Contribution of BI on Flipkart

Amrit Mohanty

Introduction

One of the best Indian e-commerce company founded in 2007 by Sachin Banshal and Binny Banshal, and both are belong to IIT Delhi. Both of them worked for Amazon.com and left to create their new company. Their new company was incorporated in October 2007. The first product they sold was a book Leaving Microsoft to Change the World. Our Organization employs more than 33000 people. It adopts different mode of payments like Cash on Delivery, Credit or Debit Card transactions, net banking, Card Swipe on delivery, e-gift voucher etc. The most popular model was Cash on Delivery. Company was also came to light because of its cash on delivery model which proven a great significance since credit card and net banking penetration is very low in India. Flipkart has generated revenue of $350 million. It has started making 2–3% product margin after expanding its product catalogue. Company has an intention to boost its operating margins to 8%-10% in the coming year. It has heavily invested in infrastructure especially in backend logistics as a business differentiator.

In the organization 37% of equity has been controlled by Bansals, Accel Partner and Tiger Global together control 48% and the management holds the remaining 15%. In the May 2018 the Wall mart Inc picked up 77% stake of the company for $16 billion, and it is the country’s largest acquisition and world’s biggest purchase of an ecommerce company. The deal includes $2 billion for fresh investment as WalMart tries to take on rival Amazon’s global expansion, pegging the value of company at $22 billion. But the Company brand will remain distinct from that of WalMart.

Key Corporate Events

1. In the 2007 Company was founded by Sachin Bansal and Binny Bansal
2. In the year 2008 Company was able to receive 100 orders per day.
3. In 2010 Company acquired the Bangalore based social book discovery service weRed from Lulu.com
4. In February 2012 the company started its DRM-free online music store Flyte and it shut down in June 2013
5. In May 2012 Company acquired Letsbuy an online electronics retailer.
6. In May 2014 Myntra was acquired by the company for 20 billion.
7. In February 2014 Company has partnered with Motorola Mobility.
8. On 6th October 2014, the company held a major sale since it promoted the day as “Big Billion Day”.
9. In March 2015 the company blocked access to its website on mobile devices and started its mobile app.
10. In April 2015 the company acquired Appiterate a Delhi based mobile marketing.
11. In October 2015 the company again reprised its big billion day event and this it was multi day event.
12. In December 2015 the company purchased a minority stake of MapmyIndia.
13. In 2016 the Company acquired Jabong.com and UPI mobile payments.
14. In April 2017 e Bay announced that it would sell its subsidiary to the Company

Business Model

Website — Company’s website is the B2C shopping Portal for Indian customer. The model is Portal > List sellers who want to sell their products > Get customers browsing through the products > Create appropriate discount > Customer shops for the desired products > Seller / Company ships the product to the customer > Product accepted and didn’t return back > seller gets his agreed price of the product minus the commission charged by Company. Here the core of the Model is “x% commission on the total sale value given to the seller”.

Sales can happens multiple channels

1. Via website
2. Via Web-app
3. Through Mobile App (Android)
4. Via Tele Sales (Customer Calling and Placing Order)
5. Via Affiliate networks (Bloggers, Coupon Websites)
6. Social Buy
7. Business Model from Listing and Convenience Fee — Company charging a listing fee for the seller who want to sell on its platform which as a turn adds up to the total revenue of the company.
8. PayZippy — Payzippy is a payment gateway is like other payment gateway which provides services to the company. In Payzippy they charge a transaction processing amount to every transaction and that goes through the payment gateway infrastructure.
9. Business Model of Logistics — At first Ekart logistics was used to deliver the ordered products and later it got spun off as a different entity and it ships for all other platforms. Logistic company charges sellers to deliver the goods that user have ordered. Charges are depending on within city or inter-city or size of the package.

Business Model in Digital Media

1) Co Branded Banner Opportunities on Home Page: In the company website you can see a slider in the homepage presents opportunities for sellers, brands and product launchers to represent themselves for a million views and this comes with large fees. The company earns a lot of revenue from the banner.

2) Co-Advertised Physical Product across Publications: The large ads in news paper and in magazine shared with the brand they are advertising.

3) Targeted Search Results: The moment when search something, Company’s algorithm decides which sellers products come at the top. This space also sold for additional revenue.

4) Business Model from Wholesale — Company started cash and carry which dedicated to wholesale of goods to its seller base. It is similar to other Metro cash and carry which caters to retailers and wholesalers. The revenue which has generated is exactly similar to the web portal revenue that company gets from selling goods in its platform.

5) Business Model from Product launch — Since the company has enormous user base many companies uses its platform for their product launch. This gave company revenue in terms number of product sola and also the advertising revenue.

6) Business Model from Myntra — Company was having its own fashion category but the revenues that achieved through Myntra were comparatively high. The revenue from Myntra was also added to the company total earning.

BI Overview of the Company

Company is having a big data group for engineering, product and operations. Data is the core of everything. It generally helps to understand the customers deeply, learn what to build measure impact of our work and drives all the decisions.

Company’s big data platform supports one of the largest ecommerce data sets in the world with 100M+ users. Billions of events generated by large number of users each day through their visits, search queries, page views, add to carts, wish lists, offers, payments, purchases etc.

Billions of events absorbed by data platform each day and it performs large data computation on this data and it generates thousands of business metrics and reports to drive daily decisions. Data platform also provides hundred of steaming metrics in real time to manage sales events like Big Billion Day and flash Sales.

The machine learning platform enables data scientists to build, operate and deploy models at scale in production. Machine learning also used to solve complex ecommerce problems.

Hundreds of insights are generated by insight platforms about the users each day. Insights are like predicted gender, brand affinity, price affinity, income range etc. These insights are then used to personalize user experience such as targeting the right users and personalizing their search results are based on user affinity.

Finally the Trust and Safety (TnS) platform use data and insight to find out the model risk for each commerce entity, fraud patterns and also it uses risk scores for each transaction to take preventive action to reduce fraud. By using TnS a significant costs of the company has been saved by reducing the fraudulent activities in different segments like buyer fund claims, seller fund claims, rating and review fraud, fake product listing, pricing abuse etc.

Function of Artificial Intelligence and Machine Learning and its Future

From the moment when you log into the web portal till you receive your package at almost every step of the user journey there are many machines algorithm at work like it gives personalized recommendations that figure what products to buy, Shipment details will be known by customer service bots, Anti fraud algorithm says whether the order is genuine or not, routing the package from a warehouse to the city, and assigning the shipment to a field agent to deliver it home.

No such self driven trucks, drones and robots are present at the frontier of AI. It is more focused on the problems such as personalization, product discovery, address understanding, search, and fraud detection. BI system in the company generally used to reduce cost and friction in the delivery supply chain.

Company had an analytics team from almost day one and importance of data science was understood from the very beginning. ‘AI for India’ initiative was lead by co founder and Chairman Sachin Bansal. During the announcement in December Bansal outlined the company has an intention to solve complex problems unique to India.

A Data Moat

People who can conduct research in machine learning, who have formal education in one of those areas had proven greater experience in those areas. Around 50 people are now working as scientist. The company now plans to double the headcount. Here the data acknowledges the company is having limited talent supply in comparison to Alibaba which is having 90 engineers.

The company has encouraged its engineers to take the machine learning course that Google has recently launched. Here the goal is to Up Skill everyone.

Company is now planning to launch India specific challenges. Company is looking to rope in the data science community at large through data challenges. The challenge is similar to Netflix challenge where the algorithm predicts user rating for films.

Due to the high data generation by the company there are certain complexity involved in the system.AI and machine learning somehow able to reduce the complexity. Company generally generates over 10 terabytes of data, which grows fivefold on a day during event like big billion Days sales.

In order to keep the data intact the company has invested a lot in infrastructure. Company data platform lets its business and product teams to glean insights form a self-service fashion. ML platform allows the data scientist to host a model, check the results, feed data into it, and tweak the model.

Key AI/ML use cases

The company has more than 20 projects that deploy machine learning. Some most used machine learning’s techniques are Decision Trees, Logistic Regression, Support Vector Machines and deep leanings.

The AI’s are used generally to prevent frauds, customer support, logistics and warehousing, estimating product popularity, intent modeling, personalization, conversational search, forecasting, discovery , image speech and text processing etc.

The search itself has a higher area of research in the company. Suppose a person search for Iron Table then at that time the search algorithm first links the ironing boards category and provides link to foldable iron tables on the third search.

The company also uses MI Algorithms to group addresses that have been ordered in a subarea fashion. MI also separated compound words and eliminates the bogus orders.

Company is working on a pilot projects in metro cities to optimize the routes of the delivery agents. It is also creating systems to automatically sort packages on their final routes by area. By using this technique fewer hour activities has been reduced to fewer minutes.

Pin code prediction is the other feature of AI where it helps to identify the incorrect pin codes in comparison to the address. Many times due to the incorrect PIN code the order get delayed and it is avoided by the above prediction techniques.

By using Geocoder project the latitude and longitude of the address can be easily known. Geocoder project is in partnership with MapmyIndia, which uses deep learning techniques to identify the latitude and longitude of a given project.

Predicting and forecasting of product at individual product level, category level and what should be the right price is the another area of machine learning which also has been applied.

The company uses computer vision to understand exact product attributes from images and check if it is accurate. Because literally there are 100 of products are there in the market and we also get these many numbers of products putting human resource is also too expensive so deep learning helps to identify no vulgar or objectionable content being uploaded

Cutting Fraud and Boosting Revenue

Preventing transaction fraud is another key area of application of machine learning for the company. It has a production fraud models that prevent reseller fraud.

A recent application of machine learning was helpful to find out RPI that is Revenue per Impression. It’s a measure that helps the company to figure out how much revenue it would make if it showed a particular product to you.

For identification of product category machine learning plays a measure role. For example the machine learning identifies the product was in the shoe category, black color and the brand was Nike, it had stripes. Generally the AI regress the numbers based on past data and trying to generalize from it, using logistic regression and decision trees.

By using the review analyzer the company analyzes million of reviews and demonstrated how company uses insights from million of reviews.

Speech recognition and support for Indian languages is being seen as another critical area for the company. Here the idea is to build an interface that should accommodate the people who are semi literate. The company currently supports English and smattering of Hindi. The speech recognition system is in an experimental stage and it is not implemented fully.

In the areas of drones or robotics the company is interested in partnering or acquiring certain companies.

The NPS value of the platform has increased due to the performance of AI and machine learning. NPS is the short net promoter score.

Company Leverages Business Development Activities through Machine Learning

Company uses computerized Maintenance Management Software (CMMS) that enables to implement solution and gets result quickly. Where company Leverages BDA:

1. By using the data and analytics company predict region wise demand, so stocking of inventory was done accordingly.
2. Certain capabilities have been automated in warehouse and fulfillment center that is leading to faster dispatch of accuracy.
3. Creating algorithm accurately for customer expectation by calculating user and production location.
4. Data analytics helps to predict demand for a particular region so that we can plan our inventory according to that.
5. Company has been diligently pursuing mobile-first strategy since 2014 and had given a huge investment in mobile strategy.

Myntra adopted Analytics

Company owned Myntra had adopted several technologies to expand its customer base. Myntra platform includes all backend software programs that identify the buying habits of the customers. This would really helpful in planning the inventory and supply chain. For A/B testing machine learning solutions are deployed to build new features focused on merchandizing and personalization. Merchandising helps user on mobile and website platforms.

Partnership with Microsoft

Company partnered with Microsoft to facilitate a better consumer service. In the first step company has adopted Microsoft Azure for its public cloud platform. Company is planning to leverage AI, Machine Learning and analytics and analytic capabilities that present in azure such as Crotana intelligence suit and power BI to optimize the data, marketing, advertising and customer service.

Project Mira

AI project of the company was Mira. Company studied its customers for a long time after that it has been concluded that a large chunk of their returns could have been prevented by asking one or two simple questions before their purchase.

The main goal of Mira is to gain experience of being a sales associate through artificial intelligence. When users are searching certain products then certain conversations and patterns are captured by AI system and through digital channel which would further helpful in predictions and suggestions. It is a real world application of conversational technology.

At the time when a person searched for an air conditioner because of Mira, company asks buyers certain questions like what kind of AC they want, the tonnage, brand, room size etc. It is the starting stage of helping a customer to find the exact product and it is helpful to navigate to the exact product they want.

Sillicon Valley Ambitions

The company is also expanding its presence to the Sillicon Valley in US. Company is focusing AI based products by making use of world class research facilities like F7 Labs, Company’s US based research firm in Palo Alto.

The company had a number of AI problems more interesting to Indian context. In case color recognition all the standard colors are there but in Indian context gulaabi is a separate one. To difference between kurta and kurti is also difficult for AI system. Now the F7 lab is used to resolve these kinds of issues. Also F7 lab is using Natural language processing to understand certain colloquial English which is basically used in India.

Apart from that the company is now emphasizing the Image analysis and deep Learning Analysis.

Better Ads for Consumer

Company has exclusively used the AI to crack the user segmentation. AI system generally gathers and processes multiple patterns across various categories from the same user. It helps to find to strengthen the quality data and its analysis. AI eradicates the anomalies by analyzing the historical pattern and identifies a pattern which helps to find out the brand affinity, choice categories, pocket size and frequency of purchases of users. AI also analyzes an Individual user, their habits, previous activities, likes and dislikes etc and serve ads at the individual levels. AI also assists in targeting ads on the E-Commerce and 3rd party platforms; specifically based on times of day, calendar events etc. Ads of the company have highest CTR in industry. Machine Learning uses multiple micro streams of data and conversion of funnel data to optimize the ads.

AI-ML Techniques to Identify Consumer Behavior

Now a day the company is betting on AI and Machine Learning to know the consumer behavior.

The company analyzes consumers who visit its homepage across channels like mobile app and web and desktop. Machine learning techniques predict the content of customer is most likely going to click. Based upon the past history this technique is also used to find the age and gender.

The AI and ML based techniques are also helping the firm to understand what the consumers actually meant when they search a product then the technique through up relevant result immediately.

For example, in general the consumer from metro may knows that Wrogn (Jeans) is the Virat Kohli fashion brand for jeans. But Consumers from tier-2 and tier-3 cities search for brand denim trousers as “Virat Kohli Jeans” since demand is higher in those cities. AI-ML techniques help to find out such kind of result if there would be such kind of search.

Data Protection

Company takes rigorous measures to protect the personal and financial Data. Company looks data as an aggregate perspective and garbles the stored data.

Centralized Procurement Process and Analytics

The company is looking to cut costs by merging 30% of the departments, keeping hiring minimum and centralizing purchases to steer the company towards profitability.

The Engg department, logistic Ekart and the advertising and e commerce unit will be unified into one part. Independent categories such as large appliances, furniture, home décor, kitchen and furnishing are also clubbed into one — home. Similarly sales and marketing run as a single tem after the merger. This collaboration also reduces the usage of manpower and as turn the cost reduced.

The company is also merging procurement for all functions such as media buying, IT, promotions, supplies and warehousing based upon the assumption that it would be easier to negotiate deals with centrally rather than multiple teams.

By merging the entire department their database are also merged accordingly and analysis are also being conducted in a centralized fashion which in turn reduces the requirement of IT resources.

The company is also looking at investment in automation at its warehouses to reduce costs and need for human Intervention.

Company Improves Inventory Utilization with QLIKVIEW

Qlikview is the replacement for open source BI; the company invested its first license business discovery software solution. Company was experiencing complex flow from different units. A smart system is needed to collate the data and display a meaningful and actionable analysis for decision makers.

Qlikview is the superb tool for managing inventory by optimizing stock levels and lowers costs associated with excess stock. By using Qlikview day in day to day operation the company has improved the inventory utilization by 5 percent. Qlikview application analyzes around 25 million rows of data, and it is increasing with the company’s growth. It is easier for the staffs to analyze the data since there is a direct access to company data.

Qlikview has changed the way company does business. Basically it is helping company to become more agile and responsive to the customers. By using the Qlikview the data driven decisions are also established. All the teams are no longer dependent for any kind of report analysis. By using Qlikview the knowledge shared across teams easily. Due to the user friendly and intuitive dash boards it is well accepted across teams. The Qlikview business application works with existing data sources like MySql, Microsoft Excel, Spreadsheet software etc. Now the company is planning to extend Qlikview use to tablets and smart phones.

Benefits of Qlikview

1. Provides transparent and up-to-date information for analysis.
2. Provides Embedded data driven decision to company.
3. By using the Qlikview inventory utilization has improved.
4. Delivered the first application for sales in just six weeks.

CRM at company

Company is fast growing with several employees, partners, suppliers and distributors. The information entering into the system is too huge to gather and make sense without any proper technology integration such as ERP and CRM. CRM enables the efficient gathering and also pooling of customer orders, sales information, web traffic, delivery and shipping details and analytics in real time. It also helps to control the scaling by allowing addition and compression of resources according to the demand. It also allows automating the process and as a result it also helped to reduce the labor cost.

Company reaches out to its customers through several channels such as email, customer care, chat support etc. CRM enables the integration. All channels at their back end provide quality services and consumer experiences.

Consumer Management Orientation

According to company philosophy huge discount is not the only factor by which you can sustain the business and customer satisfaction. Company has built on the core foundation of reliability, quickness, credibility, verity and quality. By providing good customer service the company enjoys high levels of customer satisfaction and generating repeat business.

To develop the customer relationship AI and Machine plays a measure role. It provides a good user interface, easy navigation to the searched items, and provides more variability of items by analyzing the shopping pattern.

Sales Force Automation

The regular sales functions have been supported by SFA and tools that are employed by users to perform administrative and other repetitive tasks. The manual processes are now converted into automated process, which helps sales representative to work in a more efficient manner. Applications such as quarterly sales reports and calendaring are now being automated by certain tools.

SFA tools assist sales representative for formulating a professional sales counter. Sales representative can be remaining in contact with customers via e mail and cell phones and hence the travelling hours get reduced. Sales agent can also receive and manage orders from customer in an easy and timely fashion. Routing and calendaring helps to increase the production and also helps to reduce the downtime.

Service Automation

Service automation is the tool for enabling business to automotive value creation between consumers and business (B2C), between business (B2B) and within business.

The goal is for each service instance the ideal configuration of resources necessary to deliver the service instance. The service instances are composed and executed dynamically and in real time.

The following questions are addressed in service automation

1. How a semantic model of business developed
2. How semantically developed business activities and services be developed
3. To what extent the plan, processes and workflows for business generated automatically
4. How businesses coordinate their activities more quickly and rapidly.

Analytical CRM

Customer Knowledge Discovery

Company has a superb recommendation system to make product discovery faster and quicker. The recommendation system used by the company predicts user’s intent and it also helps to connect them with their necessary products in an automated manner.

Ex: recommendation to buy a Motorola phone while surfing Nokia Phone.

Recommendation System = Data + Algorithm

Formula:

Recommendation Score = (i*i) / (n1*n2)

Where i = count of occurrences of items i1, i2 together. n1 = # of occurrences of the item i1 alone.

N2 = # of occurrences of the item i2 alone.

Generally it uses collaborative and content based filtering methods to generate recommendation.

Customer demographic analysis and customer behavior modeling

Company’s customers fall in to the categories of youth, technology enthusiasts, avid book readers and online shoppers. The company has carefully demarcated and segmented each of the categories and modeled the consumer behavior to understand and predict their preferences.

The categories are bought history, compare history, browse history, bought items, rated products, cart additions etc.

LAMP Architecture

LAMP stands for Linux, Apache, MySql and PHP is the common way of designing modern day website. This architecture provides a cost effective way of building websites. LAMP components also help in the backend scaling of website and integration with ERP capabilities. The company is the best example in successfully implementing the LAMP architecture.

There are four essential technology teams within the organization

1. Website development and maintenance
2. Data Analytics/Business Analytics
3. ERP
4. Platform management

Analytic Team

The analytics team not only looks after the data mining and consumer buying pattern but also the team concerned about vendor management, supply chain management and logistics management.

The team engages in managing information system and analyzing data flow to streamline the operations of the company. Some achievements of the team are “2-day delivery process”, “money back guarantee, “route optimization”, “hassle free just in time shipping” etc. The team indulges itself in data warehousing, data mining and predictive analytics.

Analytical System Used in Supply Chain

1. By using AI and Machine Learning, sales are predicted for different regions and accordingly the allocation of inventory is conducted.
2. The company stores the detail of each transaction by using MIS system.
3. Also by storing data and using tracking system a order is easily tracked and also the system updates about the status of shipment to the customer via message, email and through website.

Intelligence and Analytics in Ware House Management System

Inward Processing

Physical in warding: The area where physical delivery of goods from suppliers to warehouse are taken.Quality Check + Scan:: After receiving the goods it gores though a quality check and after that it scanned through an electronic device and subsequently the electronic entry to the records into the IT system occurs.

Pre-packing of products: In this stage initial packing of each product done.

Storage Management

Put-List Generation: Input of all the products done on IT systems and shelves generated which facilitates the placement of the product.

Order pending check: After system gets the input of the incoming products, the system then checks whether there is incoming products are pending or not. If the product is found pending then it is sent directly to the final packaging for outward processing.

Physical placement of Shelves: Based on the put list products are placed to the respective shelves.

Closing put-list: After the product placement Put-list is updated with actual placement information in IT system.

Outward Processing

Pick-list generation: Pick List is generated based on the order to be delivered.

Pick-up from shelves: The respective products are picked up from the shelves as per IT system entries and gathered together to move towards final packaging area.

Final Packaging: The picked up products are then packed in branded boxes and packaging is conducted according to the category of products.

Placement in respective hub’s bag: After the final packaging the product is placed in a specific bag and it is delivered to the destination.

Here the scanning and some other process are done manually and there is certain scope for automation.

Company uses its own ERP system to track the details of all transactions.

To do the payment the company had launched its payment system called PayZippy which provides a hassle free and safe payment system for customer.

Order Fulfillment

Orders of customers are fulfilled via inventory or JIT procurement system.

After the order is placed and it gets approved then the inventory check done at local warehouse. If it was not found then the order goes to the nearest warehouse. In the case if the product is not found in the inventory at that time it is forwarded for JIT procurement for local vendor. If the product is also no present here then it is forwarded to central procurement team. After procuring the product is packaged and delivered to the customer. Though here most of the process are automated but certain modification also required.

Status of the order has been updated via email, message and through website to the customer. Here the process is automated and the auto generated alert system has been integrated with ERP system.