Professional Documents
Culture Documents
Voice Support For E-Commerce Businesses in India
Voice Support For E-Commerce Businesses in India
Voice Support For E-Commerce Businesses in India
Businesses in India
October 20, 2021
Contents
4 Stories 17
4.1 Amazon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
4.2 Flipkart . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
5 IndiaSpeaks 18
ABSTRACT
Fuelled by tremendous reach of high speed internet, Indian e-commerce business is witness-
ing a rapid growth in Tier II+ cities. In India, 92% of population living in these areas is ready to
ride this e-commerce train. To instill trust, create long term engagement and to provide per-
sonalized experience, it is imperative that these customers are served in their preferred choice
of communication. This means that the e-commerce interfaces should be re-calibrated to in-
corporate local language interface, local language customer support, vernacular voice search
capabilities and chatbots. This document provides detailed insight into this problem with mar-
ket analysis and evidence from information aggregated from various sources.
urban India – reaching 323 million users (67 per cent of urban population) in 2020, digital adoption
continues to be propelled by rural India clocking a 13 per cent growth to 299 million internet users
(31 per cent of rural population) over the past year,” the report said. Furthermore, small towns
account for
6 warnings almost two out of five active internet users (AIU).
Smartphone Use Smartphone shipments in India increased by 23% YoY to reach 38 mil-
lion units in the first quarter of 2021, driven by new product launches and delayed demand
from 2020. Xiaomi led the Indian smartphone market with 26% shipping, followed by Sam-
sung (20%). India stood third (4.6 hours a day) on the list of average time spent by an average
user on smartphones, with Indonesia (5.2 hours a day) and Brazil (4.8 hours a day) taking the
top two spots worldwide.
With COVID-19-induced lockdowns, sale of smartphones through online channels increased
in 2020. During the festive season (Diwali), footfalls in physical stores picked up and offline
channels clocked a 5% YoY growth in the fourth quarter of 2020. According to a report by
Statista (Image above) 2 the number of smartphone users in India is expected to rise exponen-
tially in recent future owing to untapped potential in the rural parts of the country.
As said previously, internet penetration and widespread use of smartphones forms the foun-
dation for the e-commerce business to flourish. It is no surprise that, owing to the rise in inter-
net and smartphone usage the e-commerce sector shows optimistic signs in terms of growth
2
(Image Source: Statista https://www.statista.com/statistics/467163/
forecast-of-smartphone-users-in-india/ )
and customer acquisition and retention. Next, we will give a brief overview of the current state
and future projections for E-commerce business in India.
billion in 2020, driven by strong adoption of online services such as e-commerce and edtech in
the country.
The trend of increasing online sale is expected to continue every year from 2022, driven by
improved digital infrastructure, surging internet usage and rising acceptance of e-commerce.
Following the government curbs on social distancing and lockdown, there was a 39% rise in the
average time spent by an Indian user on a smartphone.
According to Bain & Company report, India’s social commerce gross merchandise value
(GMV) stood at US$ 2 billion in 2020. By 2025, it is expected to reach US$ 20 billion, with
a potentially monumental jump to US$ 70 billion by 2030, owing to high mobile usage. Driven
by beauty and personal care (BPC), India’s live commerce market is expected to reach a gross
merchandise value (GMV) of US$ 4-5 billion by 2025.
The Government of India’s policies and regulatory frameworks such as 100% Foreign Di-
rect Investment (FDI) in B2B E-commerce and 100% FDI under automatic route under the
marketplace model of B2C E-commerce are expected to further propel growth in the sector. As
per the new FDI policy, online entities through foreign investment cannot offer the products
which are sold by retailers in which they hold equity stake.
1.3.1 CORPORATE
Huge investments from global players such as Facebook, which is investing in Reliance Jio are
being recorded in the e-commerce market. Google also reported its first investment worth US$
4.5 billion in Jio Platforms. This deal was followed by the purchase of Future Group by Reliance
Retail, expanding the presence of the Ambani Group in the e-commerce space. Much of the
growth in the industry has been triggered by increasing internet and smartphone penetration.
Reliance is looking at using a hub and spoke model for JioMart, where it will look at managing
both the kirana chain as well as customer order fulfilment. Through this model, the Reliance
warehouse would act as a centralised hub and goods would travel outward to smaller locations
(kiranas) for distribution.
In July 2021, e-commerce conglomerate Amazon opened its first Digital Kendra in Surat,
Gujarat. Amazon’s Digital Kendras are physical resource centres for micro, small and medium-
sized businesses (MSMEs) to learn about the advantages of e- commerce. In May 2021, Flip-
kart strengthened its grocery infrastructure to cater to customer safety and demand across
India. In this quarter, it is planning to further expand its fulfilment center capacity for grocery
by over 8 lakh square feet across Delhi, Kolkata, Chennai, Coimbatore and Hyderabad.
Indian language conversational AI technology gathered momentum rather late. It is not a
secret that now these companies are betting high on Indian language technologies to provide
their users a rich experience. Large corporations like Amazon and Google have used regional
languages as the central theme in their recent marketing campaign to advertise their products.
Google is leading the innovation in developing Indian language technology with extensive re-
search and speech-to-text application for more than 9 Indian languages. Amazon’s flagship
voice system Alexa now supports Indian-English and Hindi; and further plans to support Tamil,
Telugu and Kannada to reach its potential customer base of 200 million who speak these lan-
guages. Recently Microsoft added two Indian languages namely Indian English and Hindi to its
neutral TTS service. Entertainment giant Netflix also recently rolled out its platform support
for Hindi. This includes video synopses, search and so on. Voice search in Indian languages and
many more innovative technologies are on the horizon.
1.3.2 GOVERNMENTS
As of June 25, 2021, the Government e-Marketplace (GeM) portal served 6.87 million orders
worth Rs.116,291 crore (US$ 15.67 billion) from 2.0 million registered sellers and service providers
for 52,651 government buyers. Through its Digital India campaign, the Government of India is
aiming to create a trillion-dollar online economy by 2025. It has formed a new steering com-
mittee that will look after the development of a government-based e-commerce platform. The
new committee, set up by the Commerce Ministry, will provide oversight on the policy for the
Open Network for Digital Commerce (ONDC), which is an e-commerce platform that the gov-
ernment is backing for the development. The ONDC will serve as the infrastructure for setting
up the final storefront, which will be similar to Flipkart and Amazon. In June 2021, Flipkart an-
nounced its partnership with the Telangana government to deploy drones to deliver medical
supplies in remote areas amid the COVID-19 pandemic.
Governments are also providing financial support to promote the research in conversa-
tional AI. Central government as well as many state governments have recently awarded project
grants, provided funds towards development of language technologies. Technology develop-
ment for Indian languages (TDIL) along with various IITs and private startups have been in-
volved in creating an ecosystem to develop machine translation, automatic speech recogni-
tion, text-to-speech, optical character recognition, handwriting recognition, etc., for Indian
languages. They are involved in developing open-source datasets as well as technologies to
promote the use of Indian languages in internet. Bahubhashak, National strategy for AI, India
AI are few of the recent projects and initiatives started by the central government.
1.3.3 STARTUPS
A comprehensive list of some of the recent startups backed by high-profile investors is given in
Table 1. Most of these startups initially started with Google’s voice and language services and
later developed their own technologies upon raising sufficient funds.
The general trend is supportive of massive growth opportunities in many business segments
which a traditional approach, limited to a few languages such as English, accented-English and
Hindi and a few applications such as chatbots, TTS and NLP, is inadequate to support. Most of
the startups in this space either provide value-added-service to businesses to reach new cus-
tomers or introduce automation in customer interactions. These demand-driven services are
targeted to ensure cost reduction for businesses and expand the reach but are largely inad-
equate in generating richer personalized user experience and lead generation leading to cus-
tomer satisfaction/retention. We are amongst the first startup that are committed to create
an ecosystem around speech technology where each components in the ecosystem interacts
with others for data, feedback and performance improvement.
• Untapped market with no loyalty base and preconceived biases; a level playing field for
emerging e-commerce businesses.
• The revenue share of Tier II+ cities has been growing and digital marketing efforts can be
targeted directly to these potential customers.
A 2019 Redseer survey 4 on 3000 consumers in 121 cities throughout India suggests that
there is a tremendous growth opportunities when it comes to online spending. Around 210
million internet users prefer digital content in vernacular. An Yourstory 5 article supports this
finding and further claims that vernacular content creation is the key to digital advertisement
to monetize the Tier II+ consumers by turning them to digital platforms. This user base can be
3
IBEF report https://www.ibef.org/industry/ecommerce.aspx
4
Redseer https://redseer.com/newsletters/vernacular-is-now-not-the-future-a-300-bn-opportunity-today/
5
yourstory https://yourstory.com/2020/09/advertising-vernacular-content-tier-two-three-markets/
amp
targeted for vernacular ad spend and online purchasing. Starting with the electronic, apparel
and beauty product, the e-commerce is ready to embark into areas such as online grocery shop-
ping, local product purchase, online food delivery and so on.
generation of consumers. Use of voice-assistant apps doubled to approximately 5–6 million monthly
users in 2021 (average until May). Web pages were translated to vernacular languages approximately
50% more in 2020 compared with 2019. Use of vernacular-language apps such as ShareChat and
Daily Hunt continued to accelerate through the pandemic.
Another report by Bain & Company 8 shows that the total number of monetizable vernac-
ular users will increase to around 340 million by 2023. Furthermore, thanks to multiple online
payment platforms such as PayTM, Google Pay etc. the online purchase has become hassle
free.
1 in 10 platform users adopt voice search, and 1 in 3 new e-retail users visit through a vernacular
platform interface. Voice and vernacular searches will increasingly become mainstream.
Bain and Company Report
There are many reasons why most of these users are currently reluctant to purchase online.
These reasons include lack of trust, complicated and unfamiliar interface and unfavourable
perceptions and lack of local language support. The same report points out that about 23%
of the users who did not purchase products online cite unavailability of local language support
as one of the reasons for their choice of doing so. The above mentioned factors are interrelated
8
Bain and Company https://realtynxt.com/wp-content/uploads/2020/12/
Future-Of-Commerce-Report-PDF-1.pdf
and affect each other. For instance, a voice search in vernacular language with local language
interface will be important in gaining the trust and with time the hesitancy will vanish. We be-
lieve that a friendlier interface and easy catalogue search will convert most of these users to
regular online purchase.
quality without worrying about advertisement, customer acquisition and reach and so on. The
vernacular support will surely help them increase their customer base and hence sales.
Automatic Speech Recognition (ASR) A good speech recognition system should be robust
to channel, speaker, age, gender, environment, noise and recording device variations present
in the voice.
• Increasing the amount of training data, typically in the order of thousands of hours.
• Improving the quality of the ground truth data (a.k.a the audio transcriptions).
• Diversifying the nature of the training data with respect to the speaker’s age, gender,
accent and dialect regions.
• Handling the morphological complexity and effective vocabulary building for the lan-
guages.
Natural language processing systems Natural language processing (NLP) systems enable ma-
chines the ability to read, analyze, understand and derive meaning from spoken or written form
of human languages. These systems process the input text from voice recognition system and
identifies the intent, context, name-entities and sentiments to automatically trigger actions.
Such systems can be built efficiently by training on large amount of annotated text data. Our
product offerings include NLP systems that can be used to build chatbot, appointment booking,
question-answering, sales lead generation, bill payments, request processing, etc. We will de-
velop the NLP modules taking into account the complex morphological structure of the Indian
languages.
Vernacular, Voice and Video will emerge as the game changers for the digital ecosystem over the
next few years.
Biswapriya Bhattacharjee, Executive Vice President, Insights Division, Kantar
4 STORIES
4.1 AMAZON
For the first time in a major online sale, customers experienced the Amazon Great Indian Fes-
tival in regional languages: Hindi, Kannada, Tamil, Telugu and Malayalam. Apart from this ver-
nacular push, Amazon is also powering up it’s app with Alexa, it’s homegrown Voice Assistant
that can invoke searches and specific actions in the app. In the run up to the Great Indian Festi-
val, Alexa answered over 100K requests from customers on the Amazon shopping app to help
navigate to their favourite stores such as the SMB Store, the Great Indian Bazaar, deals, gifting
store and the Fun Zone. On the Amazon Shopping app, Alexa received its highest single day
requests of over 1 million to guide customers to their product searches, best deals, bill pay-
ments, music and much more during Prime Exclusive Access. To get more users to engage and
try their Alexa Voice Assistant in India, Amazon even announced roping in Amitabh Bacchan
as the voice of Alexa in India! This illustrates the seriousness with which Voice is being consid-
ered inside Amazon for Indian market and also the kind of no-holds-barred investments that
are being made in the same.
4.2 FLIPKART
Introduced in Flipkart’s grocery store, ’Supermart’, the Voice Assistant will enable consumers
to discover and buy products using voice commands in multiple languages, starting with Hindi
and English.
The voice-first conversational Artificial Intelligence (AI) platform has been built by Flip-
kart’s in-house technology team with solutions for speech recognition, natural language un-
derstanding, machine translation, and text to speech for Indian languages, a statement said 9 .
9
Read more at https://economictimes.indiatimes.com/small-biz/startups/newsbuzz/
flipkarts-voice-assistant-to-help-people-shop-for-grocery/articleshow/76282759.cms?utm_
source=contentofinterest&utm_medium=text&utm_campaign=cppst
5 INDIASPEAKS
IndiaSpeaks Research Labs is a technology based B2B service provider that provides voice-
based solutions to enhance regional customer base in India. We provide voice-first mode of
interaction to the end users so that they can navigate through the mobile or computer appli-
cation and purchase products using voice commands. We provide these solutions for Indian
languages along with English so that semi-literate users from the huge untapped rural market
can be easily reached by the businesses.
Our focus lies in development and customization of Indian language voice-based chatbot
services for various businesses. These services can be effectively deployed in their customer
care call centres so as to avoid (i) long waiting time, (ii) frustrating experience, and (iii) overhead
task on the call centre executives.
Our solutions and services mainly utilize our indigenous voice recognition, text-to-speech
and natural language processing systems carefully designed for Indian languages. These solu-
tions will fill a major gap in the current product offerings of the businesses. Our services can
be easily adopted by businesses across several verticals: (i) E-commerce, (ii) BFSI, (iii) Travel &
hospitality, (iv) Media & entertainment, (v) Automobiles, (vi) Payment interfaces, etc.
Business objectives In order for our business venture to sustain and grow, the business needs
to work on the following objectives:
• Developing language technology tools for all the spoken languages of India.
• Building a large speech and text data corpus for Indian languages covering different ge-
ographical locations and dialects.
• Developing high-quality voice technology solutions by partnering with the research com-
munity.
• Automating the intents from content resource management (CRM) resources to make
voice-first interaction for the product so that ease of doing business is improved.
• Create future business plans for providing our voice services directly to the end users
rather through other businesses.
• Achieve a sales revenue of Rs. 30 crores and net profit of Rs. 10 crores by the end of 3rd
year.
Business solutions Our startup primarily offers the following two solutions to businesses
across various domains in all Indian languages:
(1) End-to-end “Voice-first” interface: Voice-first is a whole new interfacing technology which
uses voice recognition, TTS and NLP systems that allow users to interact with the applica-
tion by just speaking (which is the most natural form of communication) rather by typing. This
makes the technology more accessible to users across the board, not just senior adults and peo-
ple with disabilities. Since the voice-First technology uses AI at the back-end, the system will
remember the context according to the user’s prior conversations thus offering personalized
experience to them. We will try to provide an ideal voice-first interface to our clients’ busi-
nesses so that all the features and capabilities in their mobile/web application can be accessed
through voice commands.
(2) Automation of customer care call centres: In the traditional process, the users are re-
quested to contact the respective business customer care, business process outsourcing units
or interactive voice response systems through telephone to raise complaints or queries. The
main concerns of this approach are (i) choosing the right option in IVR by listening to all the
available options, (ii) longer waiting time for the calls to be answered, (iii) overloaded calls to
the customer care executives, etc. Automation of call centres (ACC) is designed to solve all
the pain points discussed above with the help of state-of-art core technologies such as voice
recognition, TTS and NLP by integrating these components based on a predefined workflow.
The end customers can interact with these automated systems and resolve their issues with
lesser waiting time and get better experience, thus making the entire process more efficient.
Further, it makes the system and user interaction more realistic and natural.