TL;DR

• GiveWell has published a podcast, looking back on their investments in iron fortification and their $8.2M grant to Fortify Health in 2021. When the grant was made, the projected cost-effectiveness was 5x cash transfers we delivered and estimated 12x. We want to share what this looked like from the inside.
• Over three years, our team scaled from reaching fewer than half a million people to more than 20 million, while roadblocks in our government engagement strategy forced a pivot.
• This post is about what we learned over the last 3+ years of scaling a public health intervention, the mistakes we made and the learnings we hope will be useful for funders, implementers and supporters.
• We are seeking approximately $1 million over the next two years to unlock the government channel, where we believe the next order of magnitude of impact lies. We would welcome your support.

1.    Background

GiveWell has just published a podcast discussing their thinking about iron fortification and our work at Fortify Health. In it they discuss their lookback on their 2021, $8.2M grant to Fortify Health along with their decision to continue their support of our work with an additional $10M grant. We appreciate the epistemic honesty and rigor that GiveWell has brought to this process and their desire to better understand impact and evidence generation. 

The headline result is encouraging, shifting an estimated cost-effectiveness of 5x cost of cash to 12x over the July 2022 - June 2025 period. Internally, we are more proud of what sits behind that number:

• Growing from fewer than half a million beneficiaries reached per month to more than 20 million last month;
• Building a team of almost 100 remote-working team members across 41 locations in India; and
• Driving down the cost per beneficiary from $0.86 in 2022 to approximately $0.20 currently. 

But the headline does not tell you about the failed Maharashtra government partnership we had been working on since 2019. It does not tell you about the two years we missed production targets, the various team management challenges we got wrong, or the strategies we had to abandon and build later in 2025. That is what this post is for.

I am sharing this because I appreciate the transparency of the Effective Altruism community and our combined desire to maximise cost-effective impact. My hope is that sharing what we got wrong will help others skip these pitfalls. 

Additionally, we want to continue scaling our work and are seeking support in our journey to reduce iron-deficiency anaemia at a massive scale.

2.   About Fortify Health

2.1 The problem

Anaemia is one of India's most persistent and damaging public health challenges. More than two in three children under five are anaemic. More than half of all pregnant and lactating women are anaemic, and for pregnant women, severe anaemia doubles the risk of dying during or after childbirth. For children, the consequences compound across a lifetime: impaired cognitive development, reduced school performance, lower long-term productivity.

Iron deficiency is the leading cause of anaemia. Iron deficiency anaemia occurs when depleted iron stores impair the body's ability to produce haemoglobin and carry oxygen effectively.

Despite meaningful progress on other nutrition indicators, anaemia prevalence among women and children actually worsened between India's 2015-16 and 2019-21 national surveys. This is not a problem resolving on its own.

2.2 Our solution

Fortify Health is an Indian non-profit organisation and we make it as simple as possible for millers and government programmes to transition from the distribution of unfortified wheat flour to fortified wheat flour. In practice, this means adding iron, folic acid and Vitamin B12 according to Government of India regulations to wheat flour that is already being distributed (but otherwise unfortified). We do this through our hassle-free package:

• Right Premix: the right blend of micronutrients in the right form and dose for Indian wheat flour, manufactured to a consistent specification.
• Right Equipment: installation and calibration of a microdoser that doses the premix automatically within the mill, with no manual measurements and consistent, appropriate dosage.
• Right Testing: an AI-enabled quality assurance tool that cuts testing turnaround from over ten days to fifteen seconds. It determines whether the fortified iron in a sample is within the regulatory range, and if not, runs an automated troubleshooting sequence to help millers fix any issues.
• Right Training: tools to support both mills and governments to implement wheat flour fortification at scale.

2.3 What we are building toward

In India, more than 400 million people consume some form of wheat daily through a government welfare programme, be it the Public Distribution System, the midday meal scheme known as PM POSHAN, or the Integrated Child Development Scheme, which reaches pregnant and lactating women and children under the age of three. Our goal is to support governments to supply fortified wheat flour into these systems so that governments finance and sustain the impact at national scale.

By the end of 2028, we hope to be working with more than 33% of all industrially packaged wheat flour in India through our mill partnerships, while also meaningfully engaging the government as they seek to reduce iron-deficiency anaemia. Specifically, we are working toward reaching more than 10 million beneficiaries through government and social safety net partnerships, alongside more than 60 million beneficiaries through our existing mill partnerships. If one or more state governments fully mandate fortified wheat flour, the potential reach could be significantly larger, though the variance here is high given the political forces involved.

3.   Private Miller Partnerships – from .5 to 20 Million people reached

The key financial headline across the grant period: we spent 69% of our budget while reaching more than 28% more people than projected. That combination drove an estimated 780,000+ person-years of anaemia averted over the three-year grant period, at roughly $5 per year of anaemia averted.

3.1 What drove the improvement: within our control

The most significant structural change was operationally shifting from a state-based model to a cluster-based model. Rather than pursuing broad geographic coverage, we focused on deeper support within key hubs across 27 ‘clusters’ across India. This allowed us to replicate our model effectively and cost-efficiently.

We also built significant internal capacity over the grant period. Our team leveraged its presence and partnerships to drive down time to launch from 11 months down to approximately 3 months. That same market position allowed us to drive down key input costs, including premix and equipment. We have been deliberate about building a high-quality leadership team that is not top-heavy, allowing us to maintain strong team members across the organisation while keeping executive overhead lean. We are now an organisation of almost 100 people with overhead as a share of total expenditure falling from 30% to 19%.

But perhaps most importantly, we hired and retained some of the best development practitioners in India. Most of our leadership team has been with the organisation for more than two years and we have maintained an attrition rate of below 5% every year since 2022. We have invested heavily in people (perhaps cost-ineffectively in the short-term) but these investments have paid significant dividends. 

3.2 What drove the improvement: outside our control

Two changes in GiveWell's model also contributed to the improvement in estimated cost-effectiveness, neither of which was within our control.

First, GiveWell's updated benefit model now accounts for neonatal mortality reduction and physical work capacity effects that were not included in the 2021 projection.

Second, IDinsight conducted an independent primary data collection exercise on our work, which led to a revision in the estimated per capita consumption of wheat flour from 191g per day to 132g per day. A summary of the study as well as our contextualised interpretation can be found here. While lower than previously assumed, 132g per day figure still exceeds the 75g per day threshold that WHO deems sufficient to drive anaemia reduction at the population level. The practical effect of this revision was that each tonne of fortified flour reached more individual people, partially offsetting the lower per-person iron intake and contributing positively to the overall beneficiary count.

4. Our government partnership story

As noted in the GiveWell podcast discussion, our engagement with government has produced mixed results. From 2022 until mid-2025, our team pursued a top-down approach, seeking to drive changes in state and central government mandates by engaging with the most senior bureaucrats we could speak with. The results fell short of our hopes and expectations. Key relationships stalled, officials were transferred at critical moments, and we fell into a structural catch-22 of working with the government: they want local evidence of the impact before embarking on a pilot, but we can’t embark on a pilot without local evidence. 

In late 2025, we fundamentally changed our strategy to a bottom-up approach: generating local evidence through smaller, faster programmes with local bodies, civil society organisations, and quasi-government entities. The goal is to build the case for fortification from the ground up, and ultimately to support the kind of mandate that could sustain impact at the national scale. The results so far have been more promising.

The following is a summary of our results since shifting to this new strategy:

We are seeking to expand this rapidly over the next two years, with a target of more than 10 million beneficiaries through more than 60 partnerships by December 2028. We want to be honest about the uncertainty: some of India's larger states could, if they mandated fortified wheat flour through their welfare programmes, far exceed 10 million beneficiaries on their own. But the path to that is highly uncertain, often political, and depends on having the right champion, the right evidence, and the right timing converging in the same place.

5. What we learned

Theme 1: Build the foundations before the pressure arrives

What Has Worked

One thing I am most proud of over this period is that we grew the team from approximately 25 people to almost 100 without a meaningful drop in the quality of our hiring or the culture of our team. This happened through a period of vast scaling and growth, with a fully remote team now working across more than 40 locations across India. 

What we built well and early was our people infrastructure: hiring systems, feedback loops, and HR data management. We introduced a structured OKR cycle on a six-month cadence that was long enough for meaningful measurement but short enough to pivot when it was clear the goals were inappropriate. We have iterated constantly on how we use goal setting, shifting toward more measurable and controllable targets, adjusting the number of goals over time, and more recently linking OKRs to individual performance evaluations. We also used in-person retreats deliberately and they have become absolute highlights of the year. None of this is particularly innovative, but the combination of relationship building, continuity and iteration has paid compounding returns in ways that are easy to underestimate. These are all costs, but I suspect that in the desire for cost-effectiveness many organisations in the development sector tend to underinvest in operations and people. 

What Has Not Work

Where we were less deliberate, and are now catching up, is financial and procurement infrastructure. We invested in people systems early but not in finance systems at the same pace. As the organisation has grown, these gaps have become increasingly costly. We are addressing them now, but doing so under the pressure of fast-growing programmes is materially harder than doing so in advance. It is worth noting that with the advent of AI, the financial cost and technical knowledge required to build these systems has reduced significantly, which makes early investment even more accessible for organisations at our stage.

Theme 2: Know what to measure, and be honest when you cannot

What Has Worked

Outcome and impact-based metrics are intellectually appealing, attractive to funders, and encouraging to team members and potential hires. They are also, in practice, often outside the control of the people responsible for delivering them, and when working with government, almost entirely so. The shift we made in our policy team's OKR targets, from outcomes such as the number of partnerships formed to processes such as the number of quality meetings held and projects initiated, has so far served us well. It was linked to what the team could actually influence, and it reduced the demoralisation that came from missing targets.

That shift raised a harder question we have not fully resolved: how confident are we that our processes are causally connected to our outcomes when it comes to policy? We can track the number of officials we meet. We cannot easily track whether those meetings are moving us closer to adoption, or assess the quality of those engagements. This is a limitation of government engagement work, and one where I have not heard compelling answers on how to develop high-quality measurable targets that effectively motivate the right actions.

On the implementation side, we have built something we are proud of: a structured, multi-tiered quality monitoring system paired with an AI-enabled testing that cuts testing turnaround from over ten days to fifteen seconds. This is an innovation that we believe will have a meaningful impact on the likelihood of wheat flour fortification scaling, not just through our programme but across India and globally.

What Has Not Work

One area where measurement fell short, and where we under-invested, is local evidence generation, particularly for the government channel. Government decision-makers consistently asked for data from their own states and communities. In Maharashtra, we were explicitly asked to conduct local consumer acceptability studies before discussions could progress. In retrospect, we should have invested in academic and institutional partnerships earlier, not for fundraising or impact validation, but specifically to support government decision-making and to demonstrate that fortification works in an Indian context.

We did invest in two large primary data collection activities that proved genuinely valuable. A process evaluation conducted by Development Solutions gave us deep qualitative insight into how stakeholders think and feel about wheat flour fortification, and directly informed changes to our approach. A quantitative assessment by IDinsight gave us and our key partners primary data on the beneficiaries we are reaching and how much fortified flour they are actually consuming, providing meaningful confidence that our fortification efforts are driving real impact. Both of these were important. The lesson is that similar investments, directed at government audiences and earlier in the programme cycle, could have meaningfully accelerated our policy engagement.

Theme 3: Scale and sustainability are not the same

What Has Worked

Fortify Health is the only development sector organisation supporting major implementation of wheat flour fortification in India through private millers. We currently support over 80% of all millers registered as producing fortified wheat flour nationally, with an adoption rate of approximately 40 to 50% and a retention rate of over 95%. We take pride in these numbers. But they also reveal something we are still working on: we have built extensive reach within the market, while the broader ecosystem, including industry associations, large-scale millers, demand-side actors, and government stakeholders, has been slower to adopt wheat flour fortification. Without ecosystem ownership, the model's long-term sustainability continues to depend too heavily on Fortify Health's continued financial and technical presence.

What Has Not Work

On quality and operations, mills continue to rely heavily on us for troubleshooting and quality oversight. From a financial perspective, it will continue to be difficult to have mills invest meaningfully in key variable costs such as packaging. Further innovation and exploration are required on both fronts. Our recent WhatsApp-enabled AI troubleshooting tool is one step in this direction, but we plan to experiment more.

We were also slower than we should have been to engage large millers and retail actors early on. We focused on small and medium-scale mills, which was a pragmatic decision given where the opportunities were, but the credibility that comes with large-scale mill and retailer partnerships is disproportionate and can drive broader market awareness. This is something we are now investing significantly more effort in, and we hope it will pay dividends in the next 12 to 18 months.

The demand-side gap is perhaps our largest ongoing challenge. Market awareness of fortification and anaemia continues to be extremely low. We conducted a demand generation pilot in Madhya Pradesh earlier this year and found that generic in-store promotional materials did not significantly increase sales of fortified wheat flour. We are now developing a miller marketing package to support partner mills with more targeted, brand-led approaches, but this work is early-stage and impact is uncertain. We still do not have a clear, evidence-based answer to the question of how to build consumer pull for fortification, or even whether doing so is strictly necessary for the model to work at scale. This is something we intend to invest in and learn from over the next two years.

Theme 4: It's hard to know when to pivot your strategy

A true account of our government partnership work over the past three years is one of a logical strategy running into structural barriers we did not fully anticipate, and of a significant pivot that we believe positions us better for the future.

For three years, our team pursued a top-down strategy: engaging state governments directly, securing large-scale welfare pilots, and building toward a central government mandate. The structural barriers were consistent. Finding the right entry point in each state was persistently difficult. Senior officials who were accessible were rarely decision-makers, and those who were decision-makers were frequently transferred before agreements could be reached, resetting months of relationship building. More than once, we had a final contract in front of the final decision maker for signature, only for that person to be transferred the same day. 

The evidence catch-22 was a recurring obstacle: officials demanded local evidence before approving pilots, but pilots were the only way to generate local evidence. The Maharashtra Tribal Development Department, a relationship we had been cultivating since 2019, ultimately produced a clear answer in July 2024: the Secretary would not approve the project. Andhra Pradesh's pilot was cancelled by internal tendering disputes unrelated to fortification. Punjab produced no approved pilots after 18 months of engagement.

Internally, this period was difficult.. Years of effort by a committed policy team produced only a single, one-year pilot for a government programme. Keeping the team motivated through that required being honest about what we were learning, adjusting how we measured success, and ultimately being willing to conclude that the strategy itself needed to change.

In late 2025, we pivoted. Rather than seeking top-down permission at the highest levels of state and central government, we shifted to a bottom-up approach: generating local evidence through smaller, faster programmes with local bodies, civil society organisations, and quasi-government entities. This mirrors exactly how the rice fortification evidence base was built in India, through distributed pilots accumulating into a national case over time. A national convening we co-hosted in November 2025, attended by premix suppliers, government officials, and civil society actors, reinforced this logic and helped build the coalition that this kind of change requires.

We are confident about the prospects of this approach. We have built a model that is hassle-free. We have a network of more than 290 industrial-scale millers who would form the backbone of any large-scale government fortification programme. We are demonstrating that the model is simple enough, cheap enough, and good enough to implement at scale, and we are building the evidence base to prove it. We are seeing increased momentum through midday meal schemes and centralised kitchen programmes in particular.

The impact potential through government partnerships is substantial, both in scale and in the ability to sustain wheat flour fortification with significantly reduced philanthropic support over time. Much more remains to be learned, and mistakes will undoubtedly be made. But we believe the bottom-up approach represents a step change in what we can achieve as an organisation.

6. Transferable Lessons for Others

For implementers

• Build financial and procurement infrastructure at the same time as your people systems, not once the pressure arrives. The cost of catching up under scale is significantly higher than the cost of building early.
• In government engagement, consider whether a bottom-up evidence generation strategy may be right for your context. At least in India, the top-down approach requires more time, resilience, and political capital than most implementation timelines and funding grants can absorb.
• Build a replicable model and figure out the appropriate unit of replication early. Get the operational and logistical systems around that unit right, and then focus on how fast you can move through each unit. Speed and quality at the unit level compound over time and a lot of lessons get learned through doing rather than thinking. 

For funders and supporters

• Cost-effectiveness can be a a function of organisational investment, not just programme design. The difference between 5x and 12x over three years was largely operational: premix negotiations, onboarding systems, team deployment, and overhead management. Funding organisational health and capacity matters.
• Government partnership work does not fit neatly into annual cost-effectiveness analysis cycles. It requires patient capital, realistic time horizons, and a genuine willingness to fund strategy pivots when the evidence demands them.
• The marginal dollar in Fortify Health's government channel is not building a new model. It is converting a working model into a government-sustained impact at scale. There are open questions about how best to evaluate the feasibility of this theory of change, and we welcome engagement from funders who want to think through that with us.

For those considering working in this space

• This work can be slow and demoralising at times. You will engage with actors who may be philosophically misaligned, navigate systems that move at their own pace, and seek to serve beneficiaries who may never know your organisation exists. But it also has the potential to improve the lives of millions of people. I would strongly encourage anyone considering this space to go in with clear eyes, and to look for organisations that are honest about what they know and what they do not know, and that are willing to follow where the evidence leads rather than just the good stories.

7. Help us support millions more people

The government channel is where we believe the next order of magnitude of impact lies. We are seeking approximately $1 million over the next two years to fund evidence generation, district-level pilots, coalition building, and technical advisory work with the government. Our target is to reach more than 10 million beneficiaries through 60 or more partnerships by December 2028.If we reach this, we expect the cost-effectiveness of the government channel to exceed even our private miller work. 

We want to be honest about the uncertainties: there are realistic futures in which we significantly exceed this target, and realistic futures in which we do not get the traction we are hoping for. What catalytic capital would enable is investment in the people required to build long-term, trusting relationships with government partners, and who can act as technical supporters on the implementation of key projects. These are not short-term hires. They are the foundation of a strategy that we believe could ultimately sustain wheat flour fortification across India with significantly reduced philanthropic support.

Iron deficiency anaemia affects hundreds of millions of people in India. The infrastructure to address it at scale exists. What is needed now is the evidence, the champions, and the sustained support to make fortified wheat flour the default, not the exception. If you want to be part of that, we would welcome the conversation. 

Contact: [email protected]