When the HAL 9000 computer took control of the ship in Stanley Kubrick’s “2001: A Space Odyssey,” the human fascination with Artificial Intelligence (AI) suddenly turned to apprehension. Today, there is no fear of AI. In fact, AI and machine learning have become invaluable tools that helps us humans do our jobs better.
For marketers, trying to understand the highly variable customer journey and ever-evolving consumer behavior has left them with far more questions than answers. Increasingly, they are turning to AI to gain valuable insights and a better understanding of their customers.
There is simply more data and information available than ever before, much of it being provided by the customers themselves in the form of online conversations, comments, reviews, blogs, social media interactions and more.
But the only way to sift through the sheer volume of this information – let alone extract any actionable insights – is with the assistance of AI. So how is this done? How, exactly, do marketers use AI to go through the vast amounts of unstructured consumer-generated data?
Getting Specific
Using AI-driven consumer insights platforms, like the one developed by Oculus360, brands are able to extract the precise insights needed for any particular marketing effort. Some examples of what may be extracted from these online conversations include:
- Customer segments and demographics, such as age, gender or personality type
- Product attributes that different consumer segments like and dislike
- Consumer perception of brands
- Specific occasions when products are used or that they are associated with
Without years of extensive research and an ongoing account planning effort, understanding particular attributes or events that consumers associate with a particular product is extremely difficult. Extracting more granular insights – which occasions consumers associate with your brand, for instance – is next to impossible. Machines, however, can automate this process and quickly provide a list of specific attributes that will greatly affect your targeting, your messaging and, eventually, your brand.
One AI-assisted project example would be to sift through online comments for trends around times of the day and/or events associated with your brand, from routine activities (e.g. waking up or going to bed) to the special occasion (e.g. weddings or birthdays), as well as product attributes where you may have competitive positioning over other brands.
How the Machines at Oculus360 Do It
It’s clear the gleaned insights would be invaluable. But, again, how do we get to this point? Here is the Oculus360 method:
1.Seed, Aggregate and Score:
We scour millions of content posts across relevant public sites to gather, score and refine data based on relevance to the insights we need to gather – product/brand, target segment, attributes, etc.
2. AI Platform Learns Category Themes: Using Oculus360’s patent-pending techniques, we then use our analytics models to process large volumes of unstructured data. This is accomplished using AI techniques such as:
- Natural Language Processing (NLP) uses AI and computational linguistics to understand natural human language, which helps us identify product occasions, attributes, and perceptions and uncover more complicated consumer aspects, like demographics, gender, personality and emotions.
- Semantic Networks provide a conceptual model that helps us understand the relationships of words and ideas to a specific topic or category (like Luxury Cars or eSports). We combine semantic networks that capture common knowledge with category specific networks that we’ve built from the bottom up using consumer vernacular. This enables us to relate marketing messages with consumer speak.
3. Machine Learning: Finally, we use machine learning to identify key consumer insights, and organize them into category themes. Once our platform understands the category, it begins to identify relevant themes made up of occasions, perceptions, attributes and customer segments. Armed with these themes and phrases, we can then track consumer perception drift and trends over time.
Compare the machine learning approach to traditional market research and market intelligence techniques like focus groups, surveys and interviews. These are expensive, time-consuming methods designed to understand only a small subset of customers whose views are extrapolated to a larger population. They are not entirely authentic dialogues, either -- participants know they are part of a study, and that can impact their responses. Furthermore, they measure at a single point in time and don’t represent the trends over time.
While marketers may have years of experience, award-winning creative talent and a complete history of case studies from which to learn, what team has the capacity to sift through millions of online consumer comments and reviews to extract structured consumer insights?
With the help of Artificial Intelligence and machine learning, marketers have timely, ongoing access to the types of insights and information that can create better, more targeted, more efficient, more effective campaigns.
Thank you, HAL.
