Machine learning is a topic most marketing leaders have heard discussed in the boardroom from their CTO or CIO. The ability to understand trends in data not from asking questions, but from “magical algorithms” that continuously help big data managers learn more about the customer and their next moves. In today’s world, where more and more data is flooding into companies from multiple sources, the ability to learn what your customers want next is critical. However, not all customer analytics is the same, especially when it comes to superior customer marketing.
First off, machine learning is a discipline within science dealing with the development of algorithms that can learn from data. These algorithms take data inputs and make predictions or decisions based on the insights rather than following an explicit set of instructions. It is essentially a sub-area of study of both computer science and statistics. The science of machine learning is applied to a wide range of computing tasks such as search engines, spam filtering, and in the world of marketers, data mining.
Machine Learning is used daily to certain degrees with the marketing of products and services to customers. In the most basic of applications, it is used to decide how and when to present ads to a customer on several online platforms. A customer who searches for a specific set of terms over time in Google becomes qualified to receive a particular message you pay to have presented to them. Machine learning applied to email marketing is another application. One that is often overlooked.
Email marketing is often performed with a customer’s email and some very broad definitions derived from the data connected to that email address. Often times it is the basic elements of good customer data collection like gender, age, location, etc. These characteristics are used to create a segment and the recipients receive the same product suggestions or a degree of personalized content accordingly.
Most marketers consider a good campaign to be one where they send an e-mail to a group based on their age, location and gender and target the content based on these parameters. A good, yet highly stereotypical, example would be an e-mail sent in the Spring for colourful floral umbrellas to women aged 20-30 and living in Seattle. The problem here is that the e-mail content is based on a pull of basic data and is missing significant information required for it to be more than an assumption. Machine learning can help make it more engaging.
Imagine if you were tasked with compiling tons of data from multiple sources like Web site clicks, online behaviours, and in-store transactions. The challenge of weeding through this data and knowing what to do with it can be staggering. It is immensely valuable, but few marketers have the skills to correlate what it all means. Enter the marketer’s best friend, machine learning.
Here at Coherent Path, machine learning happens within the unique hyperbolic geometry we create for our clients. Data from all of a customer’s touch points come together into a single space where our advanced machine learning algorithms take over. Our team of nuclear scientists and math prodigies spend countless hours applying advanced mathematical theorems to our platform to not only pull the data together, but use machine learning and other tools to identify common customer attributes and pathways for more probable (and profitable) engagement.
This advanced method for analyzing data means that each customer can be isolated and a journey created for them that is unique to the experience they statistically have a higher chance of engaging with. Looking back at our e-mail example, the resulting messaging could be very different. Instead of blanketing an audience with a series of colourful floral umbrellas based on a possible affinity to the items, machine learning would only send such an e-mail to recipients whose entire experience to date would qualify them as being an ideal candidate for having an interest in such a product. However, it doesn’t stop there.
Since Coherent Path’s hyperbolic geometry creates a unique product journey for each customer, the offer would only be presented if it were something that would engage him or her on his or her designed pathway leading to lifetime loyalty. This means that our machine learning isn’t simply presenting a product based on likelihood (derived from their historical or more recent behaviour) but the offer or product that is aligned to the journey our math has determined to be the most relevant and likely to drive lifetime engagement, not just another purchase.
This means that a 23-year old female in Seattle may not receive an e-mail about colourful floral umbrellas. Instead, she may receive an e-mail about a new series of designer scarves, jeans, or another item that is better aligned to the journey she is on through a retailer’s product catalogue. It doesn’t mean she won’t ever be told about colourful umbrellas.
Machine learning simply presents the most engaging recommended items based on past purchases, her likelihood to find relevance with the product, and the goal of moving her towards lifetime engagement with the retailer. Colourful floral umbrellas may be something she learns about in a few week’s time, when machine learning feels she is better suited to learn about them.
Are you using machine learning and have a pathway defined for your customer? If not, why not start?