Author Name
Frontier Tech Hub
Choose Report Type
Publication Date
Report Upload
FTH-Quaterly-Insight-june.pdf
(3.91 MB)
PDF Text
India’s Data Imperative:
The Pivot Towards Quality 1 India’s Data Imperative:
The Pivot Towards Quality
How high-quality data can unlock better outcomes, build public
trust, and power the next decade of digital governance
Knowledge Partner: India’s Data Imperative: The Pivot Towards Quality
I
am delighted to present this edition of the NITI Aayog Frontier
Technology Hub’s Quarterly Insight on “India’s Data Imperative:
The Pivot Towards Quality”. This timely report underscores a
crucial shift in India’s data journey – from achieving unprecedented
scale to ensuring uncompromising precision and trustworthiness
in our data.
Over the past decade, India has truly distinguished itself as
a global leader in digital public infrastructure. The Unified
Payments Interface (UPI) moves trillions of rupees monthly,
Aadhaar has authenticated billions of identities, and Ayushman
Bharat has extended healthcare access to more than 350 million
citizens. These are not merely statistics; they are testaments
to a foundational transformation, enabling financial inclusion,
empowering citizens, and streamlining welfare delivery on a scale
previously unimaginable. Our collective efforts have laid robust digital
rails, demonstrating to the world how technology can empower a
nation of 1.4 billion.
However, as we embark on the next phase of our digital evolution, the
focus must decisively pivot towards the quality of the data that fuels
these incredible platforms. As this report highlights, data quality is
no longer a peripheral technical consideration but a fundamental
prerequisite for good governance and sustained public trust. When
a single erroneous digit can delay a benefit, or a duplicate record
inflates welfare outlays, the true cost of poor data becomes painfully
apparent – impacting budgets, distorting policy, and eroding the
faith citizens place in our digital systems. The move from “scale to
precision” is not just an aspiration; it is a national imperative.
I particularly commend the inclusion of two practical, easy-to-use
tools in the report: The Data-Quality Scorecard and the Data-
Quality Maturity Framework that are designed to enable every
department to conveniently assess their current data landscape,
identify gaps, and chart a clear roadmap for improvement. Whether
it is implementing real-time validation at the source, assigning data
stewards, or linking grievance redressal to backend corrections,
the “Starter Kit” outlined in this report offers tangible pathways to
achieve quick, verifiable wins.
This report serves as an invaluable guide for all government Ministries
Foreword A
s India matures as a Digital Economy, we must increasingly
prioritize a critical imperative: data quality. The shift
from scale to precision makes high-quality data not just
desirable, but essential. Even a single incorrect digit can halt a
pension, deny healthcare, or misdirect subsidies, profoundly
eroding citizen trust and costing billions in fiscal leakage.
Data quality must therefore become a frontline governance
imperative, integral to building Digital India on a robust foundation
of trust. Trust is not merely a desirable outcome—it is the bedrock
upon which our digital ambitions stand and scale.
This edition of NITI Frontier Tech Hub’s Quarterly Insights,
Future Front, developed with our knowledge partner Gramener,
emphasizes the urgent need to radically transform our approach to
data quality. Strengthening data quality is foundational to realizing
our digital ambitions—whether enabling AI-driven governance,
ensuring targeted welfare delivery, or driving cross-sector innovation.
It requires collective action, clear accountability, and a shared
commitment from every department and ministry.
DEBJANI GHOSH
Distinguished Fellow, NITI Aayog;
Chief Architect, NITI Frontier Tech Hub
and Departments. It meticulously dissects the challenges across the
entire data value chain – from generation and storage to sharing,
use, and retirement. I urge the Ministries and Departments across
the Government of India and the State Governments to thoroughly
engage with this report. The insights and recommendations in the
report are aimed at embedding a culture of data excellence – one
where quality is not an afterthought but a shared responsibility,
deeply ingrained in our institutional roles, incentivized effectively,
and supported by robust interoperability.
Our digital future will be defined not just by the platforms we build,
but by the unwavering trust we cultivate in them. This trust begins
with data that is precise, reliable, and truly ready to serve. Let us
champion clean, usable data as the foundation of good governance,
ensuring that the dividends of our digital journey reach every citizen
with precision and equity.
DR. SAURABH GARG
Secretary, MoSPI India’s Data Imperative:
The Pivot Towards Quality 1
EXECUTIVE SUMMARY:
India’s digital public infrastructure has grown at a breakneck speed. UPI now moves trillions of
rupees each month, Aadhaar verifies billions of identities, and Ayushman Bharat touches half a
billion citizens. As data volumes explode, it is imperative for data quality to keep pace across both
the torrents generated today and the decades of legacy records still in use. Records need to be
precise, complete, and up-to-date by reducing or eliminating gaps, duplicates, or timing lags that
may exist today. When one wrong digit can freeze a pension or misroute a subsidy, quality is no
longer a technical afterthought but a frontline service obligation. India’s digital governance will only
be as strong as the quality of the data that powers it. As India moves from scale to precision, data
quality must become a national priority—on par with infrastructure and platforms.
Why Data Quality Matters:
1. Fiscal leakage, erroneous or duplicate beneficiary records drain budgets, inflating
welfare outlays by an estimated 4–7 per cent annually.
2. Policy blind spots, inconsistent, non-standardised datasets distort evidence, leading to
mis-targeted schemes or delayed course-corrections.
3. Public trust erosion, citizens confronting mismatched records, rejected claims, or
endless rectification queues lose confidence in digital governance.
This edition of the NITI Frontier Tech Hub Quarterly Insight, created in partnership with Gramener,
lays out data-quality challenges across the full data value chain, analyses their root causes, and
offers two easy-to-use tools to measure data quality and maturity in any department.
1. SETTING THE STAGE: INDIA’S DIGITAL & DATA MOMENT
India’s digital public infrastructure now operates at a scale few imagined a decade ago. In April
2025, the Unified Payments Interface (UPI) processed 17.89 billion transactions worth ₹23.9trillion,
1
rivalling the monthly GDP of several mid-sized economies. Aadhaar authenticated 27.07 billion
identity requests in FY 2024-25,
2
confirming its place as the common key for banking, welfare and
a widening array of private-sector services. On the social-protection front, more than 369 million
Ayushman cards are now in circulation,
3
bringing cashless hospital care to roughly half a billion
citizens.
But the next test is precision, not reach. What if just 5 % of records are flawed, and those belong
to our most vulnerable citizens? The true test of digital infrastructure is how precisely it represents
each individual, not just how many people it onboards. As platforms mature, the focus must shift
from expansion to fidelity. Digital highways only create public value when every record is accurate,
complete, and current. And no stack of audits can rescue accuracy if the people who create or rely
on a record have no sight or stake in it. Confidence endures only when the people who depend on
a record can also see and correct it.
1.1 Small errors carry huge costs
A single wrong digit during Aadhaar enrolment can block a pension. A duplicated mobile
number can stall a hospital claim. A misspelt name in a land record can delay compensation
after a natural disaster.
Multiply such slips across several central schemes and thousands of state programmes and
the fiscal dent is staggering. Every wasted rupee crowds out an honest claim; every wrong
dashboard alert delays a life-saving decision. Bad data costs real lives and real money. India’s Data Imperative:
The Pivot Towards Quality 2
1.2 From scale to fidelity
Over the last decade India laid the rails, identity, payments, registries, that make inclusive
innovation possible. The coming decade’s services, farm-credit via account aggregators,
portable skill credentials, AI-powered health alerts, depend on the integrity of the data flowing
over those rails. Where records are rigorous at the moment of capture, new services can
scale at marginal cost; where they are inconsistent, each added layer inherits and amplifies a
structural deficit.
Recent clean-ups underline the pay-off
ActionEstimated Savings
Deleted 17.1 m ineligible PM-Kisan names
4
₹90 billion (FY 2024)
Weed out 35 m bogus LPG connections
5
₹210 billion (in two years)
Drop 16 m fake ration cards
6
~₹100 billion per year
1.3 What holds us back
Across ministries and states, platforms are built as separate silos that store data in clashing
formats and use incompatible identifiers, forcing even routine joins to be stitched together
by hand. Many of these systems run on ageing back-ends that lack basic validation rules,
audit trails or version control, so a small tweak in one module can topple an entire workflow.
Because records shuttle between teams without a named custodian, obvious errors linger
unchecked, and field programmes are often driven by enrolment targets that prize speed
over accuracy. Over time this breeds a “good-enough” culture, once 80 percent accuracy
is accepted as normal, the system quietly stops trying to prevent error and the data quality
debt compounds.
1.4 Towards “data confidence”
Building confidence begins at the point of capture: automated checks and standard pick-lists
must stop bad entries before they enter the system. Every high-value dataset then needs a
clearly identified steward—someone empowered, and accountable, to correct and continually
improve the records. Finally, common schemas for people, places and programmes must be
adopted so that data can travel seamlessly across agencies without constant translation.
Together, these three practices turn today’s fragmented information into a reliable asset that
analytics, AI and service delivery can use without sacrificing trust.
The pivot from scale to fidelity demands a new scorecard—one that values precision and
trust as highly as reach.
The table below shows how priorities must shift, toward iterative architecture, stronger
interoperability rails, and shared stewardship roles for AI-ready data.
Scale EraFidelity Era
Number of users onboardedPrecision in every record
Platform speed and uptimeTrust in data at every stage
Build core digital public goods, identity,
payments, registries
Iterate architecture, strengthen interoperability
rails, and embed shared stewardship roles
Without quality, data is just noise.
India has proved it can scale digital public goods. Showing the world we can sustain high-
fidelity data will cement India Stack as a governance model, not just a technology export, India’s Data Imperative:
The Pivot Towards Quality 3
and ensure the first dividends reach those who need them most. Getting there will demand
as much organisational reform as technology: frontline training, budget lines for data
stewardship, and lightweight policy guardrails that make quality a default requirement rather
than a discretionary upgrade.
2. SEEING THE WHOLE ELEPHANT: THE PUBLIC SECTOR DATA VALUE CHAIN
A farmer’s subsidy application, a child’s vaccination record, or a woman’s property title all
start as a few keystrokes or biometric scans. Those fragments soon underpin service delivery
and policy. Between capture and a final decision, data travels an often-hidden chain - stored,
transmitted, transformed, analysed, and eventually archived. Quality can slip at any link, with
losses that snowball for citizens and the agencies that rely on the record.
These five stages of the data value chain – generation, storage, sharing, use, and retirement –
present distinct challenges and play a role in shaping data quality. We’ll illustrate each stage
with Lakshmi Devi’s PM-Kisan record, a single row that travels the chain.
2.1. Generation - Where It Begins
Most data-quality faults are born at the moment of capture, whether a field agent taps details
into an enrolment app, a clerk uploads a survey sheet, or a sensor pushes logs to the cloud. In
resource-tight settings a single typo in a name, Aadhaar, or bank IFSC can ricochet through
every downstream system, blocking payments, skewing analytics, and triggering costly
reconciliations. Fixing them later often requires time-consuming manual reconciliations and
frustrated citizens.
Example - At the village service centre, Lakshmi Devi’s PM-Kisan details were entered with
two digits of her bank IFSC transposed. Because the enrolment screen showed only a
confirmation tick, neither she nor the officer noticed the slip.
2.2. Storage: The Quiet Fade of Confidence
Once generated, data must be stored in a way that preserves its structure, context, and
integrity. This includes defining how data is formatted, tagged, and secured; how it is
updated over time; and how different departments manage their own databases. In practice,
storage practices are fragmented. Some systems now employ cloud-native, version-controlled
stores, while others still depend on legacy databases with no audit trail.
Example - When the day’s data synchronises, Lakshmi’s record is copied from the enrolment
app to the state PM-Kisan server and then to a legacy backup database. Because neither
system runs cross-check or version control on incoming fields, the transposed IFSC is stored
unchanged in both locations, quietly duplicating the error.
2.3. Sharing: When Silos Collide
Sharing data across ministries, departments, or platforms is where digital governance can truly
shine and where friction most often flares. Crucially, inconsistency is not only horizontal (one
ministry versus another) but also vertical (different schemes, dashboards, or district offices
inside the same programme). A pension portal may store household size at the family level,
while a village survey logs each individual; a health-worker app records a child’s age in months,
yet the state immunisation dashboard expects years. The grain of data, the frequency of
refresh, and even the date formats vary wildly, turning simple joins into costly clean-ups that
stall service delivery.
Example - At quarter-end, the agriculture portal exports beneficiary data to the treasury
gateway for bulk transfers. Lakshmi’s row fails the bank’s validation check - “invalid IFSC” India’s Data Imperative:
The Pivot Towards Quality 4
—and is returned as an exception. Her payment is held until staff review the file, correct the
code, and resubmit the batch.
2.4. Use: Where the Stakes Get Real - and the Rewards too
When government data feeds a service queue, a policy dashboard, or an AI model, any
upstream flaw becomes a public-facing fault. Clean, time-stamped records can power tools
that flag disease outbreaks, fine-tune subsidy targeting, or route traffic in real time. But if two
datasets record farms at different granularities or age a child in months versus years, even
the smartest algorithm will misfire.
Architecture matters too: a single data lake risks breaches, while sealed silos hide cross-
domain patterns. Newer methods—federated learning, synthetic data, privacy-enhancing
computation, let models train where data sits, but only if every record carries consistent
schemas, consent flags, and update logs. In short, garbage in, algorithmic garbage out:
disciplined capture, harmonised storage, and governed sharing are the non-negotiable tripod
beneath any citizen-centric AI.
Example - The programme manager builds a drought-relief model that relies on “successful
payment” history to spot cash-constrained clusters. Because Lakshmi’s transfer shows as
failed, her village appears slightly better off than it is, nudging fertiliser stock toward other
blocks.
2.5. Retirement: The Data We Forget
Archival is the stage most systems skip. With no clear “sunset” rules, obsolete entries linger,
inflating counts, slowing queries, and sometimes exposing private details. Welfare rolls still
list families that have moved away; backup servers hold duplicates long after production
tables are cleaned.
Fixing this is less about storage and more about policy: time-bound retention schedules,
annual “alive-and-eligible” checks, and safe deletion once legal purpose lapses. Regular
pruning keeps databases lean, dashboards truthful, and public trust intact, freeing analysts
to focus on trends that matter, not ghosts in the machine.
Example - Six months later a data-quality drive updates Lakshmi’s IFSC in the main server,
but the duplicate row in an older backup is missed. Unless that orphan entry is archived
or reconciled, it will keep resurfacing during audits and quarterly reconciliations, requiring
manual checks each time.
Together, these five stages show how one mistyped field can ripple from capture to archive.
The next chapter tallies the real-world cost of letting those cracks persist, and why they
demand urgent repair.
This journey reveals how cracks can form at any link—and why fixing them demands a closer
look at the systems behind the data.
3. THE NATURE OF THE CHALLENGE
India’s public data engines hum louder each year, powering cash transfers, hospital authorisations,
fertiliser caps, and pandemic dashboards. Yet familiar cracks reappear: errors slip through, gaps
don’t close, fixes don’t hold. These aren’t stray glitches, they stem from how data is captured,
shared, and rewarded.
This chapter identifies six recurring design flaws, pairs each with a real-world incident, and highlights
the risks associated with leaving them unaddressed. India’s Data Imperative:
The Pivot Towards Quality 5
3.1 Systemic, not Isolated
Data quality lives or dies in system design. Every dataset, whether a 1995 ledger or a 2025
API stream, inherits the roles, incentives and validation rules baked into its first mile. When
incentives favour quantity over correctness, mistakes recur, no matter how many downstream
audits we conduct.
Example - During a 2017 linkage drive, three regions uncovered 4.4 lakh “ghost
students” claiming midday meal funds. In one state alone, more than two lakh fictitious
names vanished once biometric attendance became mandatory, proof that lasting quality
follows when accuracy is rewarded.
Risk: phantom beneficiaries drain budgets and trust every time a new scheme launches.
No stack of audits can substitute for giving the record’s true owner - teacher, farmer, patient -
live visibility and the right to correct it. The “two screen” strategy used at Aadhaar enrolment
centres – exposing the data to the Aadhaar holder to view and correct the record – is an
excellent example of implementing this.
3.2 Fragmented by Design
Platforms often start with good intentions but end up maturing in isolation. A ministry
launches a portal tailored to its flagship scheme; states then clone and tweak that code
to meet local rules; over time, each branch adds its own columns, codes, and shortcuts.
Formats drift apart, hectares in one table, acres in another, full names in one file, initials in the
next, dates logged as DD/MM/YYYY in one district but MM-DD-YY next door. What should
be a simple join across systems turns into days of manual cross-walks and spreadsheet
contortions, delaying benefits and muddying analysis.
Example - In 2021, roughly one-third of crop insurance claims in an eastern district stalled
because land titles listed parents as owners while crop-loss forms named the farmers
themselves. About 1.5 lakh applications sat idle until staff reconciled the two schemas
Risk: fractured views create blind spots and delay help when citizens need it most.
3.3 Legacy Systems, Modern Pressures
Many core registries still run on infrastructure built for a different era. They’ve been patched
over the years, but the core design remains unchanged. Basic features, such as validation
logic, audit trails, or version control, are often missing. Dashboards may have been added to
improve visibility, but they sit atop brittle backends. As policies evolve and field realities shift,
these systems struggle to keep pace. Even minor changes, such as updating a beneficiary
status, can require custom workarounds.
Example- In 2024, elderly ration holders in a western district queued for months after
fingerprint readers failed to recognise worn prints. Iris scanners arrived late, but thousands
missed subsidised grain at the peak of inflation.
Risk: brittle systems turn minor tweaks into outages for households on the margin.
3.4 Everyone’s Job, No One’s Responsibility
Data moves across departments and systems but rarely comes with a clear owner. One team
collects it, another uses it, and a third reports on it. When errors show up, it’s not always clear
who is responsible for correcting them or whether anyone has the authority to do so. This
lack of custodianship means even obvious mistakes can persist for years, not because they’re
hard to spot but because they fall between the cracks of institutional roles. India’s Data Imperative:
The Pivot Towards Quality 6
Example - In late 2022, 17,000 health insurance cards were stuck because beneficiary names
or biometrics failed identity checks. Health staff labelled it an “ID problem,” the ID team saw
a “health-database issue,” and elderly citizens paid cash until a joint task force cleaned the
queue.
Risk: obvious errors linger for years, eroding faith and loading staff with avoidable rework.
Industry practice shows that durable quality emerges when systems expose errors to the
people who benefit most from clean data, instead of layering extra audits.
3.5 When Speed Trumps Quality
Most programmes are evaluated by how fast they move: how many people were enrolled,
how quickly benefits were delivered, and how soon dashboard indicators turn green.
These indicators are helpful, but they can create unintended consequences. When speed is
rewarded, but accuracy is not, frontline staff may rush through data entry. Mid-level officials
may hesitate to reopen past data for fear of triggering audit flags. Over time, this creates a
culture where getting it done is more important than getting it right.
Example- During a large LPG subsidy roll-out in 2013, ID bank linkages covered only 60 per
cent of households; the remaining 40 per cent were bulk-rejected to keep the launch on
schedule. It took two years of painstaking review to unlock the full savings—showing that
haste can cost more than delay.
Risk: performance metrics turn into vanity stats, hiding debts that grow costlier over time.
3.6 The Cost of Low Expectation
Once systems learn to live with error, they stop trying to prevent it. A database that’s 80%
accurate becomes “good enough.” A mismatch between two registries is seen as a minor
technical hiccup rather than a barrier to service delivery. This mindset shapes everything:
how platforms are designed, how staff are trained, and how performance is judged. Slowly,
quietly, quality becomes optional.
Example- A large northern state declared itself open-defecation-free in 2019; however, an
audit of 590 rural houses found that nearly half lacked toilets. Dashboards still showed 100
per cent coverage, and funds kept moving—evidence that data judged “close enough” can
override field reality.
Risk: policy rides on numbers no one fully trusts, undercutting evidence-based governance.
These six flaws reveal a design defect, not a tech shortfall. Quality will improve only when
priorities shift from speed to precision, ownership, and aligned incentives. The next chapter
calculates the cost of persistent cracks in rupees, including delays and the erosion of public
trust. It explains why fixing them now is more cost-effective than patching them indefinitel.
To move forward, we need a shared benchmark of what quality truly means—beyond error
rates, into attributes that build trust. India’s Data Imperative:
The Pivot Towards Quality 7
4. WHAT GOOD LOOKS LIKE: ANATOMY OF HIGH-QUALITY
DATA - SIX ATTRIBUTES
Picture a GPS that shows every road but scrambles half the street names—you’d still get lost. Data
behaves the same way.
Before we can improve data quality, we must define it—clearly and measurably. Without a shared
understanding, every clean-up drive veers off course, leaving efforts inconsistent and results short-
lived. Six core attributes—accuracy, completeness, timeliness, consistency, validity, and uniqueness—
provide a practical framework to assess and improve public datasets.
Six attributes of ‘good’ data
Attribute What ‘Good’ Looks Like
Accuracy
Names, codes, and coordinates match the trusted source on the first try, no corrections
needed.
Completeness
Every required field is filled so the record can trigger a payment, call, or report without
chasing extra details.
Consistency
The same person or place carries identical values wherever it appears, no double spellings,
no age mismatch
Timeliness
Records are refreshed quickly enough to drive the intended action: minutes for payments,
days for planning, never “stale.”
Validity
Every entry passes format and logic checks—dates in range, PIN codes that exist, IDs that
clear checksums.
Uniqueness
One row equals one real-world entity; duplicates are flagged and merged before numbers
go to the dashboard.
Defining “good” is only half the battle; But even the clearest definition needs visibility to be useful.
A simple Data-Quality Scorecard does precisely that. Designed for programme officers, MIS teams,
and data stewards, the scorecard transforms the six attributes into a one-page, colour-banded
tracker that shows, quarter by quarter, where a dataset meets the mark, where gaps persist, and
who is responsible for each fix. Departments can adapt the template and instantly gain the visibility
they need to prioritise corrections and hold teams accountable.
Here’s a sample structure that departments can adapt to monitor key dimensions of data quality.
Sample Data Quality Scorecard
Illustrative example for programme registries or sectoral databases
Attribute DefinitionIndicator Example
Target
Threshold
Current
Performance
Accuracy
Correctness of data compared
to real-world values
% of beneficiary bank-account records that
pass NPCI’s Account Verification Service
(IFSC + account-number check)
≥ 98% 96.40%
Completeness
Presence of all required fields
and records
% of records with non-null mobile number
and bank details
≥ 95% 89.70%
Consistency
Internal coherence across fields
and systems
% of matched age entries across health and
education databases
≥ 90% 81.20%
Timeliness
Data is up-to-date and reflects
current status
Average time lag between scheme
approval and registry update (days)
≤ 3 days 6.1 days
Validity
Data conforms to expected
format, range, and logic
% of records passing format validation rules≥ 99% 97.80%
Uniqueness
Each real-world entity appears
only once in the dataset
% of duplicate household IDs identified in
deduplication audit
≤ 1% 3.50%
Note: The values shown in the “Current Performance” column are illustrative dummy figures, meant only to
demonstrate how you can use this scorecard to benchmark and compare your own real metrics. Adjust each cell
with your actual data when you run an assessment. India’s Data Imperative:
The Pivot Towards Quality 8
By turning abstract principles into trackable metrics, the scorecard makes quality visible—and
therefore actionable.
Used consistently, the scorecard turns abstract ideals into visible gaps—making data quality a
problem that teams can see, own, and fix.
And once the scorecard makes the red flags impossible to ignore, the next question is obvious: can
the organisation keep data healthy—day in, day out—without endless fire drills?
High-quality systems do more than patch errors; they bake safeguards into every step: structured
capture, schema controls at rest, real-time checks in motion, and planned archival. The goal
is persistence, not perfection. No stack of audits can rescue accuracy if the people who rely on a
record can’t see and correct it themselves.
Durable quality follows only when roles are clear, incentives reward fidelity, and platforms prevent
decay by design.
That’s where the Data-Quality Maturity Framework comes in
• What it is: a four-pillar self-assessment—Governance, Validation, Monitoring,
Integration—that rates each pillar from ad-hoc to best practice.
• Who uses it: department heads, nodal officers, and cross-functional data leads.
• Why it matters: an annual (with mid-year pulse) check turns “fix the data” into a
timed, budgeted roadmap—fund a steward, automate rule engines, adopt shared
schemas—rather than vague good intentions.
With scorecards tracking today’s vitals and the maturity framework securing tomorrow’s health,
agencies gain both a dashboard and a maintenance plan for trustworthy public data.
Data Quality Maturity Framework
A self-assessment guide for public institutions
How to Use This Framework:
Self-assess each row by identifying which description best matches your current state.
Use the result to identify gaps and prioritise improvements; e.g., move from “Emerging” to “Established” over the next year.
This tool can be applied at the department, scheme, or even platform level (e.g., PMAY, eShram, DigiLocker).
It can be used to guide capacity-building, budget planning, and platform redesigns.
Dimension
Level 1:
Foundational
Level 2:
Emerging
Level 3:
Established
Level 4:
Advanced
Level 5:
Institutionalised
1. Data
Governance &
Ownership
No defined
data owners or
custodians
Named data focal
points exist, but
roles are informal or
limited in scope
Department-level
data stewards
identified
with clear
responsibilities
Cross-department
governance with
coordination
mechanisms in
place
Stewardship embedded
in institutional roles;
linked to performance
and budgets
2. Standards &
Metadata
Data fields and
formats vary
across systems
Some common
formats used, but
documentation is
sparse
Data dictionaries
and naming
conventions
adopted within
departments
Cross-sector
metadata
standards adopted;
documented and
updated regularly
Standards
institutionalised across
levels of government;
aligned with national
frameworks
3. Data Capture
& Validation
Controls
Minimal
validation at
entry; errors are
common and go
undetected
Basic field-level
checks (e.g., format,
mandatory fields) in
place
Automated
validation and
duplication
checks used
during data entry
Real-time validation
against reference
data; alerts and
escalation built in
Validation built into
service design; field staff
trained and supported
to capture clean data India’s Data Imperative:
The Pivot Towards Quality 9
Dimension
Level 1:
Foundational
Level 2:
Emerging
Level 3:
Established
Level 4:
Advanced
Level 5:
Institutionalised
4. Quality
Monitoring &
Reporting
No routine
monitoring;
quality issues
identified only
during audits
Ad hoc reviews;
some manual
corrections made
Periodic checks
using scripts,
reports or
third-party
assessments
Dashboards track
quality metrics (e.g.,
error rates, missing
fields)
Data quality KPIs
tracked and published
regularly; linked to
planning and delivery
reviews
5. Correction
& Feedback
Loops
Corrections
made on
request; no
formal process
Manual correction
workflows exist but
are slow or informal
Designated
process for
correction
and grievance
redressal
Feedback loops
exist from users
to data owners;
tracking of
correction status
Continuous feedback
integrated into
workflows; audit trails
and accountability
mechanisms in place
6. Interoperability
& Integration
Systems operate
in silos; no data
exchange across
departments
Bilateral data
exchanges through
manual reports or
shared files
APIs exist
between key
systems; limited
real-time data
flows
Federated
systems with
shared identifiers
and automated
exchange
Seamless, consented
data flow across
services; common
reference data and
vocabularies shared at
scale
7. Culture &
Capacity
Data quality
not prioritised;
limited
awareness or
training
Staff recognise the
need but lack tools
or time to act
Regular training
and quality
awareness
activities
underway
Quality practices
seen as core
to programme
performance
Culture of data
ownership; quality is
part of organisational
identity and decision-
making
The next question is how far each dataset already stands from these targets
Institutions can use this framework to move from reactive fixes to a planned, progressive approach
to data quality, aligned with their own context and capacity. This is not a ranking exercise, but a
diagnostic aid. It helps organisations identify gaps, prioritise actions, and create a shared language
around what it means to build data systems that are not only functional, but resilient and future-
ready.
Used together, the scorecard and maturity framework turn firefighting into foresight—exposing
blind spots, ordering fixes, and giving every team a shared language for resilient, future-ready
data. Yet diagnosis is only the prologue. The next step is turning insight into quick, verifiable wins.
Chapter 5—the Starter Kit— distils proven tactics that any programme can pilot rapidly, translating
“what good looks like” into measurable improvements on the ground.
5. STARTER KIT: QUICK WINS AND TACTICAL PATHWAYS
With a shared yard-stick is in hand—the six core attributes, a scorecard to flag red cells, and a
maturity lens to show whether fixes will stick—the next question is what to tackle first.
Drawing on proven enterprise playbooks, three field-ready moves stand out—Fix it at the Source,
Keep it Clean, and Make it Matter. Each can be rolled out in under a quarter and tracked on the
existing scorecard. No sweeping rewrites or new legislation are required—just light tweaks to
capture screens, monitoring cycles, and accountability loops that any public programme can pilot
and scale.
5.1 Fix it at the source
(Focus on validation, deduplication, and metadata)
A. Real-time validation during data entry
Most errors originate at the point of capture. Preventing them upfront saves time and
effort later.
• Enforce field-level checks (e.g., PIN code format, date logic).
• Use dropdowns or autocomplete for standard entries (e.g., district names, job
categories). India’s Data Imperative:
The Pivot Towards Quality 10
• Flag missing mandatory fields before submission.
Some digital identity platforms now include real-time validation features, like auto-
formatting fields, verifying entries against known datasets, or flagging duplicates, to catch
errors before submission.
B. Deduplicate registries using open-source tools
Duplicates inflate counts, misdirect benefits, and complicate audits.
• Use lightweight tools (like Dedupe.io or Python-based matching) to flag probable
duplicates.
• Prioritise deduplication before scheme expansion or transfers.
C. Publish and maintain a basic data dictionary
Ambiguity in field definitions creates inconsistency in data entry and analytics.
• List field names, accepted formats, mandatory/optional status, and update logic.
• Share internally and with vendors or state-level partners.
When organisations use standardised field definitions and data templates, they reduce
ambiguity at the point of capture, making downstream data cleaner and easier to work
with. Simpler the data is to understand, the more useful it becomes.
5.2 Keep it clean
Audits, stewards, and correction workflows
A. Conduct regular, sample-based data audits
Frequent, light-touch reviews can catch minor issues before they compound.
• Sample recent entries (e.g., 5–10%) each month for key fields.
• Check for blanks, invalid values, or mismatches.
• Document recurring patterns for system or training updates.
In many settings, even basic monthly reviews of key fields have helped reduce avoidable
errors and improve staff awareness.
B. Assign internal data stewards
Designated points of contact reduce ambiguity and improve responsiveness.
• Identify one staff member per dataset or registry.
• Give them edit rights and visibility on known issues.
• Review data quality performance as part of regular team meetings.
C. Link grievance redressal with backend correction
Citizens often notice errors before systems do.
• Add “Report a mistake” options on portals or SMS updates.
• Route flagged issues to the correction workflows.
• Notify users once the update is complete.
Some grievance redressal systems now link user complaints directly to backend databases,
allowing reported issues to trigger updates in service records, not just status tickets. India’s Data Imperative:
The Pivot Towards Quality 11
5.3 Make it matter
Dashboards, recognition, and feedback loops to make data available in a highly accessible
manner.
A. Use dashboards to track quality, not just outputs
Service delivery dashboards often monitor coverage, but not data integrity.
• Add indicators like % of blank fields, error rates, and duplication flags.
• Show trends over time by district or block.
• Highlight gaps for targeted support.
B. Recognise and reward data quality improvements
Positive reinforcement builds a culture of care.
• Highlight top-performing blocks in monthly reviews.
• Publicly acknowledge field teams with high-quality submissions.
• Include data quality in annual appraisals or incentive frameworks.
C. Run a ‘Data Quality Week’ as an internal campaign
Short, focused campaigns help rally attention and clean up long-standing issues.
• Set goals (e.g., fill missing phone numbers, remove dead records).
• Track and share progress across teams.
• Debrief and document what worked for future replication.
These interventions do not require new laws, budgets, or platforms. They need ownership, clarity,
and consistent follow-through. They’re a way to start today, and build habits that last. When
repeated and scaled, small moves like these shift the culture around data. They reduce the cost of
correction. They make systems more responsive. And they signal to staff, leadership, and citizens
that quality is not optional. It is essential. Over time, these efforts lay the foundation for a shared
culture of quality that must be reinforced, coordinated, and made visible across institutions. The
next step is to move from individual habits to collective norms, where quality becomes a shared
expectation across systems, institutions, and levels of government.
But to turn quick wins into lasting norms, data quality needs deeper roots—anchored in roles,
incentives, and systems that don’t reset with each programme cycle.
6. BUILDING BLOCKS FOR CHANGE: INSTITUTIONS,
INCENTIVES & INTEROPERABILITY
Improving data quality isn’t just a matter of better tools or tighter rules; it’s about shifting how
institutions think, behave, and collaborate. The previous chapters laid out what quality looks like
and how to measure it. But making quality last requires something more enduring: ownership,
incentives, and infrastructure that support good data as a habit, not a one-off intervention. This
chapter offers three practical levers to institutionalise quality from within the system.
6.1 Institutionalising Ownership
Quality suffers when data has no clear owner. In many programmes, data moves across
departments and levels, but no single entity is responsible for maintaining its integrity from
end to end. The fix lies in assigning stewardship, not just for compliance, but for ongoing
care. This means designating custodians at each level (national, state, district) who are India’s Data Imperative:
The Pivot Towards Quality 12
empowered and accountable for data health. These roles don’t need to be new hires; they
can be embedded within existing structures with a defined mandate, clear escalation paths,
and regular review cycles.
In addition, quality must be seen as a shared responsibility. Programme heads, IT teams, and
field staff all have roles to play. When Ownership is structured and supported, errors don’t
just get corrected—they get prevented.
Under the U.S. Foundations for Evidence-Based Policymaking Act (2019), every federal
agency was required to designate a Chief Data Officer responsible for data governance and
quality. By October 2022, the Government Accountability Office reported that 82 agencies
had officially appointed CDOs, embedding stewardship at the heart of their data operations.
7
6.2 Getting Incentives Right: Rewarding Quality, Not Just Speed
Most public programmes reward what they can count: enrolments completed, forms
submitted, and dashboards updated. These indicators are useful, but they can create
unintended consequences. When speed is rewarded but accuracy is not, field staff may rush
through data entry. When outputs matter more than inputs, verification becomes a burden
rather than a standard. Data quality indicators—such as error rates, completion levels, and
timeliness—can be tracked and factored into programme reviews, not as punitive audits but
as a reflection of delivery strength.
Some states have implemented scorecards to rank departments based on data quality,
fostering a constructive sense of competition. Others have tied data improvements to budget
flexibility or expedited approvals. Although the approaches may differ, the fundamental
principle remains the same: when quality is measured, recognized, and incentivized, it
becomes significant.
6.3 Making Systems Work Together: The Case for Interoperability
Even clean data loses value if systems can’t use it together. Interoperability—across platforms,
departments, and time—is essential to making public data work harder. It depends on
shared schemas, reference codes, and APIs that allow secure, controlled access between
programmes.
India’s toolbox is steadily growing:
• IndEA provides a common vocabulary and reference architecture.
• NDGFP sets rules for sharing, consent, and anonymisation.
• Account Aggregator, DigiLocker, DEPA show how consent-driven flows can
reduce duplication.
• Community-led tools—from CDPI to iSPIRT—offer off-the-shelf templates, APIs,
and governance playbooks.
Still, most systems remain siloed, relying on PDFs or manual copy-paste. The next step is
scale: making these standards the default, not the exception. That means using common data
models where possible—for people, households, land parcels—and building new systems on
existing rails rather than starting from scratch.
Singapore’s Government Data Architecture shows what’s possible: structured datasets, clear
access rules, shared protocols. India already has the building blocks. The challenge now is
connection—between systems, teams, and incentives. That’s what makes quality stick. India’s Data Imperative:
The Pivot Towards Quality 13
7. THE ROAD AHEAD: TOWARDS A SHARED CULTURE OF DATA EXCELLENCE
India’s digital foundations are in place. The challenge now is ensuring that what flows through
them—our data—is accurate, complete, and trusted. Data quality is no longer a backend concern;
it is central to public trust, effective service delivery, and the success of India’s own AI ecosystem.
But this isn’t a problem technology alone can solve. True change requires a cultural shift—where
quality is everyone’s responsibility, not just the IT team’s. From frontline staff to policymakers, every
actor must see themselves as a data steward.
This shift must be supported by light-touch, collaborative mechanisms: voluntary scorecards, peer-
learning networks, flexible incentives, and playbooks of best practices that evolve with experience.
These resources—maintained by a neutral body like NITI Aayog or the proposed India Data
Management Office—should blend technical, institutional, and behavioural insights from across
government and sectors.
States can lead the way by embedding data quality cells, linking quality to service outcomes, and
recognising excellence. Capacity building, leadership development, and hands-on support will be
essential to sustain momentum.
Above all, this transformation needs visible commitment from the top. A government that champions
clean, usable data sends a powerful signal: that quality is not an afterthought—it is the foundation
of good governance.
India’s digital future will be shaped not just by how many platforms we build, but by how much trust
we build into them. And that begins with data that’s ready to serve.
REFERENCES:
1. https://www.financialexpress.com/business/banking-finance-upi-transactions-up-34-on-year-
in-april-down-a-tad-from-march-3829424/
2. https://economictimes.indiatimes.com/news/india/aadhaar-authentications-touch-2707-crore-
in-2024-25/articleshow/120703093.cms?
3. https://www.linkedin.com/pulse/24th-march-2025-over-369-crore-ayushman-cards-have-
m91vc/
4. https://www.business-standard.com/india-news/centre-to-save-rs-18-000-cr-by-removing-
fake-accounts-from-welfare-schemes-123101700528_1.html?
5. https://timesofindia.indiatimes.com/business/india-business/govt-sraps-rs-3-5-crore-bogus-
lpg-connections-saves-rs-21000-crore-subsidy-in-2-years/articleshow/53326003.cms?
6. https://cag.gov.in/uploads/download_audit_report/2023/Report-No.-11-of-2023_PA-on-
PMJAY_English-PDF-A-064d22bab2b83b5.38721048.pdf ?
7. https://www.gao.gov/assets/gao-23-105514.pdf India’s Data Imperative:
The Pivot Towards Quality 15
The Pivot Towards Quality 1 India’s Data Imperative:
The Pivot Towards Quality
How high-quality data can unlock better outcomes, build public
trust, and power the next decade of digital governance
Knowledge Partner: India’s Data Imperative: The Pivot Towards Quality
I
am delighted to present this edition of the NITI Aayog Frontier
Technology Hub’s Quarterly Insight on “India’s Data Imperative:
The Pivot Towards Quality”. This timely report underscores a
crucial shift in India’s data journey – from achieving unprecedented
scale to ensuring uncompromising precision and trustworthiness
in our data.
Over the past decade, India has truly distinguished itself as
a global leader in digital public infrastructure. The Unified
Payments Interface (UPI) moves trillions of rupees monthly,
Aadhaar has authenticated billions of identities, and Ayushman
Bharat has extended healthcare access to more than 350 million
citizens. These are not merely statistics; they are testaments
to a foundational transformation, enabling financial inclusion,
empowering citizens, and streamlining welfare delivery on a scale
previously unimaginable. Our collective efforts have laid robust digital
rails, demonstrating to the world how technology can empower a
nation of 1.4 billion.
However, as we embark on the next phase of our digital evolution, the
focus must decisively pivot towards the quality of the data that fuels
these incredible platforms. As this report highlights, data quality is
no longer a peripheral technical consideration but a fundamental
prerequisite for good governance and sustained public trust. When
a single erroneous digit can delay a benefit, or a duplicate record
inflates welfare outlays, the true cost of poor data becomes painfully
apparent – impacting budgets, distorting policy, and eroding the
faith citizens place in our digital systems. The move from “scale to
precision” is not just an aspiration; it is a national imperative.
I particularly commend the inclusion of two practical, easy-to-use
tools in the report: The Data-Quality Scorecard and the Data-
Quality Maturity Framework that are designed to enable every
department to conveniently assess their current data landscape,
identify gaps, and chart a clear roadmap for improvement. Whether
it is implementing real-time validation at the source, assigning data
stewards, or linking grievance redressal to backend corrections,
the “Starter Kit” outlined in this report offers tangible pathways to
achieve quick, verifiable wins.
This report serves as an invaluable guide for all government Ministries
Foreword A
s India matures as a Digital Economy, we must increasingly
prioritize a critical imperative: data quality. The shift
from scale to precision makes high-quality data not just
desirable, but essential. Even a single incorrect digit can halt a
pension, deny healthcare, or misdirect subsidies, profoundly
eroding citizen trust and costing billions in fiscal leakage.
Data quality must therefore become a frontline governance
imperative, integral to building Digital India on a robust foundation
of trust. Trust is not merely a desirable outcome—it is the bedrock
upon which our digital ambitions stand and scale.
This edition of NITI Frontier Tech Hub’s Quarterly Insights,
Future Front, developed with our knowledge partner Gramener,
emphasizes the urgent need to radically transform our approach to
data quality. Strengthening data quality is foundational to realizing
our digital ambitions—whether enabling AI-driven governance,
ensuring targeted welfare delivery, or driving cross-sector innovation.
It requires collective action, clear accountability, and a shared
commitment from every department and ministry.
DEBJANI GHOSH
Distinguished Fellow, NITI Aayog;
Chief Architect, NITI Frontier Tech Hub
and Departments. It meticulously dissects the challenges across the
entire data value chain – from generation and storage to sharing,
use, and retirement. I urge the Ministries and Departments across
the Government of India and the State Governments to thoroughly
engage with this report. The insights and recommendations in the
report are aimed at embedding a culture of data excellence – one
where quality is not an afterthought but a shared responsibility,
deeply ingrained in our institutional roles, incentivized effectively,
and supported by robust interoperability.
Our digital future will be defined not just by the platforms we build,
but by the unwavering trust we cultivate in them. This trust begins
with data that is precise, reliable, and truly ready to serve. Let us
champion clean, usable data as the foundation of good governance,
ensuring that the dividends of our digital journey reach every citizen
with precision and equity.
DR. SAURABH GARG
Secretary, MoSPI India’s Data Imperative:
The Pivot Towards Quality 1
EXECUTIVE SUMMARY:
India’s digital public infrastructure has grown at a breakneck speed. UPI now moves trillions of
rupees each month, Aadhaar verifies billions of identities, and Ayushman Bharat touches half a
billion citizens. As data volumes explode, it is imperative for data quality to keep pace across both
the torrents generated today and the decades of legacy records still in use. Records need to be
precise, complete, and up-to-date by reducing or eliminating gaps, duplicates, or timing lags that
may exist today. When one wrong digit can freeze a pension or misroute a subsidy, quality is no
longer a technical afterthought but a frontline service obligation. India’s digital governance will only
be as strong as the quality of the data that powers it. As India moves from scale to precision, data
quality must become a national priority—on par with infrastructure and platforms.
Why Data Quality Matters:
1. Fiscal leakage, erroneous or duplicate beneficiary records drain budgets, inflating
welfare outlays by an estimated 4–7 per cent annually.
2. Policy blind spots, inconsistent, non-standardised datasets distort evidence, leading to
mis-targeted schemes or delayed course-corrections.
3. Public trust erosion, citizens confronting mismatched records, rejected claims, or
endless rectification queues lose confidence in digital governance.
This edition of the NITI Frontier Tech Hub Quarterly Insight, created in partnership with Gramener,
lays out data-quality challenges across the full data value chain, analyses their root causes, and
offers two easy-to-use tools to measure data quality and maturity in any department.
1. SETTING THE STAGE: INDIA’S DIGITAL & DATA MOMENT
India’s digital public infrastructure now operates at a scale few imagined a decade ago. In April
2025, the Unified Payments Interface (UPI) processed 17.89 billion transactions worth ₹23.9trillion,
1
rivalling the monthly GDP of several mid-sized economies. Aadhaar authenticated 27.07 billion
identity requests in FY 2024-25,
2
confirming its place as the common key for banking, welfare and
a widening array of private-sector services. On the social-protection front, more than 369 million
Ayushman cards are now in circulation,
3
bringing cashless hospital care to roughly half a billion
citizens.
But the next test is precision, not reach. What if just 5 % of records are flawed, and those belong
to our most vulnerable citizens? The true test of digital infrastructure is how precisely it represents
each individual, not just how many people it onboards. As platforms mature, the focus must shift
from expansion to fidelity. Digital highways only create public value when every record is accurate,
complete, and current. And no stack of audits can rescue accuracy if the people who create or rely
on a record have no sight or stake in it. Confidence endures only when the people who depend on
a record can also see and correct it.
1.1 Small errors carry huge costs
A single wrong digit during Aadhaar enrolment can block a pension. A duplicated mobile
number can stall a hospital claim. A misspelt name in a land record can delay compensation
after a natural disaster.
Multiply such slips across several central schemes and thousands of state programmes and
the fiscal dent is staggering. Every wasted rupee crowds out an honest claim; every wrong
dashboard alert delays a life-saving decision. Bad data costs real lives and real money. India’s Data Imperative:
The Pivot Towards Quality 2
1.2 From scale to fidelity
Over the last decade India laid the rails, identity, payments, registries, that make inclusive
innovation possible. The coming decade’s services, farm-credit via account aggregators,
portable skill credentials, AI-powered health alerts, depend on the integrity of the data flowing
over those rails. Where records are rigorous at the moment of capture, new services can
scale at marginal cost; where they are inconsistent, each added layer inherits and amplifies a
structural deficit.
Recent clean-ups underline the pay-off
ActionEstimated Savings
Deleted 17.1 m ineligible PM-Kisan names
4
₹90 billion (FY 2024)
Weed out 35 m bogus LPG connections
5
₹210 billion (in two years)
Drop 16 m fake ration cards
6
~₹100 billion per year
1.3 What holds us back
Across ministries and states, platforms are built as separate silos that store data in clashing
formats and use incompatible identifiers, forcing even routine joins to be stitched together
by hand. Many of these systems run on ageing back-ends that lack basic validation rules,
audit trails or version control, so a small tweak in one module can topple an entire workflow.
Because records shuttle between teams without a named custodian, obvious errors linger
unchecked, and field programmes are often driven by enrolment targets that prize speed
over accuracy. Over time this breeds a “good-enough” culture, once 80 percent accuracy
is accepted as normal, the system quietly stops trying to prevent error and the data quality
debt compounds.
1.4 Towards “data confidence”
Building confidence begins at the point of capture: automated checks and standard pick-lists
must stop bad entries before they enter the system. Every high-value dataset then needs a
clearly identified steward—someone empowered, and accountable, to correct and continually
improve the records. Finally, common schemas for people, places and programmes must be
adopted so that data can travel seamlessly across agencies without constant translation.
Together, these three practices turn today’s fragmented information into a reliable asset that
analytics, AI and service delivery can use without sacrificing trust.
The pivot from scale to fidelity demands a new scorecard—one that values precision and
trust as highly as reach.
The table below shows how priorities must shift, toward iterative architecture, stronger
interoperability rails, and shared stewardship roles for AI-ready data.
Scale EraFidelity Era
Number of users onboardedPrecision in every record
Platform speed and uptimeTrust in data at every stage
Build core digital public goods, identity,
payments, registries
Iterate architecture, strengthen interoperability
rails, and embed shared stewardship roles
Without quality, data is just noise.
India has proved it can scale digital public goods. Showing the world we can sustain high-
fidelity data will cement India Stack as a governance model, not just a technology export, India’s Data Imperative:
The Pivot Towards Quality 3
and ensure the first dividends reach those who need them most. Getting there will demand
as much organisational reform as technology: frontline training, budget lines for data
stewardship, and lightweight policy guardrails that make quality a default requirement rather
than a discretionary upgrade.
2. SEEING THE WHOLE ELEPHANT: THE PUBLIC SECTOR DATA VALUE CHAIN
A farmer’s subsidy application, a child’s vaccination record, or a woman’s property title all
start as a few keystrokes or biometric scans. Those fragments soon underpin service delivery
and policy. Between capture and a final decision, data travels an often-hidden chain - stored,
transmitted, transformed, analysed, and eventually archived. Quality can slip at any link, with
losses that snowball for citizens and the agencies that rely on the record.
These five stages of the data value chain – generation, storage, sharing, use, and retirement –
present distinct challenges and play a role in shaping data quality. We’ll illustrate each stage
with Lakshmi Devi’s PM-Kisan record, a single row that travels the chain.
2.1. Generation - Where It Begins
Most data-quality faults are born at the moment of capture, whether a field agent taps details
into an enrolment app, a clerk uploads a survey sheet, or a sensor pushes logs to the cloud. In
resource-tight settings a single typo in a name, Aadhaar, or bank IFSC can ricochet through
every downstream system, blocking payments, skewing analytics, and triggering costly
reconciliations. Fixing them later often requires time-consuming manual reconciliations and
frustrated citizens.
Example - At the village service centre, Lakshmi Devi’s PM-Kisan details were entered with
two digits of her bank IFSC transposed. Because the enrolment screen showed only a
confirmation tick, neither she nor the officer noticed the slip.
2.2. Storage: The Quiet Fade of Confidence
Once generated, data must be stored in a way that preserves its structure, context, and
integrity. This includes defining how data is formatted, tagged, and secured; how it is
updated over time; and how different departments manage their own databases. In practice,
storage practices are fragmented. Some systems now employ cloud-native, version-controlled
stores, while others still depend on legacy databases with no audit trail.
Example - When the day’s data synchronises, Lakshmi’s record is copied from the enrolment
app to the state PM-Kisan server and then to a legacy backup database. Because neither
system runs cross-check or version control on incoming fields, the transposed IFSC is stored
unchanged in both locations, quietly duplicating the error.
2.3. Sharing: When Silos Collide
Sharing data across ministries, departments, or platforms is where digital governance can truly
shine and where friction most often flares. Crucially, inconsistency is not only horizontal (one
ministry versus another) but also vertical (different schemes, dashboards, or district offices
inside the same programme). A pension portal may store household size at the family level,
while a village survey logs each individual; a health-worker app records a child’s age in months,
yet the state immunisation dashboard expects years. The grain of data, the frequency of
refresh, and even the date formats vary wildly, turning simple joins into costly clean-ups that
stall service delivery.
Example - At quarter-end, the agriculture portal exports beneficiary data to the treasury
gateway for bulk transfers. Lakshmi’s row fails the bank’s validation check - “invalid IFSC” India’s Data Imperative:
The Pivot Towards Quality 4
—and is returned as an exception. Her payment is held until staff review the file, correct the
code, and resubmit the batch.
2.4. Use: Where the Stakes Get Real - and the Rewards too
When government data feeds a service queue, a policy dashboard, or an AI model, any
upstream flaw becomes a public-facing fault. Clean, time-stamped records can power tools
that flag disease outbreaks, fine-tune subsidy targeting, or route traffic in real time. But if two
datasets record farms at different granularities or age a child in months versus years, even
the smartest algorithm will misfire.
Architecture matters too: a single data lake risks breaches, while sealed silos hide cross-
domain patterns. Newer methods—federated learning, synthetic data, privacy-enhancing
computation, let models train where data sits, but only if every record carries consistent
schemas, consent flags, and update logs. In short, garbage in, algorithmic garbage out:
disciplined capture, harmonised storage, and governed sharing are the non-negotiable tripod
beneath any citizen-centric AI.
Example - The programme manager builds a drought-relief model that relies on “successful
payment” history to spot cash-constrained clusters. Because Lakshmi’s transfer shows as
failed, her village appears slightly better off than it is, nudging fertiliser stock toward other
blocks.
2.5. Retirement: The Data We Forget
Archival is the stage most systems skip. With no clear “sunset” rules, obsolete entries linger,
inflating counts, slowing queries, and sometimes exposing private details. Welfare rolls still
list families that have moved away; backup servers hold duplicates long after production
tables are cleaned.
Fixing this is less about storage and more about policy: time-bound retention schedules,
annual “alive-and-eligible” checks, and safe deletion once legal purpose lapses. Regular
pruning keeps databases lean, dashboards truthful, and public trust intact, freeing analysts
to focus on trends that matter, not ghosts in the machine.
Example - Six months later a data-quality drive updates Lakshmi’s IFSC in the main server,
but the duplicate row in an older backup is missed. Unless that orphan entry is archived
or reconciled, it will keep resurfacing during audits and quarterly reconciliations, requiring
manual checks each time.
Together, these five stages show how one mistyped field can ripple from capture to archive.
The next chapter tallies the real-world cost of letting those cracks persist, and why they
demand urgent repair.
This journey reveals how cracks can form at any link—and why fixing them demands a closer
look at the systems behind the data.
3. THE NATURE OF THE CHALLENGE
India’s public data engines hum louder each year, powering cash transfers, hospital authorisations,
fertiliser caps, and pandemic dashboards. Yet familiar cracks reappear: errors slip through, gaps
don’t close, fixes don’t hold. These aren’t stray glitches, they stem from how data is captured,
shared, and rewarded.
This chapter identifies six recurring design flaws, pairs each with a real-world incident, and highlights
the risks associated with leaving them unaddressed. India’s Data Imperative:
The Pivot Towards Quality 5
3.1 Systemic, not Isolated
Data quality lives or dies in system design. Every dataset, whether a 1995 ledger or a 2025
API stream, inherits the roles, incentives and validation rules baked into its first mile. When
incentives favour quantity over correctness, mistakes recur, no matter how many downstream
audits we conduct.
Example - During a 2017 linkage drive, three regions uncovered 4.4 lakh “ghost
students” claiming midday meal funds. In one state alone, more than two lakh fictitious
names vanished once biometric attendance became mandatory, proof that lasting quality
follows when accuracy is rewarded.
Risk: phantom beneficiaries drain budgets and trust every time a new scheme launches.
No stack of audits can substitute for giving the record’s true owner - teacher, farmer, patient -
live visibility and the right to correct it. The “two screen” strategy used at Aadhaar enrolment
centres – exposing the data to the Aadhaar holder to view and correct the record – is an
excellent example of implementing this.
3.2 Fragmented by Design
Platforms often start with good intentions but end up maturing in isolation. A ministry
launches a portal tailored to its flagship scheme; states then clone and tweak that code
to meet local rules; over time, each branch adds its own columns, codes, and shortcuts.
Formats drift apart, hectares in one table, acres in another, full names in one file, initials in the
next, dates logged as DD/MM/YYYY in one district but MM-DD-YY next door. What should
be a simple join across systems turns into days of manual cross-walks and spreadsheet
contortions, delaying benefits and muddying analysis.
Example - In 2021, roughly one-third of crop insurance claims in an eastern district stalled
because land titles listed parents as owners while crop-loss forms named the farmers
themselves. About 1.5 lakh applications sat idle until staff reconciled the two schemas
Risk: fractured views create blind spots and delay help when citizens need it most.
3.3 Legacy Systems, Modern Pressures
Many core registries still run on infrastructure built for a different era. They’ve been patched
over the years, but the core design remains unchanged. Basic features, such as validation
logic, audit trails, or version control, are often missing. Dashboards may have been added to
improve visibility, but they sit atop brittle backends. As policies evolve and field realities shift,
these systems struggle to keep pace. Even minor changes, such as updating a beneficiary
status, can require custom workarounds.
Example- In 2024, elderly ration holders in a western district queued for months after
fingerprint readers failed to recognise worn prints. Iris scanners arrived late, but thousands
missed subsidised grain at the peak of inflation.
Risk: brittle systems turn minor tweaks into outages for households on the margin.
3.4 Everyone’s Job, No One’s Responsibility
Data moves across departments and systems but rarely comes with a clear owner. One team
collects it, another uses it, and a third reports on it. When errors show up, it’s not always clear
who is responsible for correcting them or whether anyone has the authority to do so. This
lack of custodianship means even obvious mistakes can persist for years, not because they’re
hard to spot but because they fall between the cracks of institutional roles. India’s Data Imperative:
The Pivot Towards Quality 6
Example - In late 2022, 17,000 health insurance cards were stuck because beneficiary names
or biometrics failed identity checks. Health staff labelled it an “ID problem,” the ID team saw
a “health-database issue,” and elderly citizens paid cash until a joint task force cleaned the
queue.
Risk: obvious errors linger for years, eroding faith and loading staff with avoidable rework.
Industry practice shows that durable quality emerges when systems expose errors to the
people who benefit most from clean data, instead of layering extra audits.
3.5 When Speed Trumps Quality
Most programmes are evaluated by how fast they move: how many people were enrolled,
how quickly benefits were delivered, and how soon dashboard indicators turn green.
These indicators are helpful, but they can create unintended consequences. When speed is
rewarded, but accuracy is not, frontline staff may rush through data entry. Mid-level officials
may hesitate to reopen past data for fear of triggering audit flags. Over time, this creates a
culture where getting it done is more important than getting it right.
Example- During a large LPG subsidy roll-out in 2013, ID bank linkages covered only 60 per
cent of households; the remaining 40 per cent were bulk-rejected to keep the launch on
schedule. It took two years of painstaking review to unlock the full savings—showing that
haste can cost more than delay.
Risk: performance metrics turn into vanity stats, hiding debts that grow costlier over time.
3.6 The Cost of Low Expectation
Once systems learn to live with error, they stop trying to prevent it. A database that’s 80%
accurate becomes “good enough.” A mismatch between two registries is seen as a minor
technical hiccup rather than a barrier to service delivery. This mindset shapes everything:
how platforms are designed, how staff are trained, and how performance is judged. Slowly,
quietly, quality becomes optional.
Example- A large northern state declared itself open-defecation-free in 2019; however, an
audit of 590 rural houses found that nearly half lacked toilets. Dashboards still showed 100
per cent coverage, and funds kept moving—evidence that data judged “close enough” can
override field reality.
Risk: policy rides on numbers no one fully trusts, undercutting evidence-based governance.
These six flaws reveal a design defect, not a tech shortfall. Quality will improve only when
priorities shift from speed to precision, ownership, and aligned incentives. The next chapter
calculates the cost of persistent cracks in rupees, including delays and the erosion of public
trust. It explains why fixing them now is more cost-effective than patching them indefinitel.
To move forward, we need a shared benchmark of what quality truly means—beyond error
rates, into attributes that build trust. India’s Data Imperative:
The Pivot Towards Quality 7
4. WHAT GOOD LOOKS LIKE: ANATOMY OF HIGH-QUALITY
DATA - SIX ATTRIBUTES
Picture a GPS that shows every road but scrambles half the street names—you’d still get lost. Data
behaves the same way.
Before we can improve data quality, we must define it—clearly and measurably. Without a shared
understanding, every clean-up drive veers off course, leaving efforts inconsistent and results short-
lived. Six core attributes—accuracy, completeness, timeliness, consistency, validity, and uniqueness—
provide a practical framework to assess and improve public datasets.
Six attributes of ‘good’ data
Attribute What ‘Good’ Looks Like
Accuracy
Names, codes, and coordinates match the trusted source on the first try, no corrections
needed.
Completeness
Every required field is filled so the record can trigger a payment, call, or report without
chasing extra details.
Consistency
The same person or place carries identical values wherever it appears, no double spellings,
no age mismatch
Timeliness
Records are refreshed quickly enough to drive the intended action: minutes for payments,
days for planning, never “stale.”
Validity
Every entry passes format and logic checks—dates in range, PIN codes that exist, IDs that
clear checksums.
Uniqueness
One row equals one real-world entity; duplicates are flagged and merged before numbers
go to the dashboard.
Defining “good” is only half the battle; But even the clearest definition needs visibility to be useful.
A simple Data-Quality Scorecard does precisely that. Designed for programme officers, MIS teams,
and data stewards, the scorecard transforms the six attributes into a one-page, colour-banded
tracker that shows, quarter by quarter, where a dataset meets the mark, where gaps persist, and
who is responsible for each fix. Departments can adapt the template and instantly gain the visibility
they need to prioritise corrections and hold teams accountable.
Here’s a sample structure that departments can adapt to monitor key dimensions of data quality.
Sample Data Quality Scorecard
Illustrative example for programme registries or sectoral databases
Attribute DefinitionIndicator Example
Target
Threshold
Current
Performance
Accuracy
Correctness of data compared
to real-world values
% of beneficiary bank-account records that
pass NPCI’s Account Verification Service
(IFSC + account-number check)
≥ 98% 96.40%
Completeness
Presence of all required fields
and records
% of records with non-null mobile number
and bank details
≥ 95% 89.70%
Consistency
Internal coherence across fields
and systems
% of matched age entries across health and
education databases
≥ 90% 81.20%
Timeliness
Data is up-to-date and reflects
current status
Average time lag between scheme
approval and registry update (days)
≤ 3 days 6.1 days
Validity
Data conforms to expected
format, range, and logic
% of records passing format validation rules≥ 99% 97.80%
Uniqueness
Each real-world entity appears
only once in the dataset
% of duplicate household IDs identified in
deduplication audit
≤ 1% 3.50%
Note: The values shown in the “Current Performance” column are illustrative dummy figures, meant only to
demonstrate how you can use this scorecard to benchmark and compare your own real metrics. Adjust each cell
with your actual data when you run an assessment. India’s Data Imperative:
The Pivot Towards Quality 8
By turning abstract principles into trackable metrics, the scorecard makes quality visible—and
therefore actionable.
Used consistently, the scorecard turns abstract ideals into visible gaps—making data quality a
problem that teams can see, own, and fix.
And once the scorecard makes the red flags impossible to ignore, the next question is obvious: can
the organisation keep data healthy—day in, day out—without endless fire drills?
High-quality systems do more than patch errors; they bake safeguards into every step: structured
capture, schema controls at rest, real-time checks in motion, and planned archival. The goal
is persistence, not perfection. No stack of audits can rescue accuracy if the people who rely on a
record can’t see and correct it themselves.
Durable quality follows only when roles are clear, incentives reward fidelity, and platforms prevent
decay by design.
That’s where the Data-Quality Maturity Framework comes in
• What it is: a four-pillar self-assessment—Governance, Validation, Monitoring,
Integration—that rates each pillar from ad-hoc to best practice.
• Who uses it: department heads, nodal officers, and cross-functional data leads.
• Why it matters: an annual (with mid-year pulse) check turns “fix the data” into a
timed, budgeted roadmap—fund a steward, automate rule engines, adopt shared
schemas—rather than vague good intentions.
With scorecards tracking today’s vitals and the maturity framework securing tomorrow’s health,
agencies gain both a dashboard and a maintenance plan for trustworthy public data.
Data Quality Maturity Framework
A self-assessment guide for public institutions
How to Use This Framework:
Self-assess each row by identifying which description best matches your current state.
Use the result to identify gaps and prioritise improvements; e.g., move from “Emerging” to “Established” over the next year.
This tool can be applied at the department, scheme, or even platform level (e.g., PMAY, eShram, DigiLocker).
It can be used to guide capacity-building, budget planning, and platform redesigns.
Dimension
Level 1:
Foundational
Level 2:
Emerging
Level 3:
Established
Level 4:
Advanced
Level 5:
Institutionalised
1. Data
Governance &
Ownership
No defined
data owners or
custodians
Named data focal
points exist, but
roles are informal or
limited in scope
Department-level
data stewards
identified
with clear
responsibilities
Cross-department
governance with
coordination
mechanisms in
place
Stewardship embedded
in institutional roles;
linked to performance
and budgets
2. Standards &
Metadata
Data fields and
formats vary
across systems
Some common
formats used, but
documentation is
sparse
Data dictionaries
and naming
conventions
adopted within
departments
Cross-sector
metadata
standards adopted;
documented and
updated regularly
Standards
institutionalised across
levels of government;
aligned with national
frameworks
3. Data Capture
& Validation
Controls
Minimal
validation at
entry; errors are
common and go
undetected
Basic field-level
checks (e.g., format,
mandatory fields) in
place
Automated
validation and
duplication
checks used
during data entry
Real-time validation
against reference
data; alerts and
escalation built in
Validation built into
service design; field staff
trained and supported
to capture clean data India’s Data Imperative:
The Pivot Towards Quality 9
Dimension
Level 1:
Foundational
Level 2:
Emerging
Level 3:
Established
Level 4:
Advanced
Level 5:
Institutionalised
4. Quality
Monitoring &
Reporting
No routine
monitoring;
quality issues
identified only
during audits
Ad hoc reviews;
some manual
corrections made
Periodic checks
using scripts,
reports or
third-party
assessments
Dashboards track
quality metrics (e.g.,
error rates, missing
fields)
Data quality KPIs
tracked and published
regularly; linked to
planning and delivery
reviews
5. Correction
& Feedback
Loops
Corrections
made on
request; no
formal process
Manual correction
workflows exist but
are slow or informal
Designated
process for
correction
and grievance
redressal
Feedback loops
exist from users
to data owners;
tracking of
correction status
Continuous feedback
integrated into
workflows; audit trails
and accountability
mechanisms in place
6. Interoperability
& Integration
Systems operate
in silos; no data
exchange across
departments
Bilateral data
exchanges through
manual reports or
shared files
APIs exist
between key
systems; limited
real-time data
flows
Federated
systems with
shared identifiers
and automated
exchange
Seamless, consented
data flow across
services; common
reference data and
vocabularies shared at
scale
7. Culture &
Capacity
Data quality
not prioritised;
limited
awareness or
training
Staff recognise the
need but lack tools
or time to act
Regular training
and quality
awareness
activities
underway
Quality practices
seen as core
to programme
performance
Culture of data
ownership; quality is
part of organisational
identity and decision-
making
The next question is how far each dataset already stands from these targets
Institutions can use this framework to move from reactive fixes to a planned, progressive approach
to data quality, aligned with their own context and capacity. This is not a ranking exercise, but a
diagnostic aid. It helps organisations identify gaps, prioritise actions, and create a shared language
around what it means to build data systems that are not only functional, but resilient and future-
ready.
Used together, the scorecard and maturity framework turn firefighting into foresight—exposing
blind spots, ordering fixes, and giving every team a shared language for resilient, future-ready
data. Yet diagnosis is only the prologue. The next step is turning insight into quick, verifiable wins.
Chapter 5—the Starter Kit— distils proven tactics that any programme can pilot rapidly, translating
“what good looks like” into measurable improvements on the ground.
5. STARTER KIT: QUICK WINS AND TACTICAL PATHWAYS
With a shared yard-stick is in hand—the six core attributes, a scorecard to flag red cells, and a
maturity lens to show whether fixes will stick—the next question is what to tackle first.
Drawing on proven enterprise playbooks, three field-ready moves stand out—Fix it at the Source,
Keep it Clean, and Make it Matter. Each can be rolled out in under a quarter and tracked on the
existing scorecard. No sweeping rewrites or new legislation are required—just light tweaks to
capture screens, monitoring cycles, and accountability loops that any public programme can pilot
and scale.
5.1 Fix it at the source
(Focus on validation, deduplication, and metadata)
A. Real-time validation during data entry
Most errors originate at the point of capture. Preventing them upfront saves time and
effort later.
• Enforce field-level checks (e.g., PIN code format, date logic).
• Use dropdowns or autocomplete for standard entries (e.g., district names, job
categories). India’s Data Imperative:
The Pivot Towards Quality 10
• Flag missing mandatory fields before submission.
Some digital identity platforms now include real-time validation features, like auto-
formatting fields, verifying entries against known datasets, or flagging duplicates, to catch
errors before submission.
B. Deduplicate registries using open-source tools
Duplicates inflate counts, misdirect benefits, and complicate audits.
• Use lightweight tools (like Dedupe.io or Python-based matching) to flag probable
duplicates.
• Prioritise deduplication before scheme expansion or transfers.
C. Publish and maintain a basic data dictionary
Ambiguity in field definitions creates inconsistency in data entry and analytics.
• List field names, accepted formats, mandatory/optional status, and update logic.
• Share internally and with vendors or state-level partners.
When organisations use standardised field definitions and data templates, they reduce
ambiguity at the point of capture, making downstream data cleaner and easier to work
with. Simpler the data is to understand, the more useful it becomes.
5.2 Keep it clean
Audits, stewards, and correction workflows
A. Conduct regular, sample-based data audits
Frequent, light-touch reviews can catch minor issues before they compound.
• Sample recent entries (e.g., 5–10%) each month for key fields.
• Check for blanks, invalid values, or mismatches.
• Document recurring patterns for system or training updates.
In many settings, even basic monthly reviews of key fields have helped reduce avoidable
errors and improve staff awareness.
B. Assign internal data stewards
Designated points of contact reduce ambiguity and improve responsiveness.
• Identify one staff member per dataset or registry.
• Give them edit rights and visibility on known issues.
• Review data quality performance as part of regular team meetings.
C. Link grievance redressal with backend correction
Citizens often notice errors before systems do.
• Add “Report a mistake” options on portals or SMS updates.
• Route flagged issues to the correction workflows.
• Notify users once the update is complete.
Some grievance redressal systems now link user complaints directly to backend databases,
allowing reported issues to trigger updates in service records, not just status tickets. India’s Data Imperative:
The Pivot Towards Quality 11
5.3 Make it matter
Dashboards, recognition, and feedback loops to make data available in a highly accessible
manner.
A. Use dashboards to track quality, not just outputs
Service delivery dashboards often monitor coverage, but not data integrity.
• Add indicators like % of blank fields, error rates, and duplication flags.
• Show trends over time by district or block.
• Highlight gaps for targeted support.
B. Recognise and reward data quality improvements
Positive reinforcement builds a culture of care.
• Highlight top-performing blocks in monthly reviews.
• Publicly acknowledge field teams with high-quality submissions.
• Include data quality in annual appraisals or incentive frameworks.
C. Run a ‘Data Quality Week’ as an internal campaign
Short, focused campaigns help rally attention and clean up long-standing issues.
• Set goals (e.g., fill missing phone numbers, remove dead records).
• Track and share progress across teams.
• Debrief and document what worked for future replication.
These interventions do not require new laws, budgets, or platforms. They need ownership, clarity,
and consistent follow-through. They’re a way to start today, and build habits that last. When
repeated and scaled, small moves like these shift the culture around data. They reduce the cost of
correction. They make systems more responsive. And they signal to staff, leadership, and citizens
that quality is not optional. It is essential. Over time, these efforts lay the foundation for a shared
culture of quality that must be reinforced, coordinated, and made visible across institutions. The
next step is to move from individual habits to collective norms, where quality becomes a shared
expectation across systems, institutions, and levels of government.
But to turn quick wins into lasting norms, data quality needs deeper roots—anchored in roles,
incentives, and systems that don’t reset with each programme cycle.
6. BUILDING BLOCKS FOR CHANGE: INSTITUTIONS,
INCENTIVES & INTEROPERABILITY
Improving data quality isn’t just a matter of better tools or tighter rules; it’s about shifting how
institutions think, behave, and collaborate. The previous chapters laid out what quality looks like
and how to measure it. But making quality last requires something more enduring: ownership,
incentives, and infrastructure that support good data as a habit, not a one-off intervention. This
chapter offers three practical levers to institutionalise quality from within the system.
6.1 Institutionalising Ownership
Quality suffers when data has no clear owner. In many programmes, data moves across
departments and levels, but no single entity is responsible for maintaining its integrity from
end to end. The fix lies in assigning stewardship, not just for compliance, but for ongoing
care. This means designating custodians at each level (national, state, district) who are India’s Data Imperative:
The Pivot Towards Quality 12
empowered and accountable for data health. These roles don’t need to be new hires; they
can be embedded within existing structures with a defined mandate, clear escalation paths,
and regular review cycles.
In addition, quality must be seen as a shared responsibility. Programme heads, IT teams, and
field staff all have roles to play. When Ownership is structured and supported, errors don’t
just get corrected—they get prevented.
Under the U.S. Foundations for Evidence-Based Policymaking Act (2019), every federal
agency was required to designate a Chief Data Officer responsible for data governance and
quality. By October 2022, the Government Accountability Office reported that 82 agencies
had officially appointed CDOs, embedding stewardship at the heart of their data operations.
7
6.2 Getting Incentives Right: Rewarding Quality, Not Just Speed
Most public programmes reward what they can count: enrolments completed, forms
submitted, and dashboards updated. These indicators are useful, but they can create
unintended consequences. When speed is rewarded but accuracy is not, field staff may rush
through data entry. When outputs matter more than inputs, verification becomes a burden
rather than a standard. Data quality indicators—such as error rates, completion levels, and
timeliness—can be tracked and factored into programme reviews, not as punitive audits but
as a reflection of delivery strength.
Some states have implemented scorecards to rank departments based on data quality,
fostering a constructive sense of competition. Others have tied data improvements to budget
flexibility or expedited approvals. Although the approaches may differ, the fundamental
principle remains the same: when quality is measured, recognized, and incentivized, it
becomes significant.
6.3 Making Systems Work Together: The Case for Interoperability
Even clean data loses value if systems can’t use it together. Interoperability—across platforms,
departments, and time—is essential to making public data work harder. It depends on
shared schemas, reference codes, and APIs that allow secure, controlled access between
programmes.
India’s toolbox is steadily growing:
• IndEA provides a common vocabulary and reference architecture.
• NDGFP sets rules for sharing, consent, and anonymisation.
• Account Aggregator, DigiLocker, DEPA show how consent-driven flows can
reduce duplication.
• Community-led tools—from CDPI to iSPIRT—offer off-the-shelf templates, APIs,
and governance playbooks.
Still, most systems remain siloed, relying on PDFs or manual copy-paste. The next step is
scale: making these standards the default, not the exception. That means using common data
models where possible—for people, households, land parcels—and building new systems on
existing rails rather than starting from scratch.
Singapore’s Government Data Architecture shows what’s possible: structured datasets, clear
access rules, shared protocols. India already has the building blocks. The challenge now is
connection—between systems, teams, and incentives. That’s what makes quality stick. India’s Data Imperative:
The Pivot Towards Quality 13
7. THE ROAD AHEAD: TOWARDS A SHARED CULTURE OF DATA EXCELLENCE
India’s digital foundations are in place. The challenge now is ensuring that what flows through
them—our data—is accurate, complete, and trusted. Data quality is no longer a backend concern;
it is central to public trust, effective service delivery, and the success of India’s own AI ecosystem.
But this isn’t a problem technology alone can solve. True change requires a cultural shift—where
quality is everyone’s responsibility, not just the IT team’s. From frontline staff to policymakers, every
actor must see themselves as a data steward.
This shift must be supported by light-touch, collaborative mechanisms: voluntary scorecards, peer-
learning networks, flexible incentives, and playbooks of best practices that evolve with experience.
These resources—maintained by a neutral body like NITI Aayog or the proposed India Data
Management Office—should blend technical, institutional, and behavioural insights from across
government and sectors.
States can lead the way by embedding data quality cells, linking quality to service outcomes, and
recognising excellence. Capacity building, leadership development, and hands-on support will be
essential to sustain momentum.
Above all, this transformation needs visible commitment from the top. A government that champions
clean, usable data sends a powerful signal: that quality is not an afterthought—it is the foundation
of good governance.
India’s digital future will be shaped not just by how many platforms we build, but by how much trust
we build into them. And that begins with data that’s ready to serve.
REFERENCES:
1. https://www.financialexpress.com/business/banking-finance-upi-transactions-up-34-on-year-
in-april-down-a-tad-from-march-3829424/
2. https://economictimes.indiatimes.com/news/india/aadhaar-authentications-touch-2707-crore-
in-2024-25/articleshow/120703093.cms?
3. https://www.linkedin.com/pulse/24th-march-2025-over-369-crore-ayushman-cards-have-
m91vc/
4. https://www.business-standard.com/india-news/centre-to-save-rs-18-000-cr-by-removing-
fake-accounts-from-welfare-schemes-123101700528_1.html?
5. https://timesofindia.indiatimes.com/business/india-business/govt-sraps-rs-3-5-crore-bogus-
lpg-connections-saves-rs-21000-crore-subsidy-in-2-years/articleshow/53326003.cms?
6. https://cag.gov.in/uploads/download_audit_report/2023/Report-No.-11-of-2023_PA-on-
PMJAY_English-PDF-A-064d22bab2b83b5.38721048.pdf ?
7. https://www.gao.gov/assets/gao-23-105514.pdf India’s Data Imperative:
The Pivot Towards Quality 15