Case Study: Inflated Meal Counts and a Connected Vendor in a School Nutrition Programme

The programme supplemented the government's mid-day meal scheme with milk, eggs, fruit and fortified biscuits, and claimed 3,500 meals a day across 40 schools for 200 days: 7,00,000 meals a year.

The funder's supply chain team worked out the raw materials 7,00,000 meals would need and asked for the vendor invoices. The quantities broadly matched. But every item came from a single vendor, and every price was 35% to 45% above wholesale.

Unannounced school visits

• Northrop visited 12 of the 40 schools without notice during the meal hour, recording whether the supplement was served, how many children received it, what was served, and obtaining a signed statement from the headmaster or meal coordinator.
• 8 schools: supplement served regularly, as described, to about 700 children in total, in line with the claimed average of roughly 88 per school.
• 2 schools: service irregular, two to three days a week instead of five.
• 2 schools: nothing received this term. One headmaster said the NGO had come for a few months the previous year and then stopped.

Extrapolated to all 40 schools: about 27 served regularly, 7 irregularly and 6 not at all. That indicates about 2,650 meals a day, or roughly 5,30,000 a year against 7,00,000 reported, an overstatement of about 24% (Grade C).

The central kitchen's equipment could process roughly 900 to 1,100 servings per cycle. The NGO's own LPG refill invoices supported two to three cycles a day, or about 2,300 to 2,800 servings. That independently corroborates the field estimate and is inconsistent with 3,500. Storage had no cold chain: milk was bought daily and eggs were kept in an unventilated room at ambient temperature.

All Rs 52 lakh of food procurement went to one proprietorship, referred to here as Vendor A. Its registered address was a residential flat. The proprietor named on its public GST registration was the NGO founder's mother-in-law, which the founder confirmed in the management interview. The relationship had not been disclosed to the funder.

Vendor A was registered three months before the CSR contract began and had no warehouse, cold storage or delivery vehicles. Its invoice numbers to the NGO ran consecutively, indicating the NGO was its only or near-only customer. Field staff told Northrop that food was actually bought for cash at the local wholesale market. Vendor A declined to share its bank statements, so where its margin ultimately went could not be traced.

At market rates, the quantities invoiced would have cost about Rs 37 lakh, so pricing above market accounts for about Rs 15 lakh. And because only about 5,30,000 meals were served, roughly a quarter of the invoiced quantities were not needed: about Rs 9 lakh at market rates. A further Rs 8 lakh of "programme coordination" was paid to two individuals whom no school or NGO staff member could identify.

Reported cost per meal was Rs 13.60. After removing the gaps above, the cost per meal actually served was about Rs 11.90.

The core activity, feeding about 2,650 children a day, was real and valuable, so the funder restructured rather than exited. Procurement moved to the funder's own supply chain team, headmasters began signing daily meal registers, Northrop was engaged for quarterly unannounced visits, and funding was resized from Rs 95 lakh to Rs 60 lakh to match actual output at market prices. The founder stepped back from programme management, and the funder appointed a new programme coordinator.

Not every finding should end a partnership. Sometimes the right answer is to keep the programme and remove the leak. That decision is only possible when you can see exactly where the leak is.