KoboCollect vs satellite imagery for a rural WASH baseline

If you've scoped a WASH baseline before, you've run into this comparison already. KoboCollect (and ODK more broadly) is a form on a phone. Satellite imagery is a picture from above. They answer different questions, and the mix-up happens when programme teams ask imagery to do a survey's job, or send enumerators to do a census job imagery handles in a tenth of the time.

What each one is built for

KoboCollect, and the ODK water point survey apps built on the same engine, exist to capture attributes nobody can see from the air: functionality status of a handpump, whether the tap stand has a working tariff system, the depth reading on a borehole log, a latrine's pit lining material. That's interview-level detail. An enumerator walks to the point, talks to the caretaker, fills the form, and the GPS tag gets recorded as part of the same visit. No satellite will ever tell you a pump handle is broken but the well itself is fine, or that a household stopped using a latrine because the door won't close.

Satellite imagery does the opposite job well. A single VHR pass over a district shows you where structures exist at all: every borehole platform, tank stand, and latrine block visible from above, dated and located. It doesn't know if the pump works. It does know that there's a concrete apron and a rising main in a settlement your last baseline list never mentioned, because that settlement grew after the list was made.

Where the sequencing matters

The failure mode we hear about most isn't "we used the wrong tool." It's "we sent the Kobo team in the wrong order." Teams build their point list from the last baseline, a district WASH inventory, or whatever shapefile got handed over at the last partner meeting, then send enumerators to visit each point on that list. The problem is the list was built from data, not from the ground, and under-mapped districts have a habit of growing new hamlets, new boreholes drilled by a different NGO, and new latrine blocks nobody logged. A KoboCollect team working strictly off an old list will walk past all of that and never know it exists.

Flip the order and the gap narrows. Pull a dated inventory of every borehole, tank stand, and latrine block that shows up on recent imagery first, then build the Kobo form list from that instead of from last year's spreadsheet. The enumerators still do the interview work imagery can't do, but they're walking to points that are there, including the ones built since anyone last updated the district register.

A rough division of labor

Treat it less as mobile data collection vs remote sensing WASH, and more as two passes over the same district. Imagery does the first pass: what's there, where, as of this year. Field teams do the second pass: what condition it's in, who uses it, what's broken. Running them in that order means the KoboCollect team spends its days on structures, not on the parts of the district where nothing has been drilled yet. It also means a facility built last dry season shows up in the baseline instead of waiting for someone to stumble on it during a different survey round.

None of this replaces the interview. A functionality score, a water quality note, a tariff collection record, those still need a person standing at the point with a form. What changes is how confidently you can say the list of points is complete before you spend a field team's time on it.

A yearly imagery pass over your draw area is what we build at Sanitation Mapping, to feed into that first step, so the inventory going into your Kobo form list reflects what's actually on the ground this year, not the register from two field seasons ago.

If your baseline list hasn't been checked against recent imagery, that's worth doing before the next field round goes out.