The first and foremost thing any marketer needs to do is get a handle on their data. Start by answering these questions: What data does my organization currently gather? What data could my organization collect?
Once you have determined the data you have access to, you can begin to look at how Artificial Intelligence (AI) and Machine Learning (ML) can generate value for your organization. The algorithms used in AI and ML need a large amount of data and a relatively continuous stream of new data – usually the more, the better.
Data can be unstructured, but the cleaner and better structured the data, the easier it will be to model, and the less work there will be to transform your data into meaningful insights. Depending on the size of your organization, the most valuable data could be found outside of marketing and sales.
Think of it this way – you are not looking to build a kitchen, but learning how to cook. Unless you have the budget to hire a team of talented data architects, think about the goals of your organization or department and not the structure or particular algorithms used to translate your data into useful content. You don’t need to evaluate deep learning; you need to identify the combination of marketing characteristics, customer and prospect behaviors, and audience segments that will lead to a one percent product revenue lift or reduction in churn.
Understanding the quality of your ingredients (data) is much more important than the quality or cost of your stove and cooking utensils (software).
Software tools make an analysis of good data easier and there are several companies seeking to democratize AI/ML, giving front-line analysts with less formal training the tools to run their own analysis. If this is your path, then the ingredients (data) are even more important.
At this stage, AI/ML software is very specific to the problem domain, which is the problem you are trying to solve with AI. For example, this could be a better use of predictions, improving customer outcomes, or reducing expenses and errors with perception (i.e., image recognition). Even generalized marketing AI software is usually an outgrowth of algorithms that focus on a specific problem. Make sure you understand the problem the software was designed to tackle and start with that specific problem you are looking to solve for your business or clients.
Where Are We in the Evolution Of AI and ML?
We currently have narrow, deep software available for many problem domains. AlphaGo can play Go, but will not help marketers understand emotional reactions to a commercial. There is existing software that analyzes human reactions, measures heart rate, and eye movement, but again, the algorithms have been trained to do just that one task.
Today, we are at peak hype- Las Vegas-style magic is everywhere. From image recognition to intelligent voice and chatbots, there is actually very little magic and lots of old-fashioned gears and humans making the magic happen.
Putting AI/ML to Work
There is no question the use of AI/ML can enrich your data. There are thousands of public data sources and research data you can use to augment and enrich your existing data sets. In my experience, we’re working with a few different companies that help enrich audience data with offline data to improve campaign performance. These AI/ML platforms get a bit better each time we use them as they begin to learn, from our human input, what it is we are trying to achieve with the data. Our goal with data is to help our clients show advertisers ROI on marketing dollars. The better we get at connecting the dots across disparate data sources, the easier it is to demonstrate campaign performance.
Gaining experience in this realm is as simple as trying out an AI chatbot on your website. There are many available and it is an easy way to get acquainted with the technology.
- Look to set some rules and information sources (such as an FAQ) that will make the bot more effective for your customers.
- Set up processes for an online chat for your teams, and monitor and measure the performance and tweak the responses as necessary.
AI/ML is already a part of our everyday lives, just look at ridesharing, social media platforms, and online shopping recommendations. Successfully adopting these technologies into our marketing strategies is a natural evolution, but must be treated with care for it to produce meaningful results that will contribute to your bottom line.