More organizations now use sentiment analysis in their media monitoring and measurement programs to gauge public sentiment toward a company and its brands. Whether done by human analysts or with computer programs, sentiment analysis grades social media comments, online reviews and news articles on a positive to negative scale. It also helps identify issues consumers may have with the organization or brand.
Companies, nonprofits and other organizations can obtain a treasure trove of insights by analyzing the enormous amount of data online. The problem is that the amount of data can be overwhelming. Making sense of it, identifying clear problems or opportunities, and developing actionable recommendations can be difficult.
Some companies and brands may equate a sizeable number of media mentions with positive brand awareness. However, more advanced techniques can uncover actionable recommendations to improve a company’s PR and marketing, design better products, and improve its customer service and other business practices. Here’s how:
Segment customers. Segmenting customers by demographics can help reveal helpful insights. Naturally, customers have varying preferences and priorities depending on their age, income, and other factors. B2B PR and marketing pros can segment prospects by company size, budget or their business objective. Examining a subset of customers can reveal information that a bird’s eye view cannot.
“When you have multiple datasets of behavioral data that you can compare against one another, your team can understand how to cater to various customer segments by understanding their motivations,” says Rohan Ayyar, regional marketing manager for India at SEMrush, in Target Marketing.
Seek patterns. Look for repeated keywords on social media and review sites to uncover common problems with customer service, product quality, pricing or other issues, such as packaging or delivery. You’ll most likely find those repetitions in negative reviews. Businesses generally focus on resolving complaints and eliminating bad reviews, but those negative comments offer a treasure trove of product research.
To gain the most from reviews, businesses can undertake a concerted effort to solicit both negative and positive reviews by explicitly requesting them at the point of sale. That requires wading through the many overly emotional comments and fake reviews.
Segment reactions by issues. Customers may have very different reactions to different aspects of the business. They may be happy with customer service, but dissatisfied with its packaging. Separating sentiment by issues can reveal high level of negative or positive sentiment for specific aspects of the business when overall sentiment seems balanced.
Employ human analysts. While automated sentiment analysis can review large amounts of data more affordably, trained human analysts are more likely to deliver useful insights. Despite advances in automated programs, human reviews remain more accurate, as trained analysts can better understand the slang, irony and sarcasm common on social media. Appropriate training and consistent criteria for evaluating posts improve accuracy.
To evaluate automated programs, compare results produced by each system against the results produced by human analysts. Some media monitoring services can combine automated sentiment analysis with evaluation by human analysts.
Analyze in real-time. Public sentiment toward an organization can change overnight – or even within hours as negative comments spread virally online. In some scenarios, the news media promptly reports their complaints. When a passenger was forcibly removed from a United Airlines aircraft, other passengers posted videos of the violent removal. Videos and damaging comments spread virally within hours, leading to extensive critical media coverage the following day. Real-time sentiment analysis could have alerted United Airlines about the imminent PR crisis and depth of public outrage, and encouraged it to respond more humbly rather than issue a feeble apology.
Real-time sentiment analysis combined with email notifications can alert PR personnel of a spike in negative sentiment that portends a coming PR crisis. PR can then take swift actions to respond on social media and to media outlets, if needed, to limit negative reactions.
Examine changes in sentiment over time. Examining how brand sentiment changes, hopefully for the better, over time can demonstrate the effectiveness of PR and marketing. Changes in sentiment is especially useful for examining a brand’s PR responses during and after a PR crisis. Combining sentiment analysis with share of voice can show how well the organization’s PR performs compared to competitors.
Combine sentiment analysis with other tools. While online sentiment analysis is generally faster and more affordable than traditional market research tools, such as surveys and focus panels, combining it with other research methods can deliver a more comprehensive and possibly more accurate results. Depending on the platform, social media commentators may be younger, more liberal, and more educated than the public at large or a brand’s overall customer base. Such demographic differences may skew sentiment reports.
Bottom Line: Sentiment analysis can produce an overwhelming abundance of data. Parsing that data can uncover insights that benefit PR and marketing as well as other departments.
This article was first published on the Glean.info blog.