The food and beverage industry is under extreme pressure to reduce costs, improve customer service and drive incremental revenue. To stay ahead of changing consumer tastes, capture new market share and grow revenue, food and beverage companies find they must introduce new products at a faster rate than ever before. More products distributed through more channels may improve customer satisfaction and increase revenue while also creating unintended consequences such as lower forecast accuracy, higher total inventory, increased distressed inventorand lower in-stock availability.
New product introductions often have limited demand histories to base future projections and can display both erratic and localized demand patterns. In today’s connected and always-on world, the impact of a social post can quickly and significantly impact the demand for a new product. Celebrities, for example, often serve as arbiters of taste, style and public opinion. A positive tweet from one can send demand soaring, while a negative post can quickly drive a product out of the market.
The effects of social media on business has led to an emerging practice of measuring the emotions behind social media mentions—social sentiment. Social sentiment measures the tone of the message and assigns a value or score to it based on several factors such as: is the comment very positive, positive, neutral, negative or very negative? Without sentiment, data can be misleading. Just because you receive a high volume of mentions following the launch of a new product does not necessarily mean the new product has been well-received.
For any size company the process of manually sorting through large volumes of data to determine sentiment can be a significant time commitment. Through the use of artificial intelligence (AI), it is now feasible to capture and mine social media data to determine social sentiment and then extend this information to show the impact on demand to help supply chain teams more accurately plan their operations. Today’s machine learning algorithms have the ability to correctly categorize the majority of social media sentiment. As these solutions work through more data they are able to learn the differences between humor, sarcasm, irony and so on to improve their success.
Examples where social sentiment can impact operations:
- Evaluate the Health of a Brand: An understanding of how your target market feels about your company, product and services through analysis of overall sentiment can provide valuable insight into the health of your brand.
- Address a Crisis: Analysis of social sentiment might reveal a spike in negative posts and provide an early warning to a potential product or service issue. Through alerts and analysis, the root cause of the issue can be uncovered and corrected.
- Research the Competition: Social sentiment analysis can help you understand how you are positioned against the competition.
- Improve Demand Prediction: Companies can now use the ‘Voice of the Consumer’ to drive improvements in forecasting and inventory positioning.
Social Sentiment for Food & Beverage
For a food and beverage manufacturer, social sentiment analysis can help influence and improve product forecasts and new product introductions. At first glance, looking at the conversations for a beverage category on Twitter can be quite daunting. The challenge is separating the noise, evaluating relevant posts and developing a sentiment score.
Algorithms process each tweet, evaluate the sentiment of the text and then sort the tweets into positive, negative and neutral sentiment (see Figure 1). Over time the machine learning system gets smarter and more detailed in its understanding of this Twitter data. As more tweets are evaluated, the system automatically refines its categorization values and sentiment scores. This same data can be used to evaluate the potential sources of risk such as competitive new product introductions, supplier challenges or shifts in consumer tastes.
The food and beverage manufacturer can now use the newly created sentiment score in the demand planning process to improve forecast accuracy and as an input to the development and launch of new products. For example, the forecast for a product with a highly positive social sentiment may be increased to reflect higher than expected demand (see Figure 2). Sentiment data is updated weekly to catch changes quickly. Sometimes all it takes is a famous person tweeting about a product to drastically change demand.
When introducing new products, these machine learning clustering techniques can also be used to identify existing groups of look-alike products. A new product is associated with an existing product cluster based on product attributes and assigned a starting sentiment score based on the product cluster score. Sentiment scores can then help to determine when and where a new product should be launched and the likely base demand and launch curve.
The food and beverage industry is one of the world’s most highly competitive and socially visible business sectors. Creating accurate forecasts can be challenging as customer tastes change frequently forcing a constant stream of new product introductions. Small improvements in forecast accuracy can lead to big gains in manufacturing efficiencies, lower inventory, less distressed product and higher fill rates all of which drive top and bottom line improvements. Although the use of social media to improve supply chain planning is still very new, the potential impact to highly competitive industries could be significant. Food and beverage supply chain leaders should actively investigate how social media and analytics can improve their operations.