Mobile Vaani: Reducing Exclusions

The digitally powered ID systems that form the basis of delivery of public services (including welfare) in India have been prone to failure in recent times. Even with a well-designed digitized ID system such as the Aadhaar in India, there are plenty of issues faced by citizens that are unaddressed. Our solution is to build an inclusive platform to help vulnerable citizens in rural areas report complaints and failures in accessing welfare entitlements which are increasingly linked to a digital ID system. This will also help create a complaints database that offers a way to identify the reasons for failure and offers a simple yet unique way to measure the efficiency of welfare delivery. Our solution provides an easy way for citizens to register complaints and can act as a pivotal feedback loop for local administrations to understand where systems are failing, and take accountability to address the gaps.

This is a voice-mail left by a farmer in the Shivpuri district in Madhya Pradesh, a state in central India. The farmer details his failed attempt to enroll himself in a cash transfer scheme for vulnerable farmers. After eight months and spending a sum of Rs. 1,000, he was not enrolled and had been given no clear answer as to why. The inability to track one’s application, despite digitization of records, along with the time and money spent in getting “identified” by the state for welfare showcase the lack of any feedback mechanisms in place to address the grievances of citizens. During the COVID-19 crisis, studies show high transaction failure rates of 39%. There are various reasons ranging from biometric mismatches to issues with matching beneficiary details on the ID system with details in the databases held by different welfare departments and banks. The biggest problem is that the citizen is often unaware of where the issue that leading to the failure thereby making it harder to get it addresses. Additionally, this process of welfare delivery through banking channels and private agents creates an ambiguous environment for the citizen to identify which institution he or she must approach for redressal. 

Gram Vaani engages with rural, informal sector households across the country through an intelligent IVR (interactive voice response) system that allows people to call into a number and leave a message about their community or listen to messages left by others. The Mobile Vaani platform with its flagship deployment in Jharkhand, Bihar, Madhya Pradesh, Uttar Pradesh, Delhi NCR, and Tamil Nadu, now has over 100,000 monthly unique users that call over 10,000 times per day to report complaints or distress and get guidance. Our solution will address the challenge of exclusion from welfare by:

1. Creating an inclusive, easy, and effective tool for aggregating complaints - A voice-mail based platform provides a simple way for citizens to register complaints as supposed to filling and submitting physical forms at government offices or through inherently opaque and exclusionary web-based platforms.

2. Tracking the process of redressal at every node -  With a well-defined complaint coding system, these are routed through a network of trained local volunteers to begin investigating the complaint and documenting the issues. 

3. Creating SOPs (standard operating procedures) for resolution of different types of exclusions - Technological improvements and improvisation of operational processes will improve welfare delivery through linked ID systems. 

We propose a solution that will help vulnerable and marginalized groups to avail social protection measures that are linked through a digital ID system, including the most basic of welfare services such as cash transfers, food rations, and rural employment. Our solution of creating a simple, safe and inclusive voice-mail based platform to register grievances in availing welfare entitlements will help us understand the nature and the extent of issues related to ID systems and other exclusionary factors that deny citizens their basic entitlements. We have already piloted this solution and received more than 7,000 grievances during the COVID-19 lockdown. 

The use of voice through IVR (interactive voice response) systems helps jump literacy barriers and the digital divide, making the solution accessible to even those who do not have smartphones or cannot use the Internet. Users simply give a missed-call to our platform, automatically receive a call back, and over this call they can listen to audio messages or record their own messages. Messages thus recorded by them are processed by a moderation team, and in the case of grievances, are passed on to a network of community volunteers. The volunteers help take these problems up for redressal with the local administration. 

This platform called Mobile Vaani, was able to deliver a very strong response during the COVID-19 lockdown when several relief measures were announced by the government but many people faced exclusion due to a variety of failures, a significant component of which were failures due to linkage problems with Aadhaar, the digital ID system in use in India. India has achieved a near-universal coverage of Aadhaar and therefore the key problems are not so much about registering for a new digital ID, but to do with use of this digital ID to access different social protection schemes. Our solution provides a way for people to register their complaints, get assistance for redressal, and in the process build a detailed understanding of the gaps in the technology architecture and procedures, to identify SOPs (Standard Operating Procedures) to bridge these gaps in a systematic manner and reduce exclusions. 

During the COVID-19 lockdown, our solution was used extensively in our areas of work in approximately 25 districts spread across the states of Bihar, Jharkhand, Madhya Pradesh, Uttar Pradesh, Delhi NCR, and Tamil Nadu. Over 20,000 voice recordings were contributed by Mobile Vaani users, more than 7,000 of which were related to problems faced by people in availing relief services. A policy brief codes this content to represent the extent of problems faced by people, and a consultative campaign #NotStatusQuo describes the specific reasons why people faced these exclusions. More than 800 of these reported issues were also resolved and acknowledged by the beneficiaries, who were able to access the relief measures. Our live dashboard reports updated numbers. The Mobile Vaani solution has also been extensively used in other contexts, such as for awareness and behavior change communication in health and nutrition, hyperlocal news and community media for development, cultural expression for community building, and data collection. These impact pathways have been extensively documented and reported in research papers by our team. 

Reflections from Practical Experiences of Managing Participatory Media Platforms for Development - A. Seth, A. Gupta, A. Moitra, D. Kumar, D. Chakraborty, L. Enoch, O. Ruthven, P. Panjal, R.A. Siddiqi, R. Singh, S. Chatterjee, S. Saini, S. Ahmad, and V. Sai Pratap.
ICTD, 2020.

Design Lessons from Creating a Mobile-based Community Media Platform in Rural India - A. Moitra, V. Das, A. Kumar, Gram Vaani, and A. Seth.

ICTD 2016.

Activity: Aggregating complaints: Voicemail based system for complaints collection

Using the existing network developed with the Gram Vaani, the solution will deploy a IVR system where citizens can call in and leave a simple voice-mail system with their complaints. With a well-defined complaint coding system, these complaints will be cataloged based on the location and the nature of the complaint. 

Output: Mapping the process of redressal for each type of complaint

The cataloged complaints are distributed to a network of local volunteers to begin investigating the complaint and documenting the issues. This could be, for example, related to the incorrect linking of Aadhaar with the citizens bank accounts or failure in applications for the required identification to access Public Distribution System (PDS).  The volunteers will document the protocols prescribed for resolving this category of complaints. 

Short-term Outcome: Assisted redressal for vulnerable citizens

The complaints identified and taken up for assisted redressal by the volunteers, can be given public visibility through platforms such as Whatsapp on which administrative officers at different levels can oversee the progress while keeping the citizens anonymous.

Senior officials need not track each and every case, but an aggregated set of complaints arising from similar issues can help them guide their junior officials to improve processes. Highlighting even a few selective cases might translate more widely to improve the redressal process in the administrative region. 

Long-term Outcome: Localized IVR systems and the creation of a public complaints repository-

The creation of a complaints database (anonymized) could be used as a tool to improve the transparency and accountability of the ID system in each administrative unit. Through the collective learning from this exercise, each local administration can be provided with the IVR technology as well as a well-defined set of operating protocols that address commonly arising complaints and issues. International experiences, such as that of the Consumer Financial Protection Bureau (CFPB) in the US, show us that the creation of public available complaints repositories helps inform the service providers of the difficulties faced by users and to empower them by providing a transparent way to avail redress.

Issues with ID documents for vulnerable groups are often highlighted only when they are used to avail services such as welfare schemes and financial services. UIDAI’s provisions for updating simple details online are often inaccessible to low-income individuals. While the government has promoted a cadre of private entrepreneurs in rural areas to help with digitally updating ID documents and other services, these service providers are often unable to resolve certain issues. The citizens and the agents are often unable to decipher error codes thrown up by the digital identity systems and are often not able to avail redress for these errors. At times the errors may occur asynchronously and not even be reported to the operators or citizens. With more effective communication of the error codes, with interpretation of these errors linked to resolution pathways, standard operating procedures could be developed to guide and assist beneficiaries on next steps. We believe it will be appropriate to identify integration points associated with these error codes, to link into the Mobile Vaani solution we have proposed. By identifying beneficiaries through their digital ID to understand the specific errors they faced, and connecting it with the voice-based solution to both provide updates to citizens as well as enable them to register grievances through voice-recordings, will enable an integration of our solution into the digital ID systems. Additionally, with well documented cases of failures and processes for redress, our solution will provide citizens and agents a mechanism to redress them.

Our solutions using IVR systems have been used by over 150 partners over the last five years, to reach out to 2M+ users. An additional half a million users interacted with our system over just a period of 100 days overlapping with the COVID-19 national lockdown in India. Users spend on average 8 minutes on our platform, browsing and listening to audios, and recording their voice messages that can be shared with other users. Several other organizations using IVR systems in the development sector also include Avaaz De, CGNet Swara, 99dots, Mobile Kunji and Kilkari, etc, demonstrating the effectiveness of this technology to reach poor and vulnerable groups of people. During the lockdown, this solution led to over 20,000 user contributions, of which more than 7,000 were SOS-style grievances. 

An integration of our solution with digital ID systems will make error codes and other information published on welfare scheme related MIS (Management Information Systems) accessible to people, against which they can register a grievance, and get guidance through audio messages on next steps to follow to avail their welfare benefits. The use of machine learning and speech2text systems has also enabled us to automatically process many of these recordings and map them to the underlying causes of errors and rectification steps. We have followed user centered design principles to build and evaluate our technological systems, and fine-tune them to improve usability; we will continue to follow the same approach for integration of our solution with the relevant digital ID systems. 

Our IVR solution is already API enabled and several of our partners integrate with our systems through these APIs. We envision an integration with APIs of MIS (Management Information Systems) of different schemes, to query information for beneficiaries based on their unique digital ID, Aadhaar, and to register grievances in case they face problems with access to these schemes. Correspondingly, the MIS systems will also be able to integrate with our system by invoking APIs to push to beneficiaries information about errors faced with their identification or authentication, or bank transfers, or other operations. Our APIs are RESTful with data exchange through JSON and XML formats. These standards are actively used for APIs of other systems and therefore we do not envision technical challenges with API-level integration. 

The voice-based nature of our solution makes it easily accessible for even less literate people, and does not require the Internet or smartphones. People can interact through keypresses and browse audio messages, take surveys, forward messages to others, and record their own messages. Machine learning solutions operating at a 98% accuracy are used to automatically filter out audio messages with less noise. Speech2text solutions are then used to automatically transcribe good quality messages, identify the gender and topic of these messages, and the location spoken by the user. 

This information is passed on to a moderation team that reviews each message and selects for publishing those messages that are in line with our editorial policy. Among these messages, any grievances that are identified are assigned to a community volunteer from among a network field volunteers in our geographies of work. The volunteers are trained with awareness about government schemes and procedures, and bring from the same community themselves, are able to understand the context and respond effectively. The volunteers use various pathways to draw the attention of government officials to issues that require their action, or they assist the beneficiaries to resolve errors, and also work with the local administration to organize registration camps or run door-to-door campaigns to ensure that welfare measures reach the people. 

This combination of appropriate technology that is accessible to marginalized populations, and an integration with people-processes led by a trained network of community volunteers, is able to reach the most vulnerable communities. 

Our steady reach for complaint reporting and other information services until before the COVID-19 pandemic was 100,000 monthly unique users. Users on average call five times a month, and spend 8 minutes per call on the platform listening to audio content. Approximately 3-4% of the daily calls also lead to voice recordings where people contribute local news, views on local and national policy, grievances with welfare schemes, questions on agriculture and livelihood, etc. This reach is across 25 districts spread about primarily North Indian Hindi-speaking states. 

During the lockdown, our IVR solutions were used extensively by several partners organizations to provide information services in their geographies. We have clocked almost 1.5 million calls from approximately half a million users. Not all these calls were related to digital ID complaints, but were meant for other services like providing COVID-19 related awareness information. 

Our goal is to grow the steady reach of Mobile Vaani to 500,000 users by early-2021, and scale to 10,000,000 users in the next five years. These people will be directly impacted with reduced exclusions in the delivery of social protection measures, most of which are linked in India via the Aadhaar digital ID system. 

Our model operates at the unit-level of a district. For each district of operation, our IVR solution is accessible via a unique phone number, which carries relevant information for that district. This includes local news, grievances recorded by people, impact achieved, agricultural questions, folk songs, and a range of other programmes. The community presence is achieved through a network of field volunteers, in most districts at about 1 volunteer per block (a district in India on average has 10 blocks), but in some blocks where we have been working since several years we have up to three volunteers in a block. The volunteers come from a range of background - teachers, health workers, regional media, social workers - and their Mobile Vaani related work is supported through a modest stipend. 

Since the content on our platforms is mostly user generated, scaling for us means (a) technology scaling, and (b) people scaling of building volunteer teams: 

- Our technology is robust and can be readily scaled horizontally - we run our IVR services in a cloud hosted manner, and during the lockdown the systems scaled easily up to 100,000 calls in a day. Further scaling is not a challenge.

- We also have experience in replicating our model successfully to 25 district, and additionally also helping partners run the model in their geographies for purposes like information dissemination. We have built internal protocols for volunteer recruitment and training, and referrals to identify new volunteers. 

Access to capital to scale is the only barrier we currently face. Our business model is to achieve financial sustainability through (a) commercial advertising by companies that want to reach rural communities, and (b) awareness or other campaigns that development organizations or governments want to run for marginalized populations that are hard to reach through other media channels. We raised an early stage equity investment in 2015, and have been profitable (close to breakeven) since the last three years. 

We have successfully demonstrated our model and additional capital will help us expand our userbase and consequently increase the inventory we can service for social campaigns and commercial advertising. Further, during the COVID-19 lockdown in India, we also successfully demonstrated the wider need for a platform like ours, to hear from people about the problems they are facing, draw attention to these problems through advocacy with the government, provide instant relief through our community volunteers and a wider network of response partners, and be agile in changing our priority to emerging needs. 

While we require capital to scale, and we have a proven business model, we are also constantly identifying new opportunities to make it easier to scale. 

We actively use machine learning in speech technologies to reduce the manual labour for moderation, and are constantly using new data to improve the performance of our algorithms. We are also building new innovations like automated question-answering systems, where a user may ask a question and in real-time we will return an appropriate answer from a pre-recorded FAQ database. A prototype of this FAQ-retrieval system operates at a 70% accuracy. This solution can become a backbone to help people understand issues and corrective steps they may need to take to access welfare schemes linked with their digital IDs. 

We are also utilized Text2speech systems and have piloted already a method for citizens to access information about their welfare related transactions, such as the number of units of food ration for which they are eligible, the units of ration they received last, any other benefits such as cash transfers for which they are eligible. This information is accessible online through a web-based system that the beneficiaries are not able to use directly. We were able to crawl this information in text, convert it to audio, and make it accessible to people over an IVR system. We plan to initiate discussions with relevant government departments so that such a solution can be implemented to make G2C (government to citizen) services more easily accessible. 

Gram Vaani is a team of 70+ full-time staff, comprised of a software engineering team, a technology operations team, a content creatives team, a community mobilization field team, a programme management team, and a research team. We work across the domains of governance and media, health and health systems, gender and child rights, and workforce. 

The solution proposed is in partnership with Dvara Research, a policy research institution based in India with about 20 interdisciplinary researchers whose mission is to ensure that every individual and every enterprise has complete access to financial services.

The Gram Vaani and Dvara Research teams have already been working on this concept and are well-positioned to scale it further. 

Gram Vaani is a global leader in the design of appropriate technologies for development. We have won several national and international awards, written research papers, and are cited regularly by academics and development practitioners alike for our technology innovations coupled with careful programme and process design to ensure impact out of these innovations. On the specific solution we have proposed here, we have already demonstrated its feasibility and relevance, and have written actively in the media using the data generated through the solution, eg: India Development Review, Scroll, India Forum, India Spend.  

Dvara Research (DR) is a not-for-profit policy research and advocacy institute whose primary mission is to ensure access to financial services for all individuals and enterprises. DR has made several contributions to the Indian financial system and participated as experts in various engagements with key policy making institutions such as the Reserve Bank of India (RBI), Securities and Exchange Board of India (SEBI) and the Government of India advocating for “suitability” in the design of financial services and stronger consumer protection for low-income households. Among other efforts, DR has contributed to various bodies including the Financial Sector Legislative Reforms Committee (FSLRC), Committee of Experts on data protection under the Chairmanship of Justice B.N. Srikrishna, RBI’s inter-regulatory Working Group on Fintech & Digital Banking and SEBI’s Committee on Social Stock Exchange.

We work in a collaborative manner with other organizations to learn and take support from their expertise. 

We actively work with faculty and students at the Indian Institute of Technology Delhi in the use of machine learning technologies. Our platform generates a large volume of data and opens up new research problems for academic institutions. 

We also collaborate actively with researchers in public health, from the University of Montreal and the London School of Health and Tropical Medicine. While our primary collaboration is in behavior change communication for immunization, and real-time data collection of routine health services, the collaboration was instrumental for our COVID-19 response because our partners were able to quickly help create new awareness content related to COVID-19 that we were able to then instantly deliver to our users. 

Similarly, our collaboration with Dvara Research builds upon our complementary strengths, where Dvara's understanding of digital ID systems and links with social protection goes well with our own understanding of ground-level issues and our field presence. 

We also work with a large number of other partners in implementing our development programmes, such as the Population Foundation of India with whom we operate an entertainment-education voice platform, CREA with whom we run a channel for sexual and reproductive rights and health, Project Concern International with whom we work on health and nutrition of pregnant and young mothers associated with Self Help Groups, Enable India with whom we operate information channels for physically disabled users, etc. 

Our services are delivered free of cost to users in a B2C manner, and financed through B2B relationships such as social campaigns and commercial advertising. Our users comprise rural and urban low-income populations, who are able to benefit both through accessing useful information as well as getting assistance for their grievances on welfare and other problems. Our B2B partners and customers are able to engage with this userbase through our platform, to provide information, answer questions and provide guidance, respond with help, and also to collect data for research purposes. Our solutions are priced by volume, based on the number of users that our partners want to reach, and we additionally provide services such as content development or field training to help our partners use our platforms and network effectively. 

Our B2B2C services are unique. We are able to reach populations that are hard to access through other channels. We are also able to provide a rigorously instrumented access to this population, measured in terms of the number of users reached, time spent, content heard, content contributed, etc. Further, our platforms can be used in different ways for different objectives, making them agile to service a wide variety of programmes. So far, we have provided these services to over 150 partners, who have reached more than 2 million users through our platforms. 

Social campaigns on our platforms are primarily donor funded, either directly to us for our own development programmes, or to partners for collaborative programmes. Since our platforms are horizontal, they are agile to service different domains and we actively work on governance, health, gender, workforce, and agriculture related topics. 

Commercial advertising campaigns on our platforms have been funded by tractor companies, solar-lighting companies, FMCG, and healthcare chains. 

After an equity investment raise in 2015, we have been profitable (close to breakeven) since the last three years and are in discussion with impact investors to raise a second round for scale-up. We see our path to financial sustainability as a capital raise to expand our userbase to more districts, following which we will be able to expand our revenue from these newer geographies that we add. Our strategy is to first strengthen ourselves as a strong regional player, in all districts in two states, and then expand to more regions. 

We are also contemplating other revenue sources, such as crowd-funding platforms for specific campaigns, and also user membership fees like in a trade union to support media services in their geographies. We hope to test these revenue streams in the near future. 

Equity investment (2015): 

- Media Development Investment Fund: $400,000

- Indian Angel Network: $100,000

Significant grants: 

- Bill and Melinda Gates Foundation: $700,000 for health and nutrition, over two years

- Laudes Foundation: $800,000 for labour rights, over three years

- USAID: $200,000 for a women media collective, over two years

- Grand Challenges Canada: $100,000 for immunization, over two years

We are open to the following funding routes. They are not mutually exclusive. 

- Equity investment: $2.5M, tranched over two years, to facilitate expansion from our current reach of 25 districts to 60 districts

- Grant investment: $500K over two years, to expand coverage of our volunteer network to deeper presence in our current geographies to service the most marginalized users for improved access to social protection. 

Unlike equity investment which we consider useful for horizontal expansion, we view grant investments as essential to reach hard-to-reach groups. 

We expect our 2020 turnover to be approximately $1.3M. 

We view the prize as being important to build strategic linkages with more partners, and to draw more attention to the issues we are passionate about. Social protection of the vulnerable is of critical importance, especially during the COVID-19 crisis when many of them who already were living on meager incomes have now lost their jobs and other sources in income. The Aadhaar digital ID system is a critical component in the delivery of welfare services, and many issues persist with the use of Aadhaar for biometric authentication, linkage with government schemes, access information about eligibility, etc. It is imperative to systematically address these issues and our solution serves to clearly outline where and why digital ID systems fail and lead to unfair exclusions. This will help build corrective methods and strengthen the architecture for social protection delivery. The visibility we gain through the challenge can be important to highlight these issues. 

With the insights and documentation we will generate about the required improvements in the delivery of social protection measures, we will benefit from greater media exposure, creating short films or podcasts, and reports and policy briefs, to draw attention of policy makers and stakeholders to the challenges in social protection delivery through digital ID systems. 

We will also benefit from wider linkages with partners who could be interested in using our solution, since it can be adapted from social protection to also consumer rights in general. 

Finally, we will benefit from technology partnerships that can help improve our technology stack, absorb more of the latest innovations in machine learning and artificial intelligence. 

We would like to partner with government departments such as UIDAI (which operates the Aadhaar infrastructure), the DBT (Direct Benefits Transfer) mission which links government schemes with cash transfers, and NPCI (National Payments Corporation of India) which executes the cash transfer. 

We would also like to partner with government departments that have field cadre, such as PRI (Panchayati Raj Institutions that are responsible for local governance at the village level), MoRD (Ministry of Rural Development which operates important schemes like for national rural employment guarantee), and MLE (Ministry of Labour and Employment which looks after social protection schemes among workers). 

These partnerships will be crucial for them to improve internal systems and processes based on the insights generated by our solution.