How Nordstrom Used NLP to Fix In-store Service
Retailer brands generate a lot of data at their stores and warehouses:
- Customer data
- Inventory data
- Sales data, etc.
Big data and secure cloud computing with built-in AI and Machine Learning capabilities allow retailers to leverage these data points to optimise production and operations and deliver top-class service to customers.
One such example is American retail giant Nordstrom.
When Nordstrom decided to launch its discount department store chain, they knew they would compete directly with T.J.Maxx, Marshalls, and Ross Dress for Less.
Instead of competing with these brands on price alone, Nordstrom decided to look at their data and see where they could improve to stand out.
They stopped concentrating on what the competition was doing and put their focus on improving customer service.
Nordstrom took their customer feedback data and ran it through an NLP model to understand the most pressing issues that led to bad customer experiences at their stores.
A common complaint that kept cropping up was that customers had trouble locating in-store sales people quickly.
Nordstrom’s in-store employees wore street clothing, making it difficult for customers to differentiate between store employees and fellow shoppers.
As the NLP model analysed tons of customer feedback text and brought this issue to light (among other issues), Nordstrom quickly put their store workers in bright-colored t-shirts, making it easy for customers to spot them.
The result?
Within two days of that pilot, the company saw a 30-point jump in the key metric to evaluate sales staff effectiveness.
This is but a small example of how retailers can analyse their data and optimise business that immediately affects sales and revenue.
Actually, how does NLP work?
While machine learning models can run through tons of data very quickly, what makes them effective is labeled data.
Labeled data tells ML models where to look and what they mean.
So the following question is “who labels the data?’
It’s not like retail brands have data labellers on their payroll read to jump in.
Finding skilled text data annotators is an enormous challenge.
While many reputed data annotation companies specialise in image and video data labelling, very few are good at text data annotation.
This is because text data labelling is a particular skill, and a piece of text can be interpreted in many ways.
For example, anybody with minimal training can label image data by drawing boxes over objects on photographs. Drawing bounding boxes around video footage isn’t that difficult either — purely because there are no subjective elements in labelling visual data. However, text annotation is a different ball game altogether.
You can address the subjectivity in text data labelling by training data labellers in the relevant domain.
Do retail enterprises have the time and budget to hire and train data labellers?
That’s why we built Traindata, Inc. — a data annotation company with bespoke data labelling tools and highly skilled, certified, and qualified data labellers.
Visit www.traindata.us to learn more and hire us to label your data.
