What a flood response looks like thirty days in
The first week of a flood response is well covered by everybody. The fourth week is not, and that is where the outcome is actually decided. Water and food arrive; sanitation and menstrual provision do not, because they are treated as secondary until they become the reason people are ill.
The argument of this note is that hygiene provision belongs in the first 72-hour plan rather than the recovery plan, and that the handover between emergency response and long-term programming is the single most reliable place to lose a population.
The school days nobody counts
Attendance registers record whether a girl was in school. They do not record why she was not, and the monthly pattern that would make the cause obvious is invisible unless somebody goes looking for it.
This piece sets out the instrument we use to recover that pattern — participant recall cross-checked against the register — and why a school-days-retained figure is a more honest unit of measurement than products distributed.
Why most impact numbers do not survive an audit
Four failure modes account for most of it: no baseline, attrition absorbed into the result, attribution claimed without a comparison group, and reach reported in place of outcome. None of them require bad faith — each follows naturally from designing measurement after delivery rather than before it.
The paper works through each in turn and sets out what a result has to carry to be usable in a disclosure, an assurance process or a board paper.
Reusables only reduce waste if people keep using them
Waste-displacement figures for reusable products are routinely modelled on the product's rated lifespan. That assumes continued use, and continued use depends on conditions that have nothing to do with the product — water availability, private drying space, and household acceptance.
We set out a method that models displacement on observed reuse at follow-up rather than rated lifespan, and show how far apart the two figures can be.
Rangeland loss is reaching households before any policy does
Grazing land degradation shows up in household income and school enrolment long before it appears in any adaptation plan. By the time it is legible at a policy level, the households that were most exposed have already made the decisions that will define the next decade for them.
A short argument for treating slow-onset climate hazard with the same urgency as acute events.
Distribution is not an intervention
The most common design failure in product-based social programmes is treating the handover as the intervention. It produces a clean logistics number and no behavioural change, and the follow-up data — when anyone collects it — shows the products unused.
This paper sets out the minimum viable combination: product, education in the same session, and follow-up at a fixed interval. Remove any one and the outcome does not hold.