Can sentiment analysis improve merchandising calls?
Image: Getty Images

Can sentiment analysis improve merchandising calls?

brian kilcourse 450x450

Through a special arrangement, what follows is a summary of an article from Retail Paradox, RSR Research’s weekly analysis on emerging issues facing retailers, presented here for discussion.

Marketers have long used sentiment analysis to target offers in the digital domain, but a new generation of demand forecasting capabilities is extending its use into merchandise planning.

Sentiment analysis examines “non-transactional” data that consumers leave along their digital paths-to-purchase that can offer an early indicator of demand. A simple example is the number of positive mentions on social or e-mail streams about a new offering. Online clicks, geo-location info and other digital data can also offer clues as to whether underlying sentiment is positive, negative or neutral.

Our latest supply chain benchmark report finds retailers clearly interested in modernizing their forecasting engines to take in new data (i.e., weather, trade area, competitive pricing), but standing out on their list of “wants” was sentiment.

cht RSR sentiment analysis
Source: RSR Research, January 2019

The findings show apparel and brand managers, who are probably both more experienced and adept at using sentiment analysis in marketing, are only slightly more interested in using that data for forecasting than grocers, drug store, and convenience stores (FMCG) – and most of those that are interested in using it already are. This no doubt is the result of both those retailers’ design-focused merchandise planning processes and long supply chains; once a product is committed to, it’s much harder to turn back than it might be in faster-moving verticals.

The “sweet spot” in sentiment analysis for forecasting is with general merchandisers. Those retailers feature a lot of seasonally sensitive and popular items. But, unlike fashion and specialty merchants, they are fast-replenishment retailers and can react relatively quickly to sudden shifts in demand (for example, if an item featured in a popular movie hit). And since they tend to be high-volume retailers, a mistake can be very costly, while successfully predicting as shift in demand is very rewarding.

The news is good for solutions providers: the desire is there, and there’s plenty of upside before the “implemented/satisfied” group is as big as the “very valuable” group of retailers.

BrainTrust

"While the potential benefits of sentiment analysis are many, today it’s still more art than science."

Mark Ryski

Founder, CEO & Author, HeadCount Corporation


"It’s not only about sentiment analysis but also about hyper-granular analysis of consumer feedback in various channels from product reviews to social media to CS calls."

Nir Manor

Retail-Tech Specialist Advisor


"Driving merchandise decisions based on customer desires and needs should improve the chances that retailers will buy and stock the merchandise that customers want."

David Naumann

Marketing Strategy Lead - Retail, Travel & Distribution, Verizon


Discussion Questions

DISCUSSION QUESTIONS: Where do you see the opportunity and limits in using sentiment analysis to help gauge demand for future product assortments? Do you see it providing just as much benefit to merchants as marketers?

Poll

9 Comments
Oldest
Newest Most Voted
Inline Feedbacks
View all comments
Mark Ryski
Noble Member
5 years ago

While the potential benefits of sentiment analysis are many, today it’s still more art than science. Forecasting demand is tricky at the best of times, and including sentiment data inputs doesn’t necessarily make this any easier. The key challenge of sentiment analysis is the diversity of data sources and reliability of this data. The true value of the insights will ultimately be contingent upon the retailer’s ability to identify the signal from the noise in the data. There’s no doubt that sentiment analysis will continue to evolve and will likely play an increasingly important role in product and other important decisions.

Jamie Samson
Reply to  Mark Ryski
5 years ago

The real finesse from sentiment analysis really comes out when you apply machine learning techniques to compartmentalised data. For example, I have built a machine learning model to mine leisure & entertainment reviews and opinions. I noticed that the language used in L&E and retail store reviews for example, are vastly different. Training models to understand specific industry language will be the next step in honing in the accuracy and ultimately the insights you can extract from such techniques.

Jeff Sward
Noble Member
5 years ago

I was hoping “sentiment analysis” was the same as answering the “why?” question (apparel). WHY did the customer choose style A over style B. Style? Color? Fit? Fabric? Trim? Performance (stretch)? Which product attribute should be replicated when chasing a best seller? Which attribute should be changed to fix a slow seller? What did the customer “feel” about their choice? Getting to “why” is still a challenge.

Adrian Weidmann
Member
5 years ago

I suspect sentiment analysis (as defined by Brian) is required by, but not limited to, apparel manufacturers as part of their trend analysis — what fashion will shoppers be interested in next summer? Being able to design, manufacture and sell in the future is critical for fashion. Retailers like Lowe’s and Home Depot need to know what fashion plumbing finishes will be in favor during the next construction period. Will it be bronze, chrome, or brushed nickel finishes? This trend analysis is required for merchants and buyers. Any trend or sentiment analysis is valuable for the entire manufacturing and retailing ecosystem — regardless of the product or category. Once again we find ourselves back at the importance of supply chain visibility and optimization — a foundational lodestone of a majority of the issues presented here on RetailWire.

Nir Manor
5 years ago

It’s not only about sentiment analysis but also about hyper-granular analysis of consumer feedback in various channels from product reviews to social media to CS calls. It’s all about automatic analysis of what consumers say, what they mean and their sentiment.

Cynthia Holcomb
Member
5 years ago

Predicting sentiment based on linear representations of data, i.e.; geo, clicks, trade area, weather, etc. is a slippery slope, subjective to the interpretation and possible bias of the software or human translator. The case for sentiment analysis for general merchandisers has a higher probability of demonstrating value than retailers of apparel and other products purchased based on highly sensitive personal preferences. Like the thousands of other solutions out there angling to interpret human behavior to improve sales, supply chain, etc. sentiment analysis focuses on complex interpretations hoping to interpret the abstraction of the individual human decision to purchase.

Patricia Vekich Waldron
Active Member
5 years ago

Disappointing to see most votes cast for “unlikely” when data and tools exist today to identify both long- and short-term trends!

Gib Bassett
5 years ago

Having the capacity to test analytic use cases like this is important for retailers in many segments to possess – fast fashion, big box retail, discount, and grocery. The hypothesis that consumer sentiment expressed in social channels – their words and language, phrases, and images can serve to gauge and predict product preferences and demand is a solid one. However it’s just one signal that must be combined and analyzed with others as pointed out (sales, PII, weather, economic). It requires testing relative to how merchandising and assortment is accomplished today – what’s the goal, how much can you improve? More ideally, it’s part of a plan across the company to deliver a seamless customer experience using available data and analytics methods. There’s also an element of organizational success required here since much of these decisions are today done more qualitatively. So it needs executive support.

David Naumann
Active Member
5 years ago

According to a recent BRP survey, only 16% of retailers incorporate social media data (e.g. preferences, affinities or trends) into their merchandise planning processes. Social media still represents a huge opportunity for retailers to be more customer-focused in their planning. Driving merchandise decisions based on customer desires and needs should improve the chances that retailers will buy and stock the merchandise that customers want. However, this is easier said than done. There is an enormous amount of data to sift through to identify social media sentiment and trends and as others have mentioned, the risk of making the wrong assumptions could be costly.