Analysis of customer’s feedback

Situation

One of well-known Central Europe e-commerce merchants asked Unicsoft for cooperation to create a tool intended to analyse a customer’s feedback on goods purchased through web marketplace, The business goal was to increase customer loyalty, drive business changes, and deliver real return on investment. Customer experiences fall into three basic categories, positive, negative or neutral. Through sentiment analysis, the goal was to detect the tone and temperament of each and every word found in a customer’s social postings and categorize those sentiments as either positive, negative or neutral.

Solution

After data cleaning and munging , an initial step of words tokenization was applied; having it done, an NLTK tools was used to define synonyms, semantics, and overall mood of feedbacks; aligned with scores given by these particular feedback authors, a section of manual business logics was brought in: language specifics, abbreviation, collocations and vernacular expression played a significant role in overall semantic analysis. Furthermore, analysis of new products or product lines could significantly affect overall strategy. Alongside with NLP and semantic analysis, there were used a Data Science techniques : given a data from social network that customer logged in through, a set of demographic features was involved to model, summing up into complex analytical solution.

Result

Unicsoft set a dedicated team comprised of highly-skilled analysts and mathematicians and software developers; this particular solution helped the Client to define his marketing and sales strategy which resulted in 10% revenue increase within one year after the deployment. The Client noticed the overall high-level of Delivery and solution architecture.