Can artificial intelligence make your company’s marketing more “human”?
The answer is, Yes! But there are a few key things to consider before implementing AI to get it right. As part of an ongoing series, we will be defining what AI really is, what to consider before implementing AI, and how to use AI powered insights. Let’s jump in!
Part 1. Artificial Intelligence definition and uses for Modern Marketers
There is a growing buzz around AI. It’s commonly mentioned among trends and next big things, but that’s typically where the conversations stop. If you ask most experts in digital media to define their AI strategy and you will most likely get a puzzled look or an explanation of what they think the possibilities might be.
First, let’s dispel the myth. AI is not a magic bullet that solves all your business challenges, at least not in 2018. AI is nothing more than an algorithm designed to mimic human problem solving, and for marketing applications, a more accurate term to use would be a subset of AI, Machine Learning combined mixed with Data Mining.
In The Key Definitions Of Artificial Intelligence (AI) That Explain Its Importance, Bernard Marr does a good job framing AI, “(AI) uses human reasoning as a guide to provide better services or create better products”. Machine Learning, which according to SAS, is “a method of data analysis that automates analytical model building” uses “artificial intelligence based on the idea that systems can learn from data, identify patterns and make decisions with minimal human intervention.” Simple put, AI applies human reasoning and problem solving to raw data to come up with valuable insights.
The ability to provide insights by transforming raw information is a trend that modern marketers must take advantage of if they are going to stay relevant in a data-driven economy. The challenge is finding out how to use the data and making sure the data source is sufficient.
Here are a couple of good ways marketers can start using AI.
- Look-alike Modeling
Look-alike uses your existing client base as the data source. AI analyzes the transaction history and details of each client to build predictive models to identify likely new clients and upsell opportunities.
This a great place to start using AI, but it has its limitations. First, the data used for the models depends on your existing client base and the accuracy of your record keeping. If you want to get it right, you need to evaluate your record keeping, create a standardized process across all your company’s business units, and build a culture around maintaining data integrity. There’s a human element that will always impact the data unless its collection is automated. Minimizing the human impact will ensure your AI can produce accurate models.
The second issue is that look-alike assumes that just because two clients share the same characteristics, they have the same needs, objectives, and timing isn’t considered. It’s important to understand that the data you put in determines the quality of the models and each client has unique needs.
Even with its limitations, look-alike models can be tremendously valuable. Building a comprehensive set of client profiles will let you take advantage of customized messaging opportunities designed to resonate with a specific audience and inform your advertising budget planning. Executed properly, AI-powered look-alike models can transform your marketing capabilities.
- Realtime Behavior Modeling
We live in a digital world where our access to information is constantly increasing. Consider how technology is changing how we use and collect data. Some basic examples are things like pixel tracking being used to track online data consumption, social media providing a glimpse into our mindset for all to see, and digital display networks marketplaces connecting marketers to the audience they are targeting. The sites we visit, the content we consume, and the things we interact with all leave a digital trail of data. What do we do with all this data?
Enter AI. Artificial intelligence can be used to process real-time behavior data of your target audiences and build intent based predictive models. Knowing a prospect’s intent allows marketers to design customized messages and place them in front of the right audience at the time they are most likely to buy. The limits of AI are only set by the availability and access to data, and the ethical collection and use of publicly available information.
How would behavior data transform your marketing strategy? Have you ever published an article using a #hashtag? Think of what you could do if you could create predictive models using all relevant #hashtags to your business. Now, what would happen if that same model cross-referenced additional data sources and overlaid look-alike models to create an audience list who are your target market and are likely in the market to buy? Would that have an impact on who you present your message to and how you customize your strategy?
There are challenges to using predictive models based on behavior. We run the risk of infringing on privacy and creating ads that overreach becoming too personal. Also, while using intent models it is easier to predict buying tendencies than by using look-alike, behavior-based models still rely on the data points available to the AI algorithm. The integrity of the data will determine the value of the models.
As we continue to develop new technologies, platforms and create guidelines on personal data collection and use, the applications for AI will evolve, empowering marketing organizations who understand how to tap into its potential with competitive advantages.
In Part 2. I will cover the steps marketers can take to get their organizations AI ready.