When sentiment analysis tools started to hit the market, they were simplistic and reactive, designed to give brands straightforward feedback on what their customers were feeling in reaction to something the brand had already done.

This is the type of sentiment analysis most of us are familiar with. The feelings being analyzed are generally simple—either positive, negative, or neutral—and the analysis is usually performed on social media posts, reviews, and other forms of online content that a customer creates after having a brand interaction (keep this point in mind—we'll come back to it soon).

Enterprise-level brands looking at simple sentiment data benefit from their size—a huge number of people are going to have an opinion, for example, on the new Sonic the Hedgehog movie. It doesn't take much in the way of sentiment analysis to realize that everyone pretty much hated Sonic's new design.

We picked this example because it's easy to extract meaningful insights from even simple customer sentiment when there's a lot to gather. Sonic is a cultural phenomenon, so everyone is going to weigh in. In this case, it was an easy problem (the design of Sonic is pretty creepy) with an easy solution (redesign Sonic).

But what good is simplistic sentiment analysis like this for the medium-sized business? How actionable is it really to be able to say, "Of the few thousand people who make up the segment you're targeting, most are generally neutral toward your new eyeliner," or, "Your reviews are mostly positive."

If the takeaway is "Get more positive reviews," or "Make your eyeliner more exciting," most marketers are going to say, "Thanks for nothing."

And while understanding some basic sentiment information is critical for big brands, whose value and profits can shift dramatically as a result of overwhelming negative sentiment (let's not forget the Pepsi/Kendall Jenner incident), this kind of data just isn't that useful for most businesses. A more complex analysis of sentiment that reveals hidden opportunities (potential new audiences, a need for a new product, possible negative reactions to a proposed new service) is of much more value.

Here's what that complex analysis looks like and what it means for businesses. We'll dive into a variety of types of sentiment analysis and examine the value each holds for data-driven brands:

• Polarity classification
• Fine-grained sentiment analysis
• Emotion detection
• Aspect-based sentiment analysis
• Intent analysis
• Multilingual sentiment analysis

Polarity classification, fine-grained sentiment analysis, and emotion detection

Modern sentiment analysis can go beyond simply defining whether sentiment toward your brand, your products, or your competition is positive, negative, or neutral (this is classic polarity classification). Fine-grained sentiment analysis essentially allows a business to determine the intensity of the sentiment being felt.

You and I both know that the difference between having a slightly positive experience with a product and a very positive experience with a product can be the difference between trying a product once and becoming a massive fan of a brand, buying their products for life.

It's valuable for businesses of all shapes and sizes to be able to understand whether general sentiment is swinging up or down (or hovering in the middle), but it's far more valuable to know that, for example, a product designed for a market segment is only receiving a slightly positive reception (instead of the huge splash you imagined), or that your competitor's flagship service is only slightly disliked (and not hated), and therefore will be harder to tear existing customers away from than you once thought.

Emotion detection is yet another layer of sentiment analysis that, when combined with fine-grained sentiment analysis or even simple polarity classification, gives a deeper picture of a situation and allows brands to make more strategic decisions.

Emotion detection essentially identifies some of the most basic emotions that human beings can express via text (think anger, fear, sadness, happiness, disgust), which in itself is valuable information. Who wouldn't want to know that their customers are literally disgusted by a proposed new product, or angry about the pricing of your flagship service? This kind of information saves brands from destruction.

But now imagine that these are all combined together. Knowing that most of your customers have a very positive reaction to your latest product, but that this is flavored by an undercurrent of frustration, can allow you to catch problems that are leading to this frustration before they get out of hand.

Imagine finding out that there is a slight negative sentiment toward a competitor's product, but this is flavored by feelings of sadness—digging deeper, you see that their customers are realizing that the latest product isn't great, but worse, is just the latest in a long line of failures. These customers are realizing there's no hope for their formerly favorite brand, and may be prepared to switch.

Aspect-based sentiment analysis

Knowing that there's a general sentiment of positivity toward your brand is of limited value, but imagine if you could refine that sentiment down to a single product, a single service, or a single feature of a product or service. Very few brands subsist on the earnings of a single product or service, so even being able to refine that deeply is useful, but far more useful is identifying the specifics of what people like or dislike.

For example, let's imagine that you learn one of your division's products is surrounded by positive sentiment. However, you then learn (through emotion detection) that there is an undercurrent of frustration.

You look deeper and find that a particular feature is the source of your customer's ire. This is all too common, especially in online reviews or upon release of a new product. Who could forget Apple taking out the headphone jack from their iPhones? Everyone was pretty much happy with the new iPhone in 2016 except for that glaring misstep.

Or better yet, suppose the sentiment had been negative? Narrowing down the source of the negativity (to the missing headphone jack) is of immense value when discussing the next iteration of the iconic smartphone. Apple chose not to change direction and kept the headphone jack out, but at least they knew the choice was unpopular.

Intent analysis

Knowing how people feel is one thing (and extremely valuable), but getting insight into a customer's intentions? Into the action they may or may not take? That's some serious value.

Intent analysis simply expands on fine-grained sentiment analysis and emotion detection. Of course we want to know that our customers are frustrated, but we really want to know when a customer is about to take action based on that frustration.

Knowing that a customer is frustrated with a problem that your product or service solves, and further, learning that there is intent to purchase a solution, can help marketers target likely buyers more precisely.

Multilingual sentiment analysis

While sentiment analysis has mostly focused on data in the English language, modern sentiment analysis tools are growing in their proficiency when it comes to analyzing other languages.

Multilingual sentiment analysis has obvious value for any multinational brand, but even a small business can benefit from an understanding of sentiment across languages. The United States, which continues to be the largest consumer market in the world, is home to no fewer than 8 different languages that are spoken by more than a million people (with 40 million spanish speakers), many of whom are interacting on social media in their native languages.

Savvy marketers understand the value of analyzing the sentiment of every segment of their market, no matter the language barriers that might exist. Not to mention the value of understanding a potential desire for the product or service you offer in another market that you've simply overlooked because no one in your company speaks the language.

Traditional, limited sentiment analysis is of limited value for small-to-medium-sized businesses

Useful, actionable information is what most businesses need, but that hasn't been available through traditional sentiment analysis.

What would have been useful would be to have known beforehand if audiences were already quite happy with the stylized, cartoony version of Sonic that already exists. It would have been valuable to know that there wasn't just negative sentiment toward any change in the character's design, but outright disgust at even the idea of doing such a thing (ask the Sonic fans—the new design was practically sacrilege!).

Even a more complex breakdown of the concerns and opinions that were being analyzed (rather than "Most people hate the new Sonic") would have been more valuable—for instance, what if there's a small subsection of the community that actually loves the new design?

That's why modern sentiment analysis, backed by machine learning and the growing science of AI, strives to do two things:

1. Give brands predictive data that is truly actionable (so that corrective measures can be taken before ill-fated marketing campaigns or product implosions get out of hand or quickly while they're in progress)
2. Analyze existing audiences that have an unfulfilled need (taking sentiment analysis out of the realm of PR and into the realm of lead generation/customer acquisition/conversion)

Modern sentiment analysis is about identifying untapped potential

While sentiment analysis can be used reactively to respond to problems (negative customer service interactions, negative reviews, product failures), what we think is more exciting (and frankly more valuable) is the potential to identify and target untapped markets or market segments.

By analyzing what customers want or need that they don't have or by looking closely at what customers hate about your competitors or potential competitors, brands today (with the right tools) have the unique opportunity to provide solutions that customers may not even know they're craving.

Seth Godin once said, "Don’t find customers for your products, find products for your customers." Untapped niches are still there—segments that have a need for a new solution (or a better solution) abound. Modern sentiment analysis helps you find those market segments and tap in before anyone else can.

Once you've tapped into these segments, you can start to use sentiment analysis to do a lot more than simply evaluate what customers want. Customer acquisition only begins with awareness (or a lack thereof). You can start to actually evaluate (and respond to) intention—and that's only scratching the surface of the possibilities.

Feelings are a huge driving force in the consumer market, so make sure you don’t stay ignorant of the sentiments and emotions which are influencing the decisions of your prospects and customers. Advanced sentiment analysis technologies are your crystal ball into the emotional undercurrents which can dictate the trajectory of your brand.