Do you know what your customers are saying about you? Don’t turn a blind eye—monitoring social media needs to be an essential part of your online strategy. Not only is it important to know the good and not-so-good things people might be saying about your brand, but by listening to these conversations, you also have a great opportunity to participate and engage with your customers.
Yet the effort required to manually keep up with all the social media activity across the globe makes it an impossible task for any business. According to a 2018 Brandwatch report on social media usage, every hour there are 240 million likes generated and 14 million photos uploaded—and that’s on Facebook alone.
The solution? Only by harnessing the power of artificial intelligence (AI) can you hope to know at any point what your global customers are saying about you and your products. Let’s take a closer look at how it’s done.
Finding out what customers think
One area where AI is being used in social media monitoring is sentiment analysis. The process, also known as opinion mining or social listening, seeks to derive subjective information from a piece of content—for instance, whether the views expressed in a tweet are positive, negative or neutral.
Sentiment analysis has been generating a lot of attention in recent years, mainly due to its role in the campaign strategies for many high-profile elections. With millions of tweets and posts being published every hour in the run-up to polling day, campaign teams are increasingly analyzing these messages and quickly establishing whether they are positive or negative towards their candidates. Instead of knocking on doors and stopping people in the street to gauge opinions or waiting for the major pollsters to provide their latest data, campaign strategists can now call on real-time analysis to fine tune their efforts at reaching voters and winning them over.
In a similar way, a customer might say, “Just took the new Brand X SUV for a test drive—loved it!” or “Disappointed with the new Brand X SUV—expected so much more.” By filtering the messages according to the implied sentiment (good, bad or neutral), responses can be managed by the appropriate teams. In this case, the company’s PR team might respond to the positive message while the sales team can address objections and concerns of the unimpressed customer and open a line of communication that might eventually lead to an otherwise unlikely sale.
Not only does sentiment analysis look at the meaning behind what might be millions of social media posts, but it processes each message in a consistent way against the same set of rules, without the inevitable bias that is introduced when multiple humans take on this type of task.
Sentiment analysis becomes more complex when it’s applied to multiple markets. Users in some countries are more likely than others to show their emotions openly, and translation alone may not be enough to process slang and local humor. Although machine translation has been shown to be sufficiently accurate at detecting sentiment, an effective sentiment analysis tool is also able to pick up on the wider context behind a message. It should identify sarcasm (“Train delayed by an hour AGAIN! I just love Eastern Rail.”) and irony (“My husband just bought me this blouse from Store X; it would have been a perfect gift—for my mother.”) as well as hatred and abuse. The ability to identify and interpret emojis (including their Unicode characters) is essential here.
There are plenty of sentiment analysis tools out there that can help you do this, as this Hootsuite blog post describes. In addition to Hootsuite’s own products, competitors such as Social Mention, Talkwalker and Meaning Cloud use automated tools to help derive relevant subjective information from the mass of social data.
Analyzing non-verbal content
An increasing amount of social media content is made up of images or videos. Your customers are likely to post a picture, a funny GIF or a video along with their post about your brand in the hope of gaining more interest and engagement with their audience. How can you monitor the millions of pieces of non-verbal content being posted every day, especially on image-dominated platforms such as Instagram and Snapchat, and analyze these posts to know whether the images and clips are being used to praise or criticize you?
AI to the rescue again: it is being used to review images and establish if the people featured in them are happy or sad, angry or excited. In 2018, Instagram announced that they had adopted the use of machine learning technology to detect bullying in photos.
The ability to analyze the sentiment behind images is already out there, and it should help form a more comprehensive monitoring approach. By piecing together an analysis of images along with their hashtags, AI engines can provide instant and extremely accurate analyses of millions of posts around the world, allowing companies to get the whole picture and communicate with their customers appropriately.
It’s challenging enough to be aware of conversations about your brand taking place around the world, but you also need to know about them quickly enough to respond in the best way. Social media storms can start from a single tweet and spread across the globe before you even know about them, as a PR director found out the hard way when a misguided attempt at irony was widely interpreted as a racist comment.
You need to receive instant alerts of any unusual patterns of social media activity relating to your company, products and brands. And it’s not enough to get alerts; you must also have the right resources on hand to handle an online crisis at whatever time of day it starts.
If you have customers in multiple international markets, investing time in developing an international social media monitoring capability is no longer an option. Only by knowing what someone is saying about you, in whatever language they’re using, can you act quickly and decisively to protect your brand and your business.