Sixteen ventures in consulting, education, technology, publishing, photography, research, civic life, and archival protest work — plus a growing Innovation Advisory canvas suite — connected by a single question: how do knowledge and story change what development practice actually does?
On the name
A log of ideas in the works.
It is a practice, not a slogan. Ideas get written down, tried, kept, or retired. The ones that survive grow into ventures. The ones that do not still earn their keep because they feed the ones that do. Everything on this site is either an idea being logged, a venture already in the works, or notes from the bench in between.
The ventures
Sixteen projects at different ages. Some run in the wild. Some are still finding their shape. Some are sketches the ink has not dried on. Together they ask how knowledge, story, and evidence can shift the practice of development, not the slide decks about it.
Consulting · MEL · Social research
A registered social research and advisory practice since 2016. MEL system design, programme evaluation, applied research, and strategic advisory for development funders, implementers, and civil society. Recent engagements include the Gates Foundation, the India Health and Climate Resilience Fund, the Alliance for Responsible Aquaculture, Claylab Education, GAIN, and Populi Research Services.
Archive · Protest · Civic tech
A crowdsourced visual archive of the posters, placards, and photography from India's ongoing student protests. Co-founded with Pallavi Nayak — cobbled together in two hours after the July 2026 Rapid Action Force crackdown at Jantar Mantar. Ships with automatic face-obscuring for protester privacy, EXIF/metadata verification, cross-referencing with public sources, and curated essays that contextualise each placard. Featured in The Hindu, August 2026.
Learning · Class 1 CBSE · NCERT-mapped
A learn-and-play platform for Class 1 CBSE. Customisable worksheet generator across English, Hindi, and Mathematics; NCERT-mapped content; flashcards; sticker rewards and progress tracking; reading-voice support in Indian and device pronunciations; and local backup so caregivers can carry the record between devices. Built for one specific first-grader at home; opened for anyone.
Civic tech · Air quality · AI · Indian cities
India's first citizen-led air quality testimony repository — now expanding. Hyperlocal AQI data, citizen-reported pollution accounts, and policy-relevant analysis across Indian cities, with a new AI assistant that turns sensor readings and testimony into questions a resident can act on.
Publishing · Commentary · Indian politics
A complete commentary practice — a sharper voice on Indian policy, rights, and the things mainstream coverage quietly skips. Fifty-plus pieces in the archive, syndicated across outlets, written to be used.
Civic tech · Policy tracker · Telegram bot
A live South Asia policy and scheme tracker. Government announcements, welfare schemes, and budget lines across the region in one searchable place — plus a functional Telegram bot that delivers daily updates straight to a practitioner’s phone.
Field kits · Frameworks · MEL practice
A complete suite of free, printable, one-page frameworks for development practitioners. Fifteen canvases live — Venture Readiness, The Measurement Checklist, the Evidence Canvas, the Scale Decision, the Kill Criteria, the Causal Pathway Review (CLEAR), the Indicator Test (VALID), the Knowledge Map (Rumsfeld), the Survey Bias Audit, the Feedback Loop, the Handover Canvas, the Baseline Design Test, the Aspiration Gap Framework, the Rapid Civic Tech Canvas, and the Curriculum–Construct Fit. Built for the moment before the proposal is written, the indicator is signed, or the evaluation lands. Free to use, printable as PDFs, used in classrooms and field reviews across South Asia.
Open source · Civic tech · Research tools
A family of open-source tools for public-interest research. InsightStack for MEL and econometrics, FieldStack for field R notebooks, EquityStack for development data workflows, SignalStack as the companion knowledge base for the Research Rundown newsletter (archived issues, featured tools, method spotlights, curated resources for practitioners), plus RootStack, BridgeStack, ViewStack, and PolicyStack under the hood.
Publishing · Women-led
A women-led publishing house committed to voices at the margins of Indian literary life. A complete catalogue across fiction, children’s books, and political non-fiction.
Education · 70 courses · Free for practitioners
A free development education platform for practitioners across South Asia. 70 courses (19 flagship + 51 foundational), 34 Studios for hands-on work, 135 policy and behavioural-economics games, 163 Reading Companions to development texts, 22 Deep Dives from named scholars, 490+ ImpactLex terms, a 324-source Dataverse, 8 Data Explorers (ASER, UDISE+, NAS, NFHS, AISHE, PLFS, Migration, Election Turnout), 31 Long View data visualisations, 16 Learning Rubrics, plain-language Law Guides for the Indian social sector, and a Pro Tools tier (Practitioner / Professional / Team). Community of 13,000+ practitioners.
Inside ImpactMojo
Photography · Daily mailer · Prints
A travel photography practice with a daily photo mailer at its heart — one photograph and its story, straight to subscribers’ inboxes every day. Plus a curated catalogue of museum-quality prints from across India and beyond. Affordable wall art priced so a first-time buyer can take the risk.
Research project · India · 2004–2026
An independent data-driven research project on India’s economic and democratic paradox from 2004 to 2026. Three novel indices — Statistical Suppression, Fiscal Centralisation, and Democratic Quality — connect macroeconomic trends to individual stories, with interactive visualisations and open datasets.
Newsletter · Development research · South Asia
A bi-monthly curation of development research, evidence, and sector jobs from South Asia and beyond. Lands in inboxes once every two months, for practitioners, students, and researchers who want the interesting stuff without the slog.
Stage zero · Tech experiments
The scratchpad, not a proper venture. Small tools, proofs of concept, and side projects built to scratch an itch or test an idea before it earns a name.
Indian craft · Thoughtful gifts
Curated Indian craft and thoughtful gifts. Researched guides to craft objects and gifts built to last, from makers worth knowing about.
Digital agency · Non-profits · Social enterprises
A digital agency built for non-profits, social enterprises, and the people changing things. The in-house workshop, open to outside projects too. We make movements pop.
Ideas, tested
57 experiments in public-interest work, shuffled on each rebuild. Each has a question, a bet, and an outcome. Showing 8 of 57.
Civic tech
The bet
An AQI number is a fact. A resident’s lived experience of that AQI is also a fact. Put them in the same repository and see what happens to the conversation.
What happened
JanVayu is India’s first citizen-led AQI testimony layer, aggregating hyperlocal readings and resident accounts across Indian cities.
JanVayu
Photography
The bet
Most travel photography lives on Instagram and dies there. Print it well, price it honestly, and offer the wall — not the feed — as the final medium.
What happened
Phototales Studio sells curated travel photography as museum-quality prints. The wall is the gallery, and the price is set so a first-time buyer can take the risk.
Phototales Studio
MEL frameworks
The bet
The expensive mistakes happen at the framing meeting, not at the analysis. Put the awkward questions on one page, free to share, printable as a PDF, and practitioners will use it because the cost of the conversation is now lower than the cost of skipping it.
What happened
The Innovation Advisory canvas suite — Venture Readiness, Measurement Checklist, Evidence, Scale Decision, Kill Criteria, Causal Pathway Review (CLEAR), and Indicator Test (VALID) — lives free on idealog.works/canvas. Used by funders, implementers, and reviewers in South Asia and beyond.
Innovation Advisory canvases
Independent research
The bet
Mainstream macroeconomic indicators hide the divergence between aggregate growth and individual lives. Build three honest indices — Statistical Suppression, Fiscal Centralisation, Democratic Quality — and use them to track 2004–2026 in a way the standard dashboards refuse to.
What happened
Some Perspective is a complete data-driven research project on India 2004–2026, with three novel indices, interactive visualisations, downloadable datasets, and a comparative international layer. Independent, open, and used in classrooms and policy circles.
Some Perspective
Commentary
The bet
The mainstream press skips the policy detail. Write the pieces they skip — rights, governance, electoral patterns, the things that get filed under “too complicated” — and keep them sharp enough to be quoted.
What happened
Policy Grounds Press has published fifty-plus pieces on Indian policy, rights, and governance. The archive rotates on the homepage and the writing is syndicated across outlets.
Policy Grounds Press
Data methods
The bet
District data is too coarse for local decisions but village data does not exist at scale. Small-area estimation can bridge the gap if the model is honest about what it is interpolating and what it is guessing.
What happened
The PEARL methodology produced block-level estimates from district-level surveys, giving programme managers sub-district granularity for the first time in states where no census update had arrived in over a decade.
PEARL methodology
Education measurement
The bet
A state government commissioned a randomised trial on a critical thinking curriculum. Before you can evaluate the curriculum you have to fix the construct, or the trial just measures its own instrument.
What happened
Research advisory on a government-school RCT, covering evaluation design, instrument review, and the measurement choices that decide what the trial can and cannot conclude. Ongoing.
Critical thinking RCT
Education
The bet
Development jargon locks people out of conversations. Build a progressive web app that works without internet, covers MEL, Theory of Change, gender, and research methods — with mathematical notation where it matters.
What happened
ImpactLex is a free, offline-capable PWA with rigorous definitions of development sector terminology. Used by practitioners without reliable internet as a reference they can carry in their pocket.
ImpactLex
Learning · Children
The bet
Most learning platforms are built for a market and then narrowed toward a child. Reverse the direction: build it for one Class 1 CBSE reader in this house, tune every design choice to her, and see whether the platform still holds up when a stranger opens it.
What happened
littlek.in ships with a customisable worksheet generator across English, Hindi, and Mathematics, NCERT-mapped content, flashcards, sticker rewards and progress tracking, and reading-voice support in Indian and device pronunciations. Now used beyond the household it was designed for.
littlek.in
Field data privacy
The bet
A field researcher’s sketch map carries the very thing that identifies a place — which community lives on which street, where the temple and the well sit, the shape of the caste hamlets. Useful for analysis, unpublishable as drawn. So build a pipeline that strips the identifying detail while keeping the community geography the analysis needs.
What happened
A reproducible notebook that flattens a scanned paper map to clean white, keeps the pen colours that mark community boundaries, removes road names and identifying landmarks with inpainting and donor-texture matching so no ghost trace remains, relabels villages by number in a hand-drawn style, and swaps or shifts landmarks under pre-agreed rules so two tiers of anonymisation produce the same result no matter who runs them. Every transformation is logged, so the de-identification is itself auditable.
Map anonymisation pipeline
Gender rights
The bet
Give girls a structured platform — a Meena Manch — inside the school, with adult facilitation and peer accountability. If the girls speak up as a group, the community hears them differently than it hears one girl alone.
What happened
Meena Manch platforms were established across multiple Indian states, reaching thousands of schools. The collectives delayed marriages, increased school attendance, and created a voice that individual girls could not sustain on their own.
Meena Manch
Children’s books
The bet
Children learn shape and symmetry through their hands, not their textbooks. Build the maths inside a story about a girl, her ants, and the rice-flour patterns on her doorstep, and the geometry follows.
What happened
Kalyani’s Ant Problem published as a fully author-illustrated picture book for ages 5–10 — geometry through kolam, story-first, classroom-friendly.
Kalyani’s Ant Problem
Education
The bet
Paywalls are the wrong bottleneck. Curation, teaching voice, and practitioner-ready tools matter more.
What happened
70 courses (19 flagship + 51 foundational), 34 Studios, 135 policy/behavioural-economics games, 163 Reading Companions, 22 Deep Dives, 490+ ImpactLex terms, a 324-source Dataverse, 8 Data Explorers (ASER, UDISE+, NAS, NFHS, AISHE, PLFS, Migration, Election Turnout), 31 Long View visualisations, 16 Learning Rubrics, plain-language Law Guides, and a Pro Tools tier. Community of 13,000+ practitioners across banks, UN agencies, governments, and universities.
ImpactMojo
Behaviour change
The bet
Make the ad bold enough that people talk about it at home. The conversation the ad starts matters more than the ad itself.
What happened
Public health campaigns on family planning, breastfeeding, and complementary feeding reached tens of millions of households. The campaigns that landed were the ones audacious enough to name the behaviour plainly.
Public health mass media
Political accountability
The bet
Observers deployed to evaluate local units file reports, candidate names, and supporting documents through a portal. Turn that exhaust into a weighted, auditable score and the party can see who actually did the work.
What happened
A seven-parameter grading framework that scores observers A to D on report quality, daily activity, documentation, proposed names, and timeliness. Runs across more than 700 districts in 23 states and recomputes from the same database on a monthly cycle.
Observer grading framework
Indian craft
The bet
The craft economy is full of beautiful work that never makes it past the local market. Curate the good stuff, research the makers, and build a guide that respects both the buyer and the artisan.
What happened
All The Good Stuff curates Indian craft and thoughtful gifts with researched guides to objects and makers. The curation is the product.
All The Good Stuff
Discrimination and development
The bet
Discrimination shows up as gaps in income, mobility, and opportunity, but raw gaps confound caste with everything correlated to it. Build the study and the design so the caste effect can be separated from the rest.
What happened
A field study on caste, social capital, and economic mobility in rural South India, covering survey design, household sampling, and the discrimination measurement most studies leave as a residual.
Discrimination and Development study
Education measurement
The bet
Arts programmes get evaluated on attendance because nobody knows how to measure joy. Try anyway — use observation tools, child self-report, and teacher perceptions to triangulate something real about what the art is doing to the child.
What happened
A mixed-methods design captured socio-emotional outcomes including joy in a low-resource Mumbai school arts programme. The methodology showed that arts-based, non-linear outcomes can be evaluated honestly.
Joy in arts education
MEL design
The bet
Evaluation tells you what happened after the money is spent. A learning layer tells you what to change while the fund is still running. Build it in from the start.
What happened
The learning layer produced real-time evidence that was used to adjust fund priorities mid-cycle. Programme managers had data they could act on, not reports they could file.
Fund learning architecture
Education
The bet
Children learn systems through narrative. Tell the story of a coin — where it comes from, where it goes, who touches it, what it buys — and the economics follow the plot.
What happened
Financial literacy modules for school children, built around narratives like the journey of a coin, reached classrooms years before fintech made financial literacy a startup pitch.
Financial literacy in schools
Electoral integrity
The bet
Rolls get revised and names drop. Pull them across revisions and the deletions are not evenly spread. Map who is removed and where.
What happened
Constituency-level analysis of elector deletions across four states, feeding commentary on roll revision. One piece ran in a national daily.
Elector deletion analysis
Measurement
The bet
International school readiness scales miss what matters in Indian contexts. Build the instrument locally — local languages, local developmental markers, local validation — and prove it holds up statistically.
What happened
The School Readiness Instrument (SRI) was developed and validated for Indian ECCE contexts, measuring cognitive, socio-emotional, and physical development. One of the earliest locally built readiness tools in the country.
School Readiness Instrument (SRI)
Mobile health
The bet
An automated voice call reaches women who cannot read, in their own language, at the right gestational week. Design the calls to be welcomed rather than tolerated.
What happened
Kilkari became one of the largest mobile maternal health services in the world, delivering weekly audio messages timed to each stage of pregnancy and early infancy across multiple Indian states.
Kilkari
Publishing
The bet
Flip the contract. Publisher signs an NDA to the author. Build a durability promise into the physical book. Let authors illustrate their own work.
What happened
StoryWell Books runs on this model. Multiple titles in the catalogue, and the author-led illustration loop is the house style.
StoryWell Books
Civic data
The bet
Most Indians overestimate their economic position. Build a calculator that uses real government data and lets a user type in their income to see where they actually fall. The shock is the pedagogy.
What happened
The Real Middle reveals where users stand in India’s income distribution using government data. Makes inequality concrete and personal in under ten seconds.
The Real Middle
Animal welfare
The bet
Fish welfare in aquaculture gets asserted more than it gets measured against a counterfactual. Set up a controlled pilot. Treatment ponds get water-quality monitoring and stocking-density guidance, control ponds carry on as usual, and dissolved oxygen, ammonia, and the rest get read off both.
What happened
A randomised pilot across 38 carp ponds in coastal Andhra Pradesh, 76 measurements, built to validate the protocol before a full evaluation. It found control ponds sitting outside safe dissolved-oxygen ranges much of the time, and confirmed the parameters could separate treated ponds from untreated ones, which is what makes the larger trial worth running.
Aquaculture welfare evaluation
Representation audit
The bet
An observer’s caste and the caste of the candidates they rank first are both in the data. Compare each observer’s top picks against the caste mix of their own candidate pool and the district’s demographics, and disproportion becomes a number rather than an accusation.
What happened
A pool-adjusted bias index surfaced specific patterns — districts where both top picks came from one dominant caste, majority-Adivasi districts with no Scheduled Tribe candidate in first preference, and conflict-of-interest cases where a proposer’s phone number matched a candidate’s. Delivered as a standalone diagnostic.
Caste representation diagnostic
Welfare at slaughter
The bet
Most farmed fish die without the method ever being examined for welfare. You cannot improve what no one has watched, so the first step is structured field observation of how killing actually happens.
What happened
Field observation of ice-slurry killing, recording the practice as it is rather than as protocols describe it, to establish whether a welfare problem exists and how large it is before anyone proposes a fix.
Slaughter-welfare observation
Early childhood
The bet
Games, not apps. Ambient Indian contexts, not imported stock art. Meet the child where the phone actually lives.
What happened
India’s first Android educational games suite for early childhood, designed and evaluated in the field. Built when digital was not the default and educational apps were not yet a category.
Early childhood digital learning
Workshop tools
The bet
Most ToC and logframe work happens in PowerPoint, which fights you on every drag. Build the tools in the browser — free, no login, opinionated defaults — and the workshop runs faster and the artefact is shareable.
What happened
Twelve interactive tools shipped under ImpactMojo — Theory of Change Builder, LogFrame Builder, Stakeholder Mapper, Empathy Map Canvas, Policy Canvas, Chart Selector, AI Canvas, and a Shiny apps suite for development economics. Used in classrooms, workshops, and live programme design.
ImpactMojo workshop tools
Women & aspirations
The bet
Development surveys measure attainment (what women have) and treat aspiration as noise. Measure aspiration and attainment separately, in matched pairs, across ten life domains and five life stages, and the gap between the two becomes the finding.
What happened
Badalte Khwaab · a telephonic survey of 6,092 rural women across 15 states, led with Development Intelligence Unit and Transform Rural India Foundation. Central finding: the biggest gap between what women want and what they have is among the youngest cohort (23%), and 38% of unmarried girls aged 15–19 place themselves at the very bottom of a life ladder while 71% still expect the next five years to improve. The report reads life stage, caste, and geography together rather than reporting a national average.
Badalte Khwaab · Women’s Aspirations in a Changing India
Behaviour change
The bet
Work at the intersection of supply (health system readiness) and demand (community practices). Shape both at the same time, district by district, using frontline workers as the hinge.
What happened
The Ananya programme reached millions across Bihar, combining supply-side health strengthening with demand-side behaviour change through frontline workers. The ambition was the point — a state, not a pilot.
Ananya, Bihar
Open source
The bet
Yes, if the workflows ship as reusable notebooks with opinionated defaults and the onboarding assumes nothing.
What happened
EquityStack (Python) and FieldStack (R) run without a programming background. EquityStack is archived on Zenodo with a DOI. Eight OpenStacks components live on GitHub.
OpenStacks for Change
Civic tech · Bots
The bet
The PolicyDhara web tracker is useful but lives on a tab you have to open. Move the headlines to a daily push on the platform practitioners already keep open. If the format is short enough, the open rate beats any newsletter.
What happened
PolicyDhara now runs a functional Telegram bot that posts a curated daily digest of South Asian policy announcements, welfare schemes, and budget lines. Read in the time between meetings.
PolicyDhara Telegram bot
Archive · Protest
The bet
After a state crackdown, the visual record of a peaceful protest starts disappearing within days — placards get discarded, photographs get pulled, memories get overwritten. Set up a lightweight crowdsourced repository fast, put the safety guardrails in first, and the archive can hold what would otherwise be lost.
What happened
Democracy by Design, co-founded with Pallavi Nayak. Ships with automatic face-obscuring, EXIF/metadata verification, cross-reference against public sources, and curated essays contextualising each placard. Built in about two hours after the July 2026 Rapid Action Force crackdown at Jantar Mantar. Featured in The Hindu, August 2026.
Democracy by Design
Knowledge work
The bet
Build the editorial process around lived accounts and keep the chapters small. Real-time reporting is possible if the structure is designed for speed.
What happened
A Year of Havoc, co-edited and co-authored in 2022, collecting human experiences of the second wave of the COVID pandemic in India.
A Year of Havoc
Publishing
The bet
The policy brief is the wrong genre for what needs to be felt. Tell the story from inside the world instead.
What happened
The Very Last Superhero, a dystopian novel following a fourteen-year-old in a climate-collapsed India. Published, bestselling on Amazon.in.
StoryWell Books
Welfare standards
The bet
Welfare rules usually follow public concern, and farmed fish sit near the bottom of that ladder. Skip ahead through the technical route instead. Water-quality and stocking-density norms, and engagement with the bodies that set animal-welfare and quality standards.
What happened
Policy and standards work to move responsible aquaculture toward formal norms, running alongside the field programme with carp farmers, so the practice has a rulebook to point at rather than only a set of recommendations.
Aquaculture standards work
AI in research
The bet
Generic transcription engines garble the names, places, schemes, and code-switched English-Hindi-vernacular sentences that fill a development research interview. Build one trained on the research vocabulary and the pipeline becomes a useful tool, not a chore.
What happened
VaniScribe runs as an AI-assisted transcription and analysis tool built for development research, sitting inside the ImpactMojo workflow. Researchers spend less time fixing mistranscribed names and more time on the analysis.
VaniScribe
Entertainment-education
The bet
Embed the message inside a story people actually want to follow. If the characters change, the listeners might too.
What happened
Adhaa Ful aired as an entertainment-education drama reaching rural audiences. The format — story first, message second — outperformed direct messaging on listener recall and reported behaviour.
Adhaa Ful
Climate resilience
The bet
Build the programming around what fishing communities already know about weather, and fill the specific gaps — where to go, what to carry, how to read the warnings — through radio they already listen to.
What happened
Lifeline programming for cyclone-prone Odisha launched years before climate adaptation entered mainstream development vocabulary. The approach — local knowledge first, institutional messaging second — became a template for later resilience work.
Lifeline programming, Odisha
Welfare delivery
The bet
Direct cash transfers to women get counted as received or not. The interesting story is in between. Who controls the money once it lands, who gets judged for taking it, who absorbs the unpaid work of enrolling everyone else.
What happened
A multi-state qualitative study of stakeholder perspectives on women’s cash-transfer schemes, coding intra-household control, documentation exclusion, payment delays, social stigma, and the administrative burden pushed onto frontline workers.
Cash transfer perspectives study
Measurement
The bet
Resilience is culturally situated. A scale built in Canada will miss what matters to a child in an Indian slum. Adapt the Child and Youth Resilience Measure to read local realities — community, family, and cultural resources.
What happened
The CYRM was adapted for Indian contexts, giving practitioners a validated resilience measure that accounts for local protective factors rather than importing Western assumptions about what bouncing back looks like.
CYRM adaptation
Civic tech
The bet
Government announcements, welfare schemes, and budget lines are scattered across dozens of ministry websites, PDFs, and press releases. Put them in one searchable place and practitioners will come.
What happened
PolicyDhara is a live South Asia policy and scheme tracker aggregating government schemes, budgets, and announcements across the region in one searchable interface.
PolicyDhara
Survey methods
The bet
A large household baseline invites fishing. Run that many tests and some look significant by chance. Correct for it up front and report what survives.
What happened
Baseline design and analysis for a household survey of 2,800 households across four districts in Jharkhand, 119 bivariate analyses with Bonferroni correction.
Household baseline analysis
AI in evaluation
The bet
Human-administered follow-up surveys are expensive and slow. An AI calling system can conduct structured interviews with fellowship alumni, capturing employment, civic engagement, and social capital data at a fraction of the cost and time.
What happened
An AI-assisted call survey was designed for tracking long-term alumni trajectories, combining automated structured interviews with targeted human follow-ups for deeper probing.
AI-assisted alumni tracking
Civic data
The bet
Build state-level maps across demography, health, gender, economy, education, and environment, and let the visual divergence speak louder than the national average ever could.
What happened
How India Lives is a pedagogical tool showing 205 state-level maps of India’s diversity. Used in classrooms and policy discussions to explain why India’s states diverge and why that divergence matters for programme design.
How India Lives
Media and attitudes
The bet
Twenty years ago, cable television arriving in a village tracked lower son preference and less acceptance of wife-beating. Cable was new then, so watching marked you out. Interview three generations of women in one household, separately, and you can see whether television still carries that signal or whether the phone has taken it over.
What happened
A triadic pilot in 115 households, interviewing a grandmother, mother, and unmarried daughter in each. Daily social media ran at 80 percent among daughters and 2 percent among grandmothers. Daughters who used social media daily reported markedly lower son preference, and each extra platform a mother used tracked about eight points more say in household decisions. Television no longer pointed one way. Nothing moved the acceptance of wife-beating, in any generation. The medium that now plausibly moves these attitudes is the smartphone.
Three-generation media and gender study
Education
The bet
Give families a voucher that covers fees at a private school of their choice. If the barrier is cost and the private schools are already trusted, the voucher closes the gap without building new classrooms.
What happened
India's first school vouchers field trial for girls, conducted in Old Delhi. An early RCT in a country where randomised evaluation of education policy was still rare.
School vouchers, Old Delhi
Migration economics
The bet
Rural households in seasonal poverty stay home because the bus fare to the city is a barrier they cannot finance. Cover the fare, and the migration — and the remittances — follow.
What happened
No Lean Season was designed with Yale RISE economists and tested as an RCT across three countries. The programme showed that a modest travel subsidy unlocked seasonal migration and measurably increased household consumption during the hungry months.
No Lean Season
Education measurement
The bet
Build a common evaluation framework — a Tulna (comparison) — that standardises what counts as an outcome across edtech products, so funders and schools can compare apples to apples.
What happened
The EdTech Tulna framework became a reference tool for assessing edtech effectiveness, giving the Indian education sector a shared vocabulary for what “works” means in digital learning.
EdTech Tulna
AI · Air quality
The bet
A raw AQI number is a fact, but a fact a parent or a school principal cannot act on. Build an AI layer on top of JanVayu that turns a reading plus testimony into “should I keep the windows shut today?” — in plain language, grounded in the actual data behind the number.
What happened
JanVayu now ships with an AI assistant that takes hyperlocal AQI and resident accounts and answers practitioner-level questions about action, exposure, and timing. Built into the expansion, not bolted on.
JanVayu AI assistant
Programme design
The bet
Cash transfers alone change behaviour temporarily. Design the incentive structure around the family system — conditional on school attendance, linked to the girl child staying enrolled, embedded in the existing welfare architecture so it survives a change of government.
What happened
Incentive design for a national child protection scheme covering child labour and child marriage, working within the POCSO and JJ Act mandate to align programme incentives with existing legal frameworks.
Child protection incentive design
Knowledge curation
The bet
Practitioners want the interesting stuff from the research world without reading forty institutional websites. Curate it honestly, keep it short, send it six times a year.
What happened
The Research Rundown reaches development practitioners, students, and researchers across South Asia with curated research, evidence, and sector jobs in a format designed to be read, not filed.
The Research Rundown
Mobile health
The bet
Give the health worker a tool she trusts more than her memory. Print the counselling points on cards, link each card to a short audio clip via mobile, and let the worker play the clip during the home visit.
What happened
Mobile Kunji reached millions of frontline health workers across Bihar. The combination of physical cards and mobile audio gave workers a script they could lean on without reading from a manual.
Mobile Kunji
Digital agency
The bet
Movements need design as much as corporations do. Build the agency for the people changing things, price it for their budgets, and the word of mouth does the rest.
What happened
MakeItPOP Online runs as a boutique digital agency for nonprofits and social enterprises — campaigns, websites, and digital strategy for organisations that would otherwise go without professional design.
MakeItPOP Online
Health systems
The bet
The district is where the state meets the ground. Small enough to manage, large enough to matter, and it is where the District Collector sits. Design the fund to move at this level.
What happened
The fund adopted the district as its primary unit of intervention, with health system strengthening and climate resilience programming designed around the administrative reality of Indian governance rather than donor convenience.
District-level health systems
Values
Six working rules for how the work gets made. They get forgotten under deadline pressure, which is why they are written down here.
We write reports that practitioners actually open on a Monday morning. The number has to answer a question somebody is already asking.
MEL work that stays on a shelf cost somebody money and helped nobody.
The voices at the edge of the data frame better questions than the terms of reference. We listen there first.
If your sampling frame is polite, you are sampling the wrong people.
If a programme officer cannot follow the argument at 3pm on a Tuesday, the argument needs rewriting. Jargon is where accountability goes to hide.
We have lost good ideas to bad sentences.
We publish openly when we can. The free edition reaches the practitioner who cannot afford to pay, and that practitioner is the one the work is for.
Access changes who gets to ask the question.
Every script, template, and workflow belongs to whoever needs it next. The sector runs on shared infrastructure and we contribute to it.
A hoarded method helps one team. A shared one helps a thousand.
Peaceful in most rooms, loud in the ones that need it. Evidence that sits on the fence is usually already serving whoever put the fence there.
Silence during harm is a position, and we know which one.
The playbook
There is no magic. There is a loop that runs many times before anything gets a name. It looks like this.
Listen for a question that real practitioners are bumping into and nobody has framed cleanly yet. Read the complaint inside the complaint.
weeks, not quarters
Prototype on the cheap. A paper tool, a field survey, a notebook, a script. Show it to three people who will use it and fix what they hate.
low cost, fast feedback
Publish a public version that someone can open on a Tuesday afternoon and get value from before closing the tab. Ship, then tune.
in the wild, not in a deck
The venture grows its own team, its own story, and its own runway. Some land with a partner organisation. Some become their own thing. A few get retired with thanks.
own legs, own shape
Work & writing
Essays, dispatches, and the slower work of making sense of what the data actually says. Two places the writing lives.
Sharper commentary on Indian policy, rights, and the things mainstream coverage quietly skips. Pulled live from the homepage so the latest pieces always surface.
A bi-monthly curation of development research, evidence, and sector jobs from South Asia and beyond. Once every two months.
Longer, slower pieces on research, measurement, parenting, and the view from the edges of the development sector. Self-hosted here on the site.
Long-form pieces published natively on X — quick takes that grew past the length of a post. Hand-picked, newest first.
Free Field Kit
Twelve questions to ask before you commission, design, or fund a piece of MEL work. One page, free to share, printable as a PDF. A companion to The Measurement Trap.
Notes
40 short working notes on measurement, field practice, and what the data actually says. Shuffled on each rebuild — showing 8 of 40.
Read a child the same book twice in one sitting. The second reading is the one that actually teaches.
India has not conducted a census since 2011. Every policy decision made with that data is a bet on a country that no longer exists.
If a training does not change what a participant does on the following Monday, it was a performance, not a training.
A hand-drawn village map carries the village. Strip the labels and leave the geometry and you have published the place under a fake name. Decide what to remove first, what to swap second, and what never to publish.
Most ToC reviews are friendly. The friendly ones are the ones where the reviewer was hoping to be invited back. The useful ones are the ones where the reviewer says the awkward thing in the first half-hour.
A public archive that claims certainty is a public archive that can be used against its own subjects. An archive that says openly what it can and cannot verify — and puts the safety of contributors ahead of the completeness of the record — behaves like a witness rather than a court.
Resilience looks different in a Mumbai slum than it does in a Toronto suburb. A scale that cannot see the difference is measuring its own assumptions.
Every brief says 'strategy'. Most briefs mean 'somebody to blame later'. Read the second one, and quote for the first.
Free platforms beat paid ones on reach, paid ones beat free on focus. The useful question is which one a practitioner will actually open on a Tuesday afternoon.
The cardboard goes into the rain. The spray-paint gets walked over by Monday. What survives is what somebody decided to keep. Archives are not neutral about the future.
A learning platform built for one specific first-grader is a platform whose design decisions are all defensible. Every widget was there because a real reader needed it. When strangers start using it, the platform is more honest than most that were built for a market.
A pre-mortem is a post-mortem for a project that has not failed yet. Run it before the launch. The cause of death usually has a date on it already.
GDP counts the consultant who advises on the kitchen renovation. It does not count the woman who has cooked in that kitchen for forty years. The number is not measuring the economy. It is measuring what the economy has agreed to notice.
A grant is a bet on a person before it is a bet on a project. Fund the person with the question, not the prettiest logframe.
Every metric is gameable by the people whose work it measures. The useful question is not whether they will, but whether the gaming produces the underlying change or only the appearance of it.
Average improvement is the place inequities go to hide. If your indicator cannot split by gender, caste, and geography, you are reporting on a country that does not exist.
Some tools have to ship at the speed of the news cycle that produced their need. A civic archive built in November for a protest that ended in July is a memorial for a placard that nobody has any more.
A dashboard is a confession of what the team has decided to notice. The honest version lists what it has decided not to look at.
A small gap between what a woman wants and what she has can mean either that the aspiration has been reached, or that the aspiration has quietly narrowed. Both look the same in one number, which is why the two have to be measured separately.
A cash transfer changes behaviour for as long as the cash lasts. An incentive embedded in the welfare architecture changes behaviour for as long as the government stands.
The communities most at risk from climate change are the ones with the least data about it. The adaptation gap starts at the weather station.
AI will not save your evaluation. It will speed up the transcription of an interview you should not have run, and produce a summary of evidence you did not want to read.
A free printable framework is a quiet act of redistribution. The audit a practitioner cannot afford to commission becomes the audit she can run on a Friday afternoon.
The farmer with the most grievances is usually the one with the fewest ways to be heard. If your sampling frame is polite, you have the wrong frame.
A civic tool is only as good as its refusal. The question is not what you can show, but what you will not build even when you could.
An indicator is a promise about what you will pay attention to. Pick only the ones you can afford to keep.
A Theory of Change is a hypothesis, not a contract. Treat it like a hypothesis and you might learn something. Treat it like a contract and you will only ever defend it.
A national paper leak is not the failure. The failure is the design that put twenty-two lakh young people on one paper, on one morning, with one set of questions.
A bus ticket is a financial instrument. The family that cannot afford the fare to the city during the lean season stays poor until someone covers it.
An RCT is a photograph. A good photograph tells you something, but you can't drive on it, cook with it, or send a child to school on it.
Most household surveys treat caste as a control variable. If the question is about caste, treating it as a control is a way of refusing to look at the answer.
A country that does not measure its own inequality is not failing to measure. It is choosing not to know.
A roll revision is a quiet political act. The names that drop between two revisions are the ones nobody is counting; the absence is the finding.
Power leaves traces on the calendar before it leaves traces on the budget. Read the calendar first.
If a report needs a glossary, the glossary is where the reader gives up. Write the glossary first, then don't use any of the words in it.
The voices at the margin are not there because they are quieter. They are there because the rooms are built small. Make the room bigger.
A daily thing forces the discipline a weekly thing pretends to have. The story you write at 6am is the one you actually had, not the one you would write if you had time to revise it into something safer.
A number that's easy to count is not the same as a thing that's worth knowing. Half of good MEL is refusing the first question and asking the better one.
38 per cent of unmarried girls aged 15–19 place themselves at the very bottom of the life ladder. 71 per cent still expect the next five years to improve. The window in which that hope survives is finite. Programme design that arrives after it closes is arriving for the wrong cohort.
If you cannot measure joy in a classroom, the problem is with your instrument, not with the joy. Try harder.
On Rotation
A few things on the desk this month. A bit of reading, something to watch, something to listen to. Updated when something better shows up, and not before.
Reading
Books from the desk and the shelf behind it. The full list lives at /shelf.
Watching
Two screening clubs, an RSVP aggregator, and three films worth the whole evening.
Listening
Hindi golden age, Bengali soul, Tamil melodies, folk protest, ambient for the writing desk, and a lot of country.
On the podcast shelf
Sixteen shows in heavy rotation — development, law, climate, and a few dailies.
About
Varna Sri Raman has spent twenty years at the edges where research, story, and practice meet. She reads evidence for a living. She reads most other things for the love of it.
The questions keep coming back to the same few. What are we measuring, and does it match what we mean? Whose voice gets to frame the problem, and who has to live with the answer? Why do so many good reports end up on a shelf? Policy briefs rarely fit these questions, so she writes in whichever form will hold them — essays, working papers, a dystopian novel, children's picture books, and a handful of longer books still taking shape.
The writing is half the work.
Off the clock, she is wife, daughter, and sister inside a loud, loving family, and mum to two sons. Two cats used to live here, until they ran away for love — they spent six months quietly duping her by bringing lookalikes home at dinner time for the extra meal, an operation she respects. Heera the dog still lives here, and is much loved. Free time goes to books, watercolour painting, astrophysics rabbit holes, and a lot of music — mandolin, sitar, and piano on the speakers, guitar in her hands (barely). When not otherwise occupied with her projects, Varna can be found curled up by the window howling at the moon. A peaceful bibliophile until confronted by a fascist.
In print
Collaborate
Conversations about MEL consulting, research partnerships, civic technology, and development education are open. The work tends to start with a question no one has framed well yet and ends with something that can actually be used in the field.
If that sounds like what you are trying to do, write in. A short note about what you are working on is enough to begin.