While companies spend millions of dollars developing their products and enhancing them to match the present-day need of the customers, they often fail to understand the customer perception of a specific product or a range of products.

Take an example of a cosmetic company that launched a new product in the market. It was observed that the product had received good ratings on social media. The product received an overall positive response in ratings and reviews from customers, however, the sales were low and did not correspond to the ratings. The mismatch of the customer engagement and sales could be attributed to stock-outs of the product where the demand was high or excess inventory in places of low demand. It could also be due to negative sentiments regarding the price of the product.

Customers often mention these in their reviews, and if the company could read through all the reviews periodically, they would be in a much better place to understand what was going right or wrong with their products.

With the complication of high engagement and low sales, the global cosmetic manufacturer wanted to understand the overall sentiment around the product and the reason behind the low sales. It was difficult for them to make sense of a large number of reviews that they were receiving online.

Some of the hurdles for the client were:

  1. Analyzing the reasons for the ratings for the products/brand /category
  2. Determining the attributes of the product that were important to customers and their preferences.

The client was looking for a smart, efficient yet simple solution to boost the image of their high-selling product and the ratings.

The client did not have the end consumer or transaction data as they are a CPG manufacturer. A key product was identified after a discussion with the client for which analysis was to be conducted. Specifically, Impact Analytics extracted the relevant data from online sources and then structured it in a manner wherein a proper analysis could be carried out. Using AI and ML-based algorithms, Impact Analytics helped the client understand the customer perception around their high-selling product/brand. Impact Analytics also helped identify the cluster reviews into various significant attributes (pricing, packaging, availability, etc.) and helped them identify the sentiment of that attribute.

The main features of the analysis were:

Identified the product and the key categories to be analysed based on the discussion with the client. The sentiment around price, quality, user experience, and packaging could not be determined by the client based on reviews and ratings received.

Identified the online data sources with all the reviews/ratings of the product

Scrapped the ratings and reviews from online sources. Data extracted from various sources came in different format and cuts, i.e., different age groups, reviewer type, etc. All the data extracted was consolidated, and patterns were identified to bring in data consistency.

Used advanced algorithms to find the sentiment for each product. Clustering and sentiment analysis of reviews were done leveraging Natural Language Processing (NLP) based topic modeling.

AI-based word clouds were created to give better insights into ratings and the sentiment. Ratings and their corresponding sentiment were not correlated as derived from the analysis.

It was identified that sentiment for the attribute ‘price’ was negative across segments and the product was deemed ‘too expensive.’ It was also found that the packaging was not impacting the choice and experience of the consumers (as perceived by the company) but it was the price/value for money aspect that affected their purchase decisions. The most important attributes were the user experience and quality, as customer satisfaction and loyalty depended on these. The key features identified that lead to an increased ROI for the client were:

  1. Four key attributes were identified using NLP based engine and machine learning algorithms: price, quality, user experience, and packaging.
  2. Sentiment analysis was carried out for each attribute in key categories and was identified across segments.
  3. Overall sentiment analysis for a review was compared with regard to the attribute wise sentiments.

With the help of Impact Analytics, the global cosmetic manufacturer was able to improve its brand communication strategy by working upon their most important product attribute. The web-based sample dashboard created can be scaled up depending on the scope and breadth of data available for easy representation and interaction. Better analysis of feedback and ratings helped the client to explore innovative marketing channels to improve their consumer engagement and improve their ROI on promotional budget.