AI in Retail Analytics

Consolidated details from various sources on Data Analytics in Retail

This blog details you about the necessity of analytics in our daily life. It’s indeed a mix of content from various other websites, blogs and short notes so as to find everything one wants to know about Retail Analytics in one place. It helps one to know more about how Data Analytics have been applied to various fields and domains. You will also get to know about the evolution of analytics field towards the end. Get started to find all about Data Analytics.

Data Analytics

Data analytics — the process of examining data sets in order to draw conclusions about the information they contain, increasingly with the aid of specialised systems and software.

As the name depicts and the definition states “data” is the key to all the processes. Data could be of any form — text (structured, unstructured or semi-structured etc.), image (1D, 2D or 3D etc.), voice (single user or multi-user with noise/without noise etc.)

Data in Retail Industry — Big Data that falls essentially into any of the above stated forms. It is of great importance for those willing to take profitable decisions concerning business. Retailers manage to use data and develop a “peculiar psychological portrait” of a customer to learn his/her sore points.

Types — Data Analytics

Analysing the data to extract meaningful insights from it and for better ROI opportunities are usually done in four different stages based on requirements.

Any kind of analytics that happen using any kind of data (text, image or speech) is likely to fall under one or more of these types of analytics.

• Descriptive
• Diagnostic
• Predictive
• Prescriptive

Descriptive

• Preliminary stage of data pre-processing
• Creates summary of historical data to get useful information (or) prepare data for further analysis
• Eg: What was my revenue last year?
• In short: What happened?

Diagnostic

• Deeper look at data to understand the causes of events and behaviour
• Analysis of why outcomes resulted as they did
• Eg: Why was my revenue high?
• In short: Why did it happen?

Predictive

• Provides insights of what might happen in the future
• Gives the probability for a particular event to happen or not in future
• Eg: What will be my revenue this year?
• In short: What could happen in the future?

Prescriptive

• Best course of action for a given situation
• Related to both Descriptive and Predictive analytics
• Using known factors from Descriptive analytics and predicted probabilities from Predictive analytics, Prescriptive analytics will decide the next steps to be taken. It can also be stated as one of the trickiest steps to be taken.
• Eg: Can we launch a new product for this set of customers?
• In short: How should we respond to those potential future events?

Need — Data Analytics

• With the increasing competitiveness within the retail sector, it is extremely important that service processes are optimised in order to satisfy the customer expectations.
• Analysing data and making changes adaptable to earn customer favorability is very important to generate profits.
• Retail data analytics deal with identifying potential customers based on their past purchases, finding the most appropriate way to handle them via targeted marketing strategies and then deciding what the next offering should be.
• The need for Data Analytics in Retail Industry could be elaborated further by understanding the functions of Data Analytics.

Functions of Data Analytics

Recommendation Engines

• Main leverages on the customers’ opinion
• Increase sales and to dictate trends
• Either collaborative or content-based filtering
• Up-sell and Cross-sell recommendations depend on customer’s profile, past behaviour, demographic data, usefulness, preferences, needs, previous shopping experience, etc.

Market basket analysis

• Traditional tool of data analysis in the retail — customer purchase behaviours
• Uses transactions with more than one item, as no associations can be made with single purchases
• Association rule mining algorithm — if you buy a certain group of items, you are more (or less) likely to buy another group of items. IF {jeans, no shirt} THEN {t-shirt} within a store.
• Types : Differential (compares across stores) and Predictive (classify items that largely occur in sequence) Market Basket Analysis

Warranty analytics

• Warranty claims into actionable intelligence
• Detecting anomalies in the warranty claims
• Data and Text mining approaches

Price optimisation

• Right price both for the customer and the retailer is a significant advantage brought by the optimisation mechanisms
• Data gained from the multi-channel sources — the location, buying attitude of a customer, seasoning and the competitors’ pricing
• By means of real-time optimisation the retailers have an opportunity to attract the customers, to retain the attention and to realise personal pricing schemes

Inventory management

• Stocking goods for their future use
• Stock and the supply chains are deeply analysed via powerful machine learning algorithms that detects patterns, correlations among the elements and supply chains
• Defines the optimal stock and inventory strategies
• Spot the patterns of high demand and develop strategies for emerging sales trends, optimise delivery and manage the stock

Location of new stores

• ZIP code and location analysis — potential for the market
• Great attention to the customers’ demographic factor
• Location of other shops — for deep comparison along with retailer’s network analysis
• Solved using Machine learning algorithms

Customer Satisfaction

• Natural language processing techniques for analysing the opinion (positive, negative or neutral) a customer has towards something — Opinion mining
• Data sources: social media — reviews or comments
• Helps in leveraging the retail standards and becomes a background for overall service improvement

Merchandising

• Merchandising tricks helps to influence the customers’ decision-making process via visual channels
• Attractive packaging and branding retain customers’ attention and enhance visual appeal
• Data science analysis remains behind the scenes in this case

Lifetime value prediction

• “It costs 10 times less to sell to an existing customer than to find a new customer”
• Customer lifetime value (CLV) — assess the financial value of each customer
• Collect, classify and clean the data concerning customers’ preferences, expenses, recent purchases, demographics, transactions and behaviour to bucket them into different groups
• Target regular customers with coupons, offers, deals, packages, discounts, et cetera etc. and track pattern to find are they likely to stay on a long run

Fraud detection

• Retailer’s reputation — lies entirely on how safe and secure the customers’ returns and delivery, transactions etc. are handled
• Machine Learning algorithms (Deep Neural Networks) developed for fraud detection should not only recognise fraud and flag it to be banned but to predict future fraudulent activities

Natural Language Processing in Retail

• Natural Language Processing (NLP) is one of the fields of AI that relies on Machine Learning (ML) for processing data in the form of ‘texts’
• Used in various domains (healthcare, education, law etc.) plays an important role for data analytics in retail industry as well.
• Functionalities of NLP in Retail includes various applications that could serve retailers while some of them are detailed below.

Natural Language Processing Applications

Sentiment analysis

• Understanding customers well from customer reviews, comments, suggestions or feedback from social media, blogs or forums.
• Falls under multiple functionalities of Data Analytics — Future Performance Prediction, Demand Prediction, Customer Satisfaction, Merchandising, Lifetime Value Prediction, Inventory Management etc.

Chat-bot

• For creating a good interactive service to know your customer
• This falls under various functionalities of analytics like to find the next availability of a particular item — Inventory Management, ratings of a product — Customer Satisfaction, provide feedback via chat — analyse sentiment from user chat text — Customer Satisfaction and etc.

Automating Invoice Processing

• While manually going through the Invoices become tedious in large firms, automating the process of finding the specific feature from Invoices could be done using NLP techniques.
• This can also be extended in distinguishing malicious data from original ones (warranty, invoices, claims, cheques, etc.)

Computer Vision in Retail

• Imaging techniques in retail industries are emerging and becoming the need of the hour.
• These imaging techniques does not necessarily only help in analysing the customer to satisfy their needs but it also helps in improving the retailers business standards with one of the case being determination of the quality of goods that are manufactured or purchased from third parties.
• With few applications that are relevant to retail industry the next few slides will describe the necessity of computer vision techniques in retail.

Computer Vision Applications

Shoplifting

• Using a series of images (2D) — Video analytics customers’ expressions are analysed to understand whether he/she is prone to shoplifting or not.
• The same could be extended with 3D images for specific use case of behavioural analysis.
• This can also be used to analyse the behavioural activities of employees in the store.

Augmented Reality

• Using beacon information along with imaging techniques will be essential to customers in stores to not only know about the options available but also offers, discounts and the path to find the next suitable product with the specification one is looking into.
• It can also be used as a path finder to find products with specific requirements.

Evolution of Analytics

• Data and Analytics have been the most commonly used words in the last decade or two. Every monetary-driven industry completely relies on Data and Analytics for its survival and growth.
• The upcoming slides will help in getting to know how the terms Data and Analytics are related and also how the industry has evolved using the practical use of these terms.

Analytics 1.0 — Need for Business Intelligence

• Customer (Business) and production processes (Transactions) were centralised into one huge repository like eCDW(Enterprise Consolidated Data Warehouse).
• Data surrounding eCDW was captured, transformed and queried using ETL & BI tools. The type of analytics exploited during this phase was mainly classified as Descriptive (what happened) and Diagnostic (why something happened).
• This era only addressed on what had happened in the past but did not give any insights of what can happen in the future with respect to the past data.

Analytics 2.0 — Big Data

• Aimed for a wider (if not better) approach towards attaining a sophisticated form of analytics.
• Information from external sources (clickstreams, social media, internet, public initiatives etc.) were taken into consideration — the term Big Data was thus coined.
• Companies’ expectation from employees — handle large volumes of data with a fast-processing engine. Unexpected response they received — emerging group of individuals or what is today better known as the “Open Source Community”. This was the hallmark of Analytics 2.0.
• Roles like Big-Data Engineers & Hadoop Administrators grew in the job-sector and became critical elements to every IT organisation.
• Tech-firms built new frameworks for ingesting, transforming and processing big-data and also integrated Predictive (what is likely to happen) analytics above it.

Analytics 3.0 — Data Enriched Offering

• Big data firms began investing in analytics to support customer-facing products, services, and features.
• They attracted viewers to their websites through better search algorithms, recommendations, suggestions for products to buy, and highly targeted ads, all driven by analytics rooted in enormous amounts of data.
• Now it’s not just tech-firms and online companies that can create products and services from analysis of data, it’s practically every firm in every industry.
• Tech-savvy giants forged ahead by making more money, a majority of other enterprises & non-tech firms suffered miserably at the expense of not-knowing about the data.
• A field of study Data Science was introduced which used scientific methods, exploratory processes, algorithms and systems to extract knowledge and insights from data in various forms. The next-generation of quantitative analysts were called data scientists.
• The tech-industry exploded with the benefits of implementing Data Science techniques and leveraged the full power of predictive & prescriptive (what action to take) analytics, i.e, eliminate a future problem or take full advantage of a promising trend.
• Companies began competing on analytics not only in the traditional sense — by improving internal business decisions — but also by creating more valuable products and services.

Analytics 4.0 — Automated Capabilities

• The cost & time for deploying Machine Learning/Deep Learning models wasn’t entirely affordable and necessitated a cheaper or faster approach in 3.0.
• The need for automation through intelligent systems finally arrived, and this idea (once deemed as beyond-reach) that loomed on the horizon is where Analytics 4.0 came into existence.
• Employing data-mining techniques and machine learning algorithms along with the existing descriptive-predictive-prescriptive analytics comes to full fruition in this era.

What Companies have foreseen?

• Juniper Research predicts retailers will spend $7.3 billion on AI by 2022, compared with the approximately $2 billion spent in 2018.
• Despite the increase in store closings in 2018, 90% of global sales were still made in-store in 2018. Furthermore, McKinsey reported that 80% of the purchases are expected to still happen in stores in 2020.
• 87% of consumers begin their shopping journey with digital, a jump from 71% in 2017 — Salesforce.
• In its 2018 study entitled How Analytics and Digital will drive Next Generation Retail Merchandising, McKinsey stated, “Winning decisions are increasingly driven by analytics more than instinct, experience, or merchant ‘art’”.
• By 2025, Bain & Co. forecasts that Millennials and Generation Z will represent 45% of the global personal luxury goods market.