Case study

Driving Microtargeted Development in 290 Villages

Overview

The Tata Trusts and Government of Maharashtra partnered with us to drive rapid development in Chandrapur. We used our platform to create village development plans that improved budgeting, program implementation, and department planning.

Partners

Tata Trusts

Government of Maharashtra

Sudhir Mungantiwar

Cabinet Minister, Maharashtra

Ashutosh Salil, IAS

District Collector, Chandrapur

Sectors

Government

Philanthropy

290

villages covered

160,900

people surveyed

6.9 mil.

data points collected

900

surveyors trained

This story was featured on TED Talks India Nayi Soch

The Problem

Driving targeted development in one of India's most diverse districts

Chandrapur has vast natural resources, yet it remains underdeveloped. However, a traditional development plan can't account for the sheer diversity of Chandrapur's blocks. Chandrapur needs targeted development plans for every household, village, and block.

This type of microplanning generally takes 6-9 months, but we only had 90 days. To make matters more difficult, there were 3 additional problems: Chandrapur has poor infrastructure and connectivity (people have to walk kilometers to find cell service), only 5% of its population is computer literate, and its village lists are not accurate or up to date.

The Solution

Automating the creation of data-driven village development plans

The Tata Trusts and local governments partnered with us to help district and block officials create microtargeted village development plans in Chandrapur. We trained first a team of surveyors, who used our data collection app to create a comprehensive household-level data set for 290 villages. We then visualized this data in an interactive dashboard with automated village development plans, geoclustering, village-level comparisons, household-level views, village profiles, and intelligent querying tools.

How we brought this dream project to life

1. Questionnaire creation
1. Questionnaire creation

Our team created custom baseline surveys — each with 150-200 questions — for each village with built-in data validations.

2. Training and piloting
2. Training and piloting

We trained 50 facilitators and 900 volunteers on data collection, followed by 7 days of piloting and 4 rounds of questionnaire iteration.

3. Data collection
3. Data collection

Volunteers surveyed households, completing 3,000 surveys each day on average. A total of 6.9 million data points were collected in just 90 days.

4. Data flagging
4. Data flagging

As data was collected, it was automatically verified and bad data was flagged for re-collection.

5. Data analysis
5. Data analysis

Our data scientists processed, cleaned, and analyzed all the data to create village scorecards.

6. Data visualization
6. Data visualization

We visualized the data on an interactive dashboard with village development plans, profiles, and more.

When you have numbers, figures and data in front of you, you stop shooting in the dark.

Ashutosh Salil, IAS
Ashutosh Salil, IAS

District Collector, Chandrapur

Government of Maharashtra
Convert the app to local languages
No internet required
Geotag each household
Reduce survey time with skip logic

Data collection app

  • Many surveyors only spoke Marathi. The entire Collect app — including action buttons and instructions — could be converted to Marathi by simply changing the language setting.

  • Many parts of Chandrapur do not have mobile or internet service. Data was continuously saved to tablets’ local storage, then synced to central servers when internet was available.

  • Every household was geotagged on a map using GPS, even without internet. Surveyors also used Collect to map health centers, schools, and village infrastructure on a satellite map.

  • Every survey was automatically customized with the most relevant questions for the person being surveyed. This saved crucial time on every survey.

Learn more about Collect

Automatic backchecks

As data was collected, it was automatically checked. Any data point that fell outside of pre-set parameters or was inconsistent was flagged on Collect. Then surveyors immediately returned to check and re-collect that survey in the field.

Data transformation

1

Consistency checks

Included intra-variable (checking each variable for incorrect values) and inter-variable checks (ensuring that data is consistent across variables).

2

Village scorecard creation

Data was aggregated to score the development of each village, based on various individual, economic, health, and infrastructure development indicators.

3

Schemes matching

By matching eligibility data for each scheme with each person's data, we figured out when people were not using schemes that they were eligible for.

View insights for each block
Identify clusters for development
See each village's development plan
View comprehensive village profiles

Interactive data dashboard

  • Government officials could compare 80 demographic, economic, health, and infrastructure data for each block.

  • Mapping made it easy to find hidden patterns, trends and insights that can be leveraged for better development plans.

  • Village development plans — with the priority and cost of each intervention — helped decision makers plan budgets effectively.

  • All the data — healthcare, education, infrastructure, and more — that officials needed was available through interactive charts.

There were times when we didn't have internet, didn't have network, and were in the remotest villages, but we still collected data for each and every household.

Field volunteer

The Results

Improved budgeting, policy decisions, and program targeting

The 290 village development plans created through this deployment were used by government officials at all levels to improve their budget and policy decisions to drive rapid, effective development in Chandrapur.

Roll out of development plans across the district

The village development plans were sent to Gram Panchayat heads for all 290 villages. The Block Development Officer of Mul added 60% of the plans' suggestions to his 2016-17 budget. The Guardian Minister adopted 18 villages and planned to use to transform these villages into model villages.

Better targeting for government programs

Government officers across all departments used this data to better target their programs and policies. For example, the Electricity Department used the dashboard to find which households currently aren't receiving electricity, and the Forest Department used the dashboard to preserve forest areas by finding the households that are most likely to burn wood (i.e. households near forest areas that don't have an LPG connection).

Improved field visits and government meetings

The District Collector of Chandrapur used our dashboard to verify claims from his department officials in real time and to cross-check villagers’ development priorities against priorities from village development plans during his field visits. This helped him eeliminate false reports or less important complaints and focus on the most important issues in each village.

This project helped the district administration understand the socioeconomic dynamics and development challenges of each village by creating a robust village requirement sheet for each and every village in Mul, Pomburna and Jiwati.

Sudhir Mungantiwar
Sudhir Mungantiwar

Minister of Finance, Planning and Forest Departments

Government of Maharashtra

Watch how data-driven microplanning transformed 290 villages in Chandrapur, Maharashtra

At Atlan, we're opening up internal tools, like data catalogs, that helped us power massive data projects around the world.

Visit Atlan What Is a Data Catalog?