We outline the process and use of sentiment analysis to interpret consumer-based social media posts in a way that can provide an advantage for supply chain businesses. The development of consumer bodies who are interested in and comment on products are important to understand and model the process. Powered by AI development services, sentiment analysis offers supply chain professionals protection against malice and threat to firms that do opt to use wild data. We examine the potential that exists for sentiment analysis to unshackle supply management decisions from collaborative supply-chain-relationships. Also, this blog post highlights some risks attached to sentiment analysis and expresses several areas where further research is required.
It is complex socio-technical systems where both technical and social factors play important roles. The technical factors are system dominated and deal with technological and supply chain structural issues such as logistics, information systems, and supply chain performance (Li et al. 2015) While the technical factors are dealt with by formal mechanisms set up by organizations, the social factors such as reciprocity and mutual trust relate to the social system of the supply chain.
It has emphasized topics such as supply chain modeling issues that generally relate to the incorporation of the latest available technology in this area. Moreover, with opportunities for improvement in supply chain performance via breakthrough technical improvements are on a decline, the focus on enhancing supply chain performance by improving relationships among supply chain partners is on the rise. Unfortunately, there is scant literature on social media usage –supply chain links. Some studies have examined the impact of social media usage on supply chain events to develop contexts for these events and to build relationships between supply chain partners (e.g. O’leary 2011). In particular, we investigate whether social media usage of various supply chain partners is associated with supply chain performance.
Consumers must evaluate and discuss products online. Chatbot development services also play a key role in stimulating customer discussions and interactions across social media channels. This relationship holds that products must lead to discussion. Industrial or commercial sales in business-to-business (B2B) markets are not going to generate significant social media posts where corporate representatives vent about a particular product. Even if they do, the volumes will remain low.
Therefore, this shows relevance in business-to-consumer (B2C) markets only. Consumers are happy to evaluate and share opinions about products. They are also happy to comment on others’ opinions where they believe they are wrong or missing some important points. Furthermore, consumers often post questions about particular products’ performance or whether they live up to the hype/marketing promises; such questions often attract comments and answers from others that use the product currently, providing a discussion about the product. From this, we can see that such an opinion is most likely expressed in the B2C market. Where consumers want more information, or will take the time to comment on a product, it is likely to be more valuable or represent a greater share of their purchasing power (i.e., it is likely to be more expensive) or it may be relating to a change in a well-established product that is greatly loved (e.g., a change in recipe on a food product, such as replacing the more expensive and traditional cocoa butter with palm oil in Cadbury’s chocolates in New Zealand; in this case, with palm oil use seen by many consumers not only as a valid cost-cutting exercise but also as a contributor to rainforest destruction in southeast Asia (Adams, 2012).
Within the universe of possible purchases of a product or service, there will be some that are blithely unaware of the product but could potentially purchase, through to strong advocates that own the product. Warren (2008, pp. 345-356) provides an examination of how this range of customers can be modeled, focusing on how marketing efforts can create a ‘pipeline’ of developing customers, from an undeveloped (and thus, unlikely to purchase) customer to one that is a customer and makes an acquisition. A range of marketing efforts can be made to each of these sections of the pipeline and firms must be cognizant of how large each stock of customers is and what role they will play in creating discussions about a product.
Many firms would be cautious about supporting new initiatives when the impact on consumer buying patterns is considered. Sentiment analysis may enable a firm to seek out greater details about consumer response to supply chain sustainability initiatives (SCSI) that they seek to undertake. Research indicates that the SCSI of a buying firm can even influence the shareholder perception of the value of the supplier firms (Wang, Petkova, & Wood, 2014). SCSI from firms upstream from the market can be evaluated easily.
Assuming that competition is chain vs chain, this approach offers a new method to confound another chain’s efforts: ‘opinion warfare’. Firms can create fake 11 reviews that may mislead analysts at rival firms (Lappas, 2012). Traditional data for supply management approaches have been sourced either internally (in larger, vertically-integrated firms) or externally, from trusted partners (either collaborator within the supply chain, for from reputable firms such as those offering marketing services or market research services). In contrast, sentiment analysis relies on ‘wild’ data that is generated externally to the firm but cannot be validated or authenticated as being genuine.
Throughout this paper, we have attempted to analyze how sentiment analysis can be used to take consumer opinions, freely and openly expressed in web 2.0 and3.0 environments, and use these to feed into a system that uses them as a proxy for demand. This process enables a firm, distant from a market, to understand that particular market more effectively in a way that will allow them to make better supply decisions. We demonstrate the potential to understand the corpus of consumers more effectively. This includes targeting particular stocks of consumers, as well as using their comments to interpret how many consumers exist in each stock. The content of the consumers’ comments also provides an indication and insight into the likely supply management decisions that a firm will need to make, allowing supply management decisions to exist independent of collaborative relationships along the supply chain. However, sentiment analysis uses a ‘wild data’ that are unauthenticated; this opens up the analyst to malicious attacks from competitors through the new vector of false consumer reviews for products.
These topics are important as the increasing integration of contemporary forms of analysis, including sentiment analysis, into enterprise systems is starting, with a view to more strongly support decision-making (Mayeh, Scheepers, & Valos, 2012). A key practical limitation of the adoption of this approach in existing supply chain management teams will be the technical skills required and the change in mind-set. Many Supply Chain Managers and their staff will be more intimately familiar with management tools and methods that have been derived from engineering or production management foundations. Such tools tend to focus on quantitative data and are often used to understand elements of quality control; e.g., statistical process control (SPC). Underlying sentiment analysis, however, is a different concept: NLP. Therefore, the approach will fall outside the traditional educational programs for the foreseeable future and there would be a shortage of staff skilled in the use and analysis of NLP. Thus, we would anticipate it to be difficult for many firms to house such capabilities inhouse. However, there is no reason that such skills need to be in-house and this is a process that may be ideally outsourced to a third party of skilled specialists.