Field case
A field engagement with the MiMaji Foundation on what changes when the households paying the most for water can finally see who charges what, and whose water is safe. The Lab is running the reach, trust and behaviour readings that decide whether an open-data intervention lands where the asymmetry sits.
MiMaji publishes water-price and water-quality data as an open public map; the Lab is measuring whether that map reaches, is trusted by, and changes the behaviour of the households on the wrong side of Nairobi's water-market asymmetry.
Nairobi's water story is not primarily a story about pipes. It is a story about information. In the low-income settlements, households can pay several times more per litre than households in wealthier neighbourhoods, often for water of unknown and variable quality, bought from vendors in a market with almost no visible pricing. The technical means to test water and to map a network exist. What has been missing is transparency: a shared, public picture of who charges what, and whose water is safe.
MiMaji works on exactly this, through community water-quality testing, WASH education in schools and community groups, and MajiMap, its public open-data map of water quality, price and access across the city. The intervention is not a pipe. It is a change to the informational environment in which price is set and safety is assessed, and its success or failure is measured on the same terms.
That is a socio-technical question before it is a data question, and it is the question the Lab is engaged to answer.
The Lab's reading is running in three overlapping streams. Together they produce evidence at reach, depth and experience level, which is the same three-part discipline we describe in Impact Measurement.
A rolling in-depth interview cohort across the settlement clusters MiMaji is active in, sampled to include households on both sides of the price asymmetry, the young women and older women who most often carry the water, and vendors themselves. We are asking a small set of testable questions: do you know MajiMap exists; if you do, when did you last consult it; if you consulted it, did anything you did next change. Reach without behaviour change is diagnostic. Behaviour change without reach is a different story that needs its own explanation.
Open data changes behaviour only when it is believed. The Lab is mapping who residents credit as the source of a price or quality claim — MiMaji, a neighbour, a vendor, a chief, a health worker, a phone rumour — and where in that ranking MajiMap actually sits. In a market thick with rumour and vendor interest, credibility is the hinge. Documenting it is a first-order finding.
Whether a vendor changes a posted price when a comparable price becomes publicly known is the observable version of the whole argument. The Lab is running repeat price and quality observations at a set of vendor points over the engagement, alongside conversations with the vendors themselves. Movement in the observed price, one way or the other, is the strongest signal that the intervention has begun to bite.
None of the following is a final finding. The engagement is live, and the honest position at this stage is that the pattern the fieldwork is surfacing is consistent enough to name, and the counter-evidence is worth stating too.
Reach follows the phone that already had a browser open. The households consulting MajiMap first are the ones who already had smartphones with data, who already spoke the language the map is in, and who already had a habit of looking things up. That is exactly the population the transparency is least needed by. Extending reach to the households on the wrong side of the asymmetry is a distribution problem, not a product problem, and it requires channels that are not the map itself.
The credibility of MiMaji as source is high where MiMaji has been physically present. In settlements where testing has been visibly done in the open, on someone's tap, with a result written down and left with the household, the map is treated as evidence. In settlements where testing has happened elsewhere and only the aggregated result is on the app, the same map is treated as opinion. The physical act of testing is doing more work than the data layer sitting on top of it. That is important, because it means the map's credibility is a downstream product of MiMaji's ranger-style presence, not an intrinsic property of the data.
Vendor behaviour moves slowly and unevenly. In some sites there is early evidence of vendors adjusting posted prices when a comparable price is known and consulted, in a way that would not be visible from the utility side. In others there is no observable movement, and vendors report that the households they sell to are not the households consulting the map. Both patterns are informative.
The intervention is doing work that the map alone would not have done. WASH-education partnerships in schools and community groups appear, in the fieldwork so far, to be the mechanism by which knowledge of the map reaches households that the map alone would not have reached. That is worth naming, because it argues against reading MajiMap as a stand-alone data product and for reading it as a bundle: physical testing plus schools work plus public map, in that order.
Not everything we are hearing supports the reading above. Some households consulted the map once, found the vendor nearest to them not listed or listed with stale prices, and stopped. Coverage gaps are the fastest way to lose the credibility MiMaji has otherwise earned, and the fieldwork is surfacing that as a first-order design pressure — arguably more than the accuracy of any individual reading.
We read MajiMap not as a data product to be assessed for accuracy, but as a socio-technical intervention to be evaluated for effect. It pairs Delft-rooted water understanding with primary fieldwork in the settlements, asking who reaches the information, who trusts it, and what changes when the price of not knowing finally has a number attached.
Through the BRW framework, MajiMap reads as a weaken strategy against a soft but powerful barrier: the information asymmetry that lets an opaque water market persist. It does not build a parallel water system (bypass) or redirect the utility (repurpose); it erodes the informational foundation on which the inequity rests. That reading tells the Lab exactly where to look for success or failure, which is in whether the transparency actually reaches and is trusted by those on the wrong side of the asymmetry — which is the reading we are running.
The engagement continues through the next monitoring cycle. Two things sit at the top of the near-term list.
Household reach for the households that most need it. The distribution problem named above is the biggest single lever the intervention has right now, and the Lab's next round of fieldwork is focused on which non-map channels (health workers, school communications, community meetings, low-bandwidth messaging) actually deliver map awareness to the households the map is meant to serve.
A defensible before-and-after read on vendor pricing. Longer observation at a stable set of vendor points, with matched controls where possible, is what turns the early pattern above into either a finding or a refutation. That is a discipline, not a promise; the finding will be whatever the data supports.
Water transparency in Nairobi is a specific case of a general pattern the Lab studies everywhere: technologies whose effect is informational and distributional rather than physical. The same logic runs through open data in energy, health and agriculture. In each, the question is not whether the information is correct, but whether it lands where the power imbalance sits, and shifts it.
This engagement supports the Lab's Water & Sanitation programme. For MiMaji's own work, see mimaji.org. For the Lab's field methods, see Field Research, and for how impact is measured across reach, depth and experience, see Impact Measurement. To discuss a study of your own, see Contact.