18 Sep Champo Carpets is one of the largest carpet manufacturing companies based in India, with customers across the world, including some of the most reputed stores and catalog companies
Case Summary: Champo Carpets is one of the largest carpet manufacturing companies based in India, with customers across the world, including some of the most reputed stores and catalog companies. Champo Carpets is based out of Bhadohi, Uttar Pradesh, which is one of the most famous clusters of carpet weaving in India. This cluster is spread over 1,000 sq. km and comprises many villages and districts in and around it. The company is a vertically integrated manufacturer and exporter of carpets and floor coverings, with more than 52 years of existence. At the beginning of 2020, the company employed 1,500 people with a capacity to produce 200,000 pieces of carpets and floor coverings per month. As part of sales and marketing, Champo Carpets shared sample designs with its potential customers, based on which the customer placed an order. The sample design selection was done in various ways and the process itself is costly and elaborate. To capture industry trends, a team of the company visited various trade shows and events and sent samples to the client as per the latest fiber and color trends. However, their sample-to-order conversion ratio was low compared to the industry average. This had cost repercussions as well as lost opportunities. The company identified the cause as inaccurate targeting of products to their customers. It subsequently implemented an enterprise resource planning (ERP) application and has been capturing data at every point of production as well as sales. They believe this accumulated data can help target their products accurately to the right clients and design an appropriate recommender system.
Learning Objectives The primary objective of the case is to illustrate how machine learning algorithms can be used to manage business-to-business (B2B) sales. The learning objectives include the following:
Access attached the full case or article. After a critical review of the case, respond to the questions below.
For a better understanding of the issues related to the problem, knowledge of data visualization using Tableau, R, or Python programming will be useful.
1. With the help of data visualization, provide key insights using exploratory data analysis.
2. What kind of analytics and machine learning algorithms can be used by Champo Carpets to solve their problems and in general, for value creation?
3. Develop ML models to help identify features that contribute toward conversion (or non-conversion) of samples sent to customers.
4. Discuss the data strategy for building customer segmentation using clustering. What are the benefits Champo Carpets can expect from clustering?
5. Discuss clustering algorithms that can be used for segmenting Champo Carpets’ customers.6. Develop customer segmentation using K-means clustering. Discuss the optimal number of clusters, significant variables, and cluster characteristics.
7. Discuss the data strategy that can be used for building recommender system models.
8. Develop an association rule mining algorithm, which can be used for recommendation.
9. Build collaborative filtering techniques for recommender systems.
10. What will be your final recommendations to Champo Carpets?
Requirements:
- Your analysis will be considered complete, if it addresses each of the 9 components and subcomponents outlined above.
- Use of proper APA formatting and citations. Supporting evidence from outside resources should be used and those must be properly cited.
- Include your best critical thinking and analysis to arrive at your justification.
- Approach the assignment from the perspective of the senior executive leadership of the company.
Answer all questions thoroughly. Min 7-8 pages without reference page and main header page.
IMB 879
Priyank Goyal, Puneet Kapoor, Gyanesh Jain, and Ashish Sethia, students of the certificate program in Business Analytics and Intelligence and U Dinesh Kumar Professor of Decisions Sciences, prepared this case for classroom discussion. Champo Carpets cooperated and provided information to the Indian Institute of Management Bangalore in connection with the preparation of this case and it was reviewed and approved before publication by a company designate. No funding was sought or received from Champo Carpets for the development of this case. This case was also developed from available and permitted sources of information. This case is not intended to serve as an endorsement, source of primary data, or to show effective or inefficient handling of decision or business processes.
Copyright © 2021 by the Indian Institute of Management Bangalore. No part of the publication may be reproduced or transmitted in any form or by any means – electronic, mechanical, photocopying, recording, or otherwise (including internet) – without the permission of Indian Institute of Management Bangalore.
CHAMPO CARPETS: IMPROVING BUSINESS-TO- BUSINESS SALES USING MACHINE LEARNING
ALGORITHMS
PRIYANK GOYAL, PUNEET KAPOOR, GYANESH JAIN, ASHISH SETHIA AND U DINESH KUMAR
For the exclusive use of V. Shukla, 2022.
This document is authorized for use only by Vidhan Shukla in BADM 511 – Introduction Artificial Intelligence taught by Daniel Kanyam, University of the Cumberlands from Aug 2022 to Feb 2023.
Champo Carpets: Improving Business to Business Sales Using Machine Learning Algorithms
Dharm Kamboj, the IT advisor to Champo Carpets, was preparing for his weekly meeting with Rohil Mehrotra, the Managing Director of Champo Carpets. Kamboj anticipated that Mehrotra might want to discuss an important business problem Champo Carpets had been facing in the last few years. Due to his efforts, Champo Carpets was able to become one of the few carpet-weaving manufacturing units of Bhadohi to implement an enterprise resource planning (ERP) system for efficient collection of business data. The application had been running successfully since 2017, capturing data at every point of production.
As part of sales and marketing processes, Champo Carpets shared sample designs with its potential customers, based on which they placed orders. In many instances, the samples did not find favor with customers, who in turn sent different samples with similar designs.
Designing a new sample required a lot of effort and was a cost-intensive activity. It had to be accurate, addressing the needs of customers in terms of design and costing. Generally, the manufacturing cost of a carpet could be broken down into three components: raw material and its dyeing cost (30%), weaving cost (60%), and finishing cost (10%). Creating a new sample of the carpet typically doubled the weaving cost per unit area. Raw material and dyeing costs increased by 3 times and finishing cost expanded by 2.5 times. This escalation was due to the frequent set-up costs and higher cost of raw material and dyes when purchased in smaller quantities. In addition, the company also incurred a design cost, higher amount of wastage and labor costs every time a single sample was made. Hence, the conversion rate of samples into orders was important. In 2019, Champo Carpets had a conversion rate of about 20% against the prevalent 35% across the industry. This low conversion rate not only dented the overall profitability of the company, but also resulted in disruptions to the regular production schedule as they had to allocate substantial resources for sample production.
ABOUT THE INDUSTRY
In 2017, India accounted for close to 40% of the world’s carpet export market.1 The carpet industry was one of India’s most important industries; and in 2018, it provided employment to nearly two million rural Indians in carpet weaving and related activities such as dyeing, raw material trading, and so on. Most artisans and weavers were from the weaker section of the society. This trade provided extra and alternative employment to even farmers and other members of a rural household. The industry was scattered across the country, with Jammu & Kashmir, Haryana, Uttar Pradesh, Rajasthan, Gujarat, and Kerala being the major carpet-manufacturing states. Carpets were also a source of major export contribution (Exhibit 1).
ABOUT CHAMPO CARPETS
Champo Carpets2 was a manufacturer based out of Bhadohi, Uttar Pradesh, one of the most famous carpet- weaving clusters in India. Spread over 1000 square kilometers, the cluster comprised many villages and districts in and around Bhadohi. Champo Carpets was one of the largest carpet manufacturing companies in India, with customers spread across the world. Its clientele included several reputed stores and catalog companies, and its products were certified by leading organizations such as Wools of New Zealand and Carpet Export Promotion Council.
Champo Carpets was a vertically integrated manufacturer and exporter of carpets and floor coverings, with more than 52 years of experience and standing. The company had a large B2B clientele worldwide. In 2020, it employed 1,500 people and was capable of producing 200,000 pieces of carpets and floor coverings per month.
1 N B Das, R K Sharma, A Pandey and B G Narayanan, “The Existence of Carpet Industry in Bhadohi, India”, Trends in Textile Engineering & Fashion Technology”, 2018, 3(3), 323-332. 2 Youtube video: https://www.youtube.com/watch?v=k6NpdD9bTrk&t=6s&ab_channel=ChampoCarpets
Page 2 of 5
For the exclusive use of V. Shukla, 2022.
This document is authorized for use only by Vidhan Shukla in BADM 511 – Introduction Artificial Intelligence taught by Daniel Kanyam, University of the Cumberlands from Aug 2022 to Feb 2023.
Champo Carpets: Improving Business to Business Sales Using Machine Learning Algorithms
CARPET MANUFACTURING
The process of manufacturing a carpet started with design and passed through several steps [Exhibit 2]. Designing was driven by CAD (computer-aided design), which not only produced a visual output but also calculated how much raw material was required. This raw material was next procured and stored in a warehouse and then issued for dyeing to local dyers. The dyed yarn was re-stored in the warehouse and issued for the next process, which involved preparation for weaving or hand-tufting. The yarn was opened and wound on a suitable package. After the carpet was hand-tufted or woven, it was sent for various finishing processes. If hand-tufted, latex and fabric backing was applied at the back of the carpet to protect the tuft from fraying, and edges were secured through binding. Finally, the tufts or piles were cut to reveal the design, after which the carpet was ready for inspection and dispatch. A typical carpet took anywhere between three weeks and three months to manufacture, from design to dispatch.
Champo Carpets’ major products could be broadly classified into four categories, namely, hand-tufted carpets, hand knotted carpets, Kilims and Durries (Exhibit 3). Hand-tufted carpets had pile yarn simply anchored to the base material. These entailed the least production effort, and hence were the most popular. Hand-knotted carpets, by definition, were knotted by hand on the stretched warp and fastened with a weft. This required immense skill and production effort, and hence was the most expensive variety. Kilims were woolen flat weaves with slits, where the pattern began or ended in a different colored yarn. These were woven in tapestry style and were expensive. Dhurries or Durries were the Indian variants of Kilim. They were flat weaves traditionally used in most Indian homes.
CHAMPO CHALLENGE
It was important for Mehrotra and his team to address the issue of the low conversion rate of sample carpets sent by them to their customers. Champo Carpets, like any traditional carpet manufacturer, was sending sample designs to the clients as per demand and anticipated need. The sample design selection was done in various ways and the process itself was costly and elaborate. To capture industry trends, a team from the company visited various trade shows and events and sent samples to the client as per the latest fiber and color trends. The company also sent samples to the client based on the color and design attributes of their past purchases. While selecting samples, the company also considered the raw material availability in the inventory, and they would prefer to upsell those products, hence samples of these materials were pushed more. They also reproduced the swatches as sent by the client into samples.
Since sampling is a costly process as mentioned before, Champo is looking for a cost-efficient way of selecting appropriate sample designs that could generate maximum revenue for the organization. Mehrotra realized that the data accumulated from ERP operations could help them toward that end. Champo collected and stored several attributes related to carpets sold by them (Exhibit 4). Mehrotra believed that the carpet attributes could be used for creating customer segments, which in turn could be used for developing models such as classification to identify customer preferences and recommendation systems. They also understood that if customers could be presented with samples suited to their tastes, past preferences and trends, it would lead to better conversions.
The task seemed cut out for Kamboj – to identify the most important customers and the most important products and find a way to connect the two using suitable attributes from data and appropriate analytics models. This would help Champo Carpets recommend ideal sets of samples to customers and help increase the conversion rate.
Page 3 of 5
For the exclusive use of V. Shukla, 2022.
This document is authorized for use only by Vidhan Shukla in BADM 511 – Introduction Artificial Intelligence taught by Daniel Kanyam, University of the Cumberlands from Aug 2022 to Feb 2023.
Champo Carpets: Improving Business to Business Sales Using Machine Learning Algorithms
Exhibit 1 Carpet Industry Export
Source: Carpet Export Promotion Council of India (http://cepc.co.in/)
Exhibit 2 Carpet Manufacturing Process
Source: Champo Carpets
Exhibit 3 Major Products
Source: Champo Carpets
Page 4 of 5
For the exclusive use of V. Shukla, 2022.
This document is authorized for use only by Vidhan Shukla in BADM 511 – Introduction Artificial Intelligence taught by Daniel Kanyam, University of the Cumberlands from Aug 2022 to Feb 2023.
Champo Carpets: Improving Business to Business Sales Using Machine Learning Algorithms
Exhibit 4 List of Attributes in the Data
Feature Feature Type Description Order Type Binary Area-wise or piecewise Order Category Binary Order or sample Customer Code Categorical 45 customers (B2B customers) Country Name Categorical Originating country of customer Customer Order ID Identifier Order identification number Customer Order Data Date Date on which order was placed
Unit Name Categorical Unit of measurement (feet, meter, inch, piece and weight)
Quantity Required Numerical Number of units ordered Total Area Numerical Carpet area (in feet and meter)
Amount Numerical Revenue generated from the carpet (in USD)
Item Name Categorical Hand tufted, Durry, and so on Quality Name Categorical Quality of material used Design Name Categorical Code assigned to describe design Color Categorical Color of the carpet
Shape Name Categorical Rectangular (REC), square, round, octagon and oval
Total Area Order Total carpet area ordered (in feet)
Source: Champo Carpets
Page 5 of 5
For the exclusive use of V. Shukla, 2022.
This document is authorized for use only by Vidhan Shukla in BADM 511 – Introduction Artificial Intelligence taught by Daniel Kanyam, University of the Cumberlands from Aug 2022 to Feb 2023.
