be relevant to specific users.
Consumers prefer experiences that are tailored specifically to them. There are many tools that are currently being utilized to deliver personalized content. Starting with product recommendations and related news suggestions, to completely tailored experiences, like media campaigns. Many e-commerce websites report over 20% improvement in sales when adding personalization features.
Typical personalization involves identifying a unique customer, even if anonymously, and using algorithms to decide what would be relevant to them. The biggest hurdle in dynamically changing the customer’s online experience is accounting for a potentially infinite number of possible routes a user may take when navigating through a website.
Machine learning can use predictive models to determine which content or products would be most relevant without knowing all possible combinations of user actions. This means that dynamic pages can really be dynamic without using canned variations, and personalized communications can relay information that is relevant to the current step in the consumer’s journey. Personalization can even go beyond website and email content. An experimental prototype ran by JWT Canada was able to dynamically change digital ads based on the audience parameters gathered using the same AI-driven system.
The most known vendors in this space are Salesforce with their Commerce Cloud Einstein, which drives product recommendations and personalization, as well as Adobe Target, which also relies on artificial intelligence to decide which content to display to its users and when. There are also a few smaller players, whose platforms work with all major content and e-commerce systems.
- Evergage – an omnichannel personalization platform. It uses machine learning to deliver consumer experiences more likely to result in increased engagement.
- Sentient – an AI driven personalization engine relying on deep learning models to deliver relevant content.
This particular area of marketing automation is relatively unknown and has only started to gain traction recently thanks to advancements in Natural Language Generation, which allows for dynamic content creation that is indistinguishable from human created one. Traditional content authoring involves curate and in-house content which marketers need to match with interests of their customers. Existing platforms like Scoop.it and Curata allow content marketers to get deep insight analytics and recommendations, but still require manual validation and an editorial process before it can be syndicated for customers.
With the introduction of AI, content creation can be done by a machine.
- AI Writer – while still in beta mode, this tool promises to write articles automatically based on existing content, all without plagiarising original authors.
- Atomic AI – a content creation platform which provides real-time assistance to editors and copywriters on what to write, where and when to publish.
- Articoolo – another beta product with a serious promise to create a fully written content automatically.
While not a traditional marketing field, today’s customer experience interactions often include additional sales opportunities, as well as efforts to increase brand and product loyalty. Consequentially, many activities performed by customer service are marketing in their nature. A recent proliferation of AI-based Chatbots has allowed many companies to take advantage of technology that facilitates enhanced consumer communications, increasing user engagement and creating new sales opportunities.
Aside from customer service, interactive customer tools can also increase engagement. Shopping assistants are a good example of such a solution. North Face has recently launched an IBM Watson driven product finder. It uses AI to bring the in-store experience to online customers.
Another area where AI is helping to increase conversions and drive more revenue is churn prevention. While also not a direct marketing area, automated churn detection can take advantage of machine learning to develop a more accurate prediction.
- Goedle.io – AI based churn prediction engine, which can accurately analyze and predict when a customer is more likely to cancel.
- Deepsense.ai – AI data science agency focusing on building churn prevention solutions.
Analytics & Insights
One of the biggest areas where AI has been able to deliver significant improvements over traditional technology is in data analytics and marketing intelligence. Marketing Intelligence platforms are sets of tools that consolidate various sources of analytic data and use dashboards to display insights based on the combined information. For the most part, this intelligence information help marketers make educated decisions on how to properly balance budget, channels, and media. Some, more advanced platforms also allow campaign execution using this data.
The exponential growth of available data has created a giant sea of raw materials that can be mined in order to produce meaningful insights. This resulted in the increase in prescriptive information, which unlike descriptive, can also provide a recommendation or even an automated action. A number of marketing platforms now feature automation and activation components that can make decisions using machine learning models.
- Datorama – a specialized marketing intelligence and analytics platform. It also includes an activation component which allows it to drive marketing campaigns, something Qlik and Tableau are not able to do.
The current marketing technology marketplace has grown tremendously over the last few years. It may seem overwhelming with such a large array of vendors aiming to deliver significant improvements over existing marketing practices using AI technology. One way to prevent potential disappointment is to run a pilot or proof of concept project before committing to longer subscriptions. It is also advisable to go through a full RFP process in order to identify specific business needs and map them to features available from potential candidates.