There are constant discussions happening about your brand online, 24/7. On social media, in forums, news articles, online reviews, and more.
Some of this information contains powerful insights, some may be completely benign, while some is potentially harmful, and needs to be acted on right away to protect your brand image. But how do you keep up with all this data without employing whole teams to monitor the entirety of the internet?
Real-time sentiment analysis is an AI-powered solution to track mentions of your brand and products, wherever they may appear, and automatically analyze them with almost no human input needed.
Firstly, what is sentiment analysis?
Sentiment analysis is a text analysis tool that uses machine learning with natural language processing (NLP) to automatically read and classify text as positive, negative, neutral, and everywhere in between. It can read all manner of text (online and elsewhere) for opinion and emotion – to understand the thoughts and feelings of the writer.
See the example below from a pre-trained sentiment analyzer, which easily classifies a customer comment as negative with near 100% confidence.
Sentiment analysis can be put to work on hundreds of pages and thousands of individual opinions in just seconds, and constantly monitor Twitter, Facebook, emails, customer service tickets, etc., 24/7 and in real time.
Real-time sentiment gives you a window into what your customers and the public at large are expressing about your brand “right now” for targeted, minute-by-minute analysis, and to follow brand sentiment over time.
Some of the benefits of performing real-time sentiment analysis:
Target your analysis to follow marketing campaigns right as they launch and get a solid idea of how your messaging is working with current and potentially new customers. Find out which demographics respond most positively or negatively. Follow sentiment as it rises or falls, and compare current campaigns against previous ones.
In 2017, United Airlines forcibly removed a passenger from an overbooked flight. Other passengers posted videos of the incident to Facebook, one of which had been viewed 6.8 million times just 24 hours later. After United’s CEO responded to the incident as "reaccommoda[ting] these customers," Twitter exploded in outrage and public shaming of United.
Negative comments on social media can travel around the world in just minutes. Real-time sentiment analysis of Twitter data, for example, will allow you to put out fires from negative comments before they grow out of control, or use positive comments to your advantage. Oftentimes it’s helpful just to let them know you’re listening:
You can similarly follow feedback on new products right as they’re released. Influencers (and regular social media users) are eager to be the first commenters upon the release of new products or updates. Follow social media and online reviews to tweak products or beta releases right after release, or stimulate conversations with your customers, so they always know they’re important to you.
You can even use social media sentiment analysis for market research to find out what’s missing from the market or for competitive research to exploit the shortcomings of your competition and create new products.
Follow the real-time sentiment of any business as it rises and falls to get up-to-the-minute information on stock price changes. If a new product release is met with enthusiasm across the board, you can expect the stock to rise. While a social media PR crisis can bring even industry giants to their knees.
There are two options when it comes to performing sentiment analysis: build a model or invest in a SaaS tool.
Building a model can produce exceptional results, but it is time-consuming and costly.
SaaS tools, on the other hand, are generally ready to put into use right away, much less expensive, and you can still train custom models to the specific language, needs, and criteria of your organization.
MonkeyLearn’s powerful SaaS platform offers immediate access to sentiment analysis tools and other text analytics techniques, like the keyword extractor, survey feedback classifier, intent and email classifier, and many many more.
And with MonkeyLearn Studio, you can analyze and visualize your results in real time.
Let’s take a look at how easy it can be to perform real-time sentiment analysis.
First, decide what you want to achieve. Do you want to compare sentiment toward your brand against that of your competition? Do you want to regularly mine Twitter or perform social listening to extract brand mentions and follow your brand sentiment from minute to minute?
Maybe you need to automatically analyze email exchanges and customer support tickets to get an idea of how well your customer service is working. The use cases for real-time sentiment analysis are practically endless when you have the right tools in place.
There are a number of ways to get the data you need, from simply cutting and pasting, to using APIs. Below are some of the most common and easiest to use.
Tools like Zapier easily integrate with MonkeyLearn to pull brand mentions from Twitter or other outlets of your choice.
Website, social media, and email data often have quite a bit of “noise.” This can be repetitive text, banner ads, non-text symbols and emojis, email signatures, etc. You need to first remove this unnecessary data, or it will skew your results.
You can run spell check or scan documents for URLs and symbols, but you’re much better off automating this process – especially for accurate real-time analysis – because time is of the essence, and manual data cleaning will create an information bottleneck.
MonkeyLearn offers several models to make data cleaning quick and easy. The boilerplate extractor extracts only the text you want from HTML, removing unneeded clutter, like templates, navigation bars, ads, etc.
The email cleaner automatically removes email signatures, legal notices, and previous replies to give you only the most recent message in the chain:
And the opinion units extractor breaks up sentences or entire pages of text into individual sentiments or thoughts called “opinion units”:
It can break down hundreds of pages and thousands of opinion units automatically to prep your data for analysis.
MonkeyLearn Studio is an all-in-one real-time sentiment analysis and visualization tool. After a simple set-up, you just upload your data and visualize the results for powerful insights.
MonkeyLearn Studio allows you to chain together a number of text analysis techniques, like keyword extraction, aspect classification, intent classification, and more, along with your real-time sentiment analysis, for super fine-grained results.
If you want to learn how to build a custom sentiment analysis model to your specific criteria (and then use it with MonkeyLearn Studio), take a look at this tutorial. You can do it in just a few steps.
Once you’re ready for MonkeyLearn Studio, you can choose an existing template or create your own:
You can upload cleaned text from a CSV or Excel file, connect to integrations with Zendesk, SurveyMonkey, etc., or use simple, low-code APIs to extract directly from social media, websites, email, and more.
As you can see below, the model automatically tags the statement for Sentiment, Category, _and _Intent, all working simultaneously.
MonkeyLearn Studio’s deep learning models are able to chain together a number of text analysis techniques in a seamless process, so you just set it up and let it do the work for you. Once your real-time sentiment analyzer is trained to your criteria, it can perform analysis 24/7, with limitless accuracy.
Take a look at the MonkeyLearn Studio dashboard below. In this case we ran aspect-based sentiment analysis on customer reviews of Zoom. Each opinion unit is categorized by “aspect” or category: _Usability, Support, Reliability, _etc., then each category is run through sentiment analysis to show opinion from positive to negative.
You can see how individual reviews have been pulled by date and time for real-time analysis, and to follow categories and sentiments as they change over time.
Another analysis for “intent,” shows the reason for the comment. This is more often used to analyze emails and customer service data. In this case, as this is an analysis of customer reviews, most are simply marked as “opinion.”
The results are in! With sentiment analysis and MonkeyLearn Studio, you can be confident you’re making real-time, data-driven decisions.
Imagine you release a new product. You can perform real-time aspect-based sentiment analysis on Twitter mentions of your product, for example, to find out what aspect your customers are responding to most favorably or unfavorably.
Play around with the public dashboard to see how it works: search by date, sentiment, category, etc. With MonkeyLearn Studio you can perform new analyses and add or remove data directly in the dashboard. No more uploading and downloading data between applications – it’s all right there.
Real-time sentiment analysis can provide powerful insights about your company and your products constantly, consistently, and whenever you may need them. Find the overall sentiment toward your brand at any given moment, target analyze marketing campaigns, or pit yourself against the competition.
MonkeyLearn Studio offers an all-in-one real-time sentiment analyzer and business intelligence visualization tool to get the most from your data, whether it's online, in emails, customer support tickets, surveys, and more.
September 17th, 2020