The untold playbook: How sentiment analysis is quietly redefining FMCG, QSR and telco strategy

Last week, we discussed how consumer sentiment analysis is a secret weapon for brands in consumer packaged goods, quick service restaurants and telecommunications. Furthering that topic, we are now turning to how consumer sentiment analysis combined with AI can provide powerful predictive analytics. 

In other words, we’re going to talk about knowing how consumers feel right now and how they are likely to feel in the future. 

Consumer metrics have evolved, and consumer sentiment analysis provides far more detailed information than more traditional measures like net promoter scores. Using consumer sentiment analysis — and taking into account chat transcripts, call logs, reviews, social media mentions and other sources that reveal how customers feel — gives brands a powerful edge in the market. 

Consumer sentiment analysis gives brands the opportunity to be proactive in adjusting and adapting their strategies and communications, rather than constantly reacting to dated measures. Using predictive analytics moves the needle even more. The combination of knowing how consumers feel now as well as what they are likely to do in the future creates a strong foundation for making profitable decisions. 

Modern life moves fast and each new technological evolution adds to the speed. The telephone made communication far faster than possible with mail, and cars and planes meant moving far faster than horses and wagons could ever go. Now, social media is having a similar effect on the speed with which products come to market, find favor among consumers, and in some cases, lose favor just as quickly. 

For brands, the speed of the product life cycle requires keeping a close watch on what consumers like, feel frustrated about or are getting tired of.

According to research group Informa’s trends report for 2025: “As purchasing habits change, companies must also focus on personalization and digital engagement to maintain loyalty. Private-label brands and direct-to-consumer (DTC) models are gaining ground, giving consumers more choices and intensifying competition.” Products touted by influencers that reach millions of followers on platforms like TikTok can catch on very quickly. 

In an environment where trends take off like wildfire, and fizzle just as quickly, AI is the surest path to tracking consumer sentiment. Using AI to forecast demand increases accuracy by as much as 30%. From product development to customer support, AI and machine learning optimize and improve operations in the FMCG sector. 

One of the areas where AI is excelling is in tracking emotional shifts, emerging keywords and sentiment velocity. Because it can parse millions of conversations, videos, transcripts and other content and categorize negative and positive sentiment, emotional shifts become clearer than they might be to a human monitoring various platforms. Similarly, emerging keywords are easier to identify. The combination of the two, emotional shifts and emerging keywords, give brands an idea of sentiment velocity — or, how quickly consumers’ perception of a brand is changing. 

One example of how AI can measure sentiment velocity can be found in an Adidas ad that connected a Palestinian-American model, Bella Hadid, to a sneaker designed to recall sneakers originally designed for the 1972 Olympics held in Munich, Germany. During those Olympics, 11 Israeli athletes were killed in a terrorist attack by a Palestinian group. Hadid has a history of outspoken support for Palestine, but said she was unaware of the attack in 1972. 

The backlash online was immediate, and came from virtually every corner. Prominent Jewish groups and Israel condemned the ad. The American Jewish Committee stated, “For Adidas to pick a vocal anti-Israel model to recall this dark Olympics is either a massive oversight or intentionally inflammatory. Neither is acceptable.” 

Hadid posted, “Connecting the liberation of the Palestinian people to an attack so tragic, is something that hurts my heart. Palestine is not synonymous with terrorism and this campaign unintentionally highlighted an event that does not represent who we are.” 

Adidas responded with multiple apologies, then pulled all images of Hadid from the campaign. Yet, the change in consumer sentiment was stark, with a 400% jump in negative mentions on social media. The shift happened quickly, too. The relaunch campaign was released on July 16, 2024, and by July 19, Adidas had apologized and “revised” the campaign. 

Less dramatic changes in consumer sentiment can be detected quickly as well. For instance, “flavor fatigue” comes through in sentiment analysis, so that brands can choose to release fewer pumpkin spice-flavored products to adjust to lower demand. Conversely, brands may choose to invest in more health-conscious products when consumer sentiment improves for products with “low-sugar” or “all-natural” ingredients. 

Quick Service Restaurants (QSRs) roll out new menu items frequently, and similar to the FMCG sector, brands need to be able to respond quickly to consumer sentiment in order to optimize offerings and prices and streamline operations. Unfortunately, traditional methods for optimizing menus and ordering processes are slow and localized. By using real-time sentiment analysis, brands can discover how consumers feel about ingredients, pricing and menu changes across broad audiences. 

Recently, at a conference held in Chattanooga, Tennessee, called Project Voice, Jay Ruparel of PizzaVoice discussed how his company tested and developed an AI product specifically for pizza restaurants. He noted that in some pizza restaurants, 70% of the orders are for pepperoni pizza, and in completing market research, his company discovered there are 1800 different ways people order pepperoni pizza. A pepperoni pie. A pepperoni slice. Extra cheese and pepperoni. 

They also discovered that things like hold time and the time of day mattered a great deal to the call-to-order conversion rate. “When people are hungry,” said Ruparel, “they just want pizza.” The average hold time for a pizza restaurant during a busy time is more than one minute, meaning many people hang up and call a different restaurant. 

PizzaVoice used thousands of hours of call transcripts from real pizza orders and natural language processing to develop their AI agent. The company implemented sentiment analysis to determine that upselling shouldn’t happen during the lunch rush, when speed is the priority. When upselling is offered, it’s based on the items in the customer’s cart. 

Actionable sentiment signals

Knowing how consumers feel is one thing, being able to take action based on those feelings is another. Brands that track sentiment closely can respond to positive sentiment around specific flavor profiles. 

“Why was bacon so popular 10 years ago? There were bacon memes on the front page everyday at one point. Where and how did it start?” is a Reddit thread from about four years ago, and the conversation that follows is a lesson in how consumer sentiment can develop, coalesce and fade. 

Today, a savvy QSR brand would have been able to be slightly ahead of the bacon curve by tracking consumer sentiment. The memes, comedians and other mentions across platforms now signal different trends like meatless, protein-packed, all-natural and others. 

The flip side of the consumer sentiment signal coin is tracking negative sentiment. “Shrinkflation” and “price hikes” are common negative keywords right now, and tracking mentions of them can help brands optimize pricing without angering consumers. 

“Modern sentiment analysis is deeper than just distinguishing positive reviews from negative ones. It identifies specific aspects of service that delight customers, uncovers emerging concerns before they become trends, and highlights opportunities for strategic improvements that drive genuine business growth,” according to a Guide to Sentiment Analysis of Customer Reviews by the company iOrders. 

Micro-optimizations

Limited-time offers (LTOs), menu simplifications and regional menu variations are a few of the tools that QSRs have to personalize service, and using AI for consumer sentiment analysis can guide effective use of those tools. For example, a restaurant could use an LTO to test a new menu item before officially launching it across the brand. 

“Ask customers for their thoughts,” recommends Gordon Food Services. “Put questions out about your LTOs on social media, leave comment cards for them to rate items and have direct conversations.” This gives brands an opportunity to make changes based on that feedback, and it also builds loyalty. 

The same idea is deployed at scale with consumer sentiment analysis. Instead of depending on a few hundred responses, brands have the option to garner comments from thousands of consumers, across numerous platforms and regions. 

AI + sentiment

Another thing that PizzaVoice learned as they trained their AI agent was that regional and demographic factors make a big difference in how and when people order pizza. Some people order in English, some in Spanish and in some places people use a mix of the two — Spanglish. AI can account for localized differences, but deliver information about broad audiences so that brands can adjust as necessary. 

During the Super Bowl and other similar events, large groups of people order pizza together, passing the phone from person-to-person. Brands are generally prepared for such large events. What about smaller, more regional events? For example, if a city is hosting a high school playoff event, they may experience a surge in business. 

Accounting for demographic differences can become complicated quickly for national brands. Using an AI model to group responses and track consumer sentiment makes it possible for brands to personalize menu options without creating operational complications. 

To say that the telecom industry is competitive is a bit of an understatement. 

In order to survive, telecom companies must prioritize customer experience. Consumer sentiment analysis and AI can help do that. 

AI can categorize emotions expressed by customers in calls, social media messages, emails, reviews and other communications as positive, negative or neutral. That information can be used to proactively enhance services, target specific marketing campaigns or make other adjustments as required. 

Early warning signals

Language that expresses negative emotions, such as frustration at slow internet speeds or anger over billing, predicts churn better than satisfaction scores can and more quickly than surveys or other tools can. By tracking consumer sentiment closely and adapting, telecom companies can reduce churn proactively. 

Proactive retention strategies

Another speaker at Project Voice was Art Coombs, CEO of KomBea. KomBea provides product development services, including AI agents that respond to customer service calls in call centers. “Our goal is to combine human intelligence with artificial intelligence,” said Coombs at the beginning of his presentation. 

KomBea’s product ProtoCall AI® is implemented individually with clients, and trained on each client’s best customer service calls. “Think about how your best customer service rep would handle a problem,” says Coombs. “We use those interactions — the best — to train the AI so that all of your calls are handled in the way the very best reps would do it.”

Along with using transcripts and recordings from the best calls, the AI agent can be further programmed to tailor special offers, proactive services and more to dissatisfied customers. Sentiment analysis is built into the training and used continually throughout the calls. 

The predictive power of AI

Combining real-time sentiment trends with individual customer data reveals an extensive profile of the customer. That kind of data is one reason that 90% of telecom companies use AI. 

Predictive analytics allows companies to identify usage patterns, avoid outages, provide the right level of service, and to reduce churn. When a telco can predict that usage in a particular area will be exceptionally heavy at a certain time of day or day of the week, they can deploy extra resources and avoid an outage. 

Similarly, a family with students in high school is likely to need a different level of service than a retired couple. Preventing outages and tailoring services are two crucial ways to improve customer service and reduce churn. 

Benefits and challenges

Although the benefits of AI and consumer sentiment analysis for telecom companies are clear, some challenges exist. Integrating the newer technology with existing systems can be difficult. Managing the initial investment can present a barrier, and bridging the skills gap can also slow adoption. 

Even with those challenges, telecoms often have millions of passive customers, and traditional, surface level surveys can miss brewing dissatisfaction. Sentiment analysis passively monitors the digital signals that can help companies head off that dissatisfaction. 

Consumer sentiment analysis is a useful tool for FMCG, QSR and telco businesses because it offers a real-time view of how consumers feel. Yet, simply knowing how consumers feel at a given moment means brands are still reacting. 

Combining consumer sentiment analysis with AI to look at how consumers feel in the moment, along with historical data, buying patterns over time, individual consumer profiles and information from a variety of internal and external platforms, changes the playing field. Predictive analytics means strategies and communications can be proactive. 

For FMCG, brands can test products and garner feedback, quickly, efficiently, and with accurate results. QSR operations face less risk in making menu or operations changes. Telecom brands can avoid problems and reduce churn by using predictive customer sentiment analytics. 

Understanding how consumers feel today is important, but knowing how they are likely to feel tomorrow is a super power.