Engage, Discover, Decide – The three buzzwords of consumer interaction
Today, the main drivers for new service adoption and customer engagement are speed, convenience, and interaction with retailers. Most companies have currently access to an incredible amount of consumer data collected from a wide range of sources, from online browsing, day-to-day interaction with customer service and brand social media channels to actual purchases both in online or physical stores. Unfortunately, companies often lack the resources to leverage this huge amount of information in order to spot behavioural patterns and turn them into actionable insights.
With the rise of the global economy and the contribution of the internet, more and more companies are vying for the attention of your customers. This is why it is vital, now more than ever, to be able to understand how to capture and, more importantly, retain your consumer base interest and business. Fortunately, the ubiquity of social media in our lives provides what is essentially every marketing team's dream: a treasure trove of personal information you can draw upon in order to glean behaviours, interests, needs, and preferences.
With Facebook's announcement of reaching the milestone of over 2 billion monthly users (source: CNN) for Q3 of 2017, averaging hundreds of thousands of posts and comments a minute, manually scanning your consumer base posts for clues and identifying relevant population clusters would be a superhuman task. This is where cognitive computing comes in.
Cognitive APIs help researchers analyse the content of an unstructured text (like documents, emails, posts or tweets) and contextualise it, based on advanced semantic rules. It is also possible to construe the tone of the writing and create a psychological profile of the poster, based on their writing style.
Let's look at a few possible applications.
Brand reputation – How does the public feel about your company?
Applying sentiment analysis to brand reputation has been a sort of pioneer in using text mining to understand customer interactions. What looks like a simple problem – scraping public posts containing the name of your company and products, and searching for positive or negative wording – can prove to be very challenging and nuanced.
First of all, it is important to identify the aspects to focus on: public opinion of your company may be very positive, but your newest product might still be underperforming. How does the impression your customers have of your company's core beliefs influence what they think about the company's products? How did the most recent announcement from your CEO impact which search terms are connected to your brand? How does world news affect your sales? Apparently unrelated topics all concur with your brand image.
Secondly, it is important to make sure the cognitive system you are using can contextualise and interpret data. The difference in wording between a genuinely happy customer's tweet and a frustrated one's could be as subtle as the addition of an emoji. For example, "The "awesome" new iPhone X face recognition does not work in strong lighting. Just brilliant! :/" and "The new iPhone's facial recognition is awesome! Brilliant product, Apple!!!" convey very different meaning, which is clear to a human analyst but might trick a simple automated search for positive or negative keywords. An advanced sentiment analysis system is going to be able to infer the actual tone of a post and help you interpret the true meaning of your customers' posts.