Client. A foundation that runs social and environmental programs alongside indigenous communities in the Amazon. Four concurrent programs, a budget on the order of USD 100K, field teams, and international donor accountability standards.
Approach
This case lives at the intersection of two links in the method: the sociology of organizations and data engineering. Social impact data is not a neutral number: it is born from a human process —the field work of people on the ground— and is reported to another —the donors—, and between the two lie cultures, languages, and power asymmetries. Modeling that data without reading its human fabric produces reports that meet the format but do not tell the truth.
The thesis that orders the work is contextual data quality taken to the most demanding terrain: sensitive community data, where a poorly designed access control is not a technical flaw, it is harm to real people.
The problem identified
Field operations were recorded by hand: spreadsheets, paper, people’s memory. That produced three chained problems: the reporting effort consumed time that should have gone to the program; visibility of funds arrived late, when it was no longer possible to correct course; and sensitive information circulated without a clear control of who accesses what.
The stated problem was “we need better reports for donors.” The real one came earlier: there was no reliable, governed data layer beneath those reports.
Functional assessment
I mapped the four programs as distinct operating systems but with a common spine: inputs (resources, field activities), outputs (impact indicators), and an accountability circuit. The assessment crossed three types of indicator that almost never coexist in the same head —financial, operational, and impact— and separated, for each program, what could be digitized without breaking how the teams actually work.
Building the solution
- Digitization of field operations. I converted manual records into structured datasets, cutting manual reporting effort by roughly half —time that returns to the program.
- Operational and analytical models that integrate financial, operational, and impact indicators into a single reading, not three reports that do not talk to each other.
- Budget and resource monitoring with near real-time visibility of funds, so that a deviation is seen while there is still time to act, not at closing.
Information and data layer
Governance was the center, not an appendix. I designed access controls for sensitive social and environmental community data, and governance practices that improved compliance with donor standards and left the operation in a state of audit readiness. The rule that held it all together: what cannot be governed responsibly is not digitized —and when it is digitized, the people the data represents are protected first.
How the work was conducted
I coordinated cross-functional teams of up to ten people, in a context of low connectivity and high cultural sensitivity. The method was field-based: capture instruments adapted to how people really work, not to what a form demands; and a traceability discipline designed to withstand a donor audit. There was no automation with AI agents —it did not fit the era or the context—; there was sociotechnical design applied to one of the most demanding environments possible.
What this case proves
- The bridge in its purest form: reading the human fabric of communities and, at the same time, modeling and governing the data that represents them.
- Real governance of sensitive data: access controls and audit readiness before donors, not a policy in a PDF.
- Measurable impact: less reporting effort, more visibility of funds, better compliance —without losing sight of whom the data protects.