Designing Third-Party Monitoring for Hard-to-Reach Districts

Remote and area-based third-party monitoring is only as defensible as the chain-of-custody behind it. In districts of Somalia where direct donor staff visits are restricted, the methodology has to do the work that physical presence would otherwise do — on enumerator selection, sampling, verification, and reconciliation. This note walks through the design we use, the failure modes we have seen elsewhere in the region, and the parts that are non-negotiable.

Why Third-Party Monitoring Is a Methodology Problem, Not a Resourcing Problem

It is tempting to treat third-party monitoring (TPM) as a logistics exercise: hire enumerators, send them to verify, file a report. In practice, every weakness in that pipeline compounds. A loose sampling rule, an unbriefed enumerator, an unchecked photograph, and the deliverable is doing the opposite of what donors hired it to do — it is laundering uncertainty into apparent assurance.

The OECD-DAC criteria do not get easier to apply in fragile contexts. They get harder. The methodology has to compensate for the absence of direct donor staff visits, not assume those visits as a fallback.

“The honest test of a TPM design is whether a sceptical donor evaluator can follow your chain from sampling frame to verified observation without you in the room.”

That sentence is the whole specification. Everything below is how we operationalise it for Somali Federal Member States and the donor programmes funded in them.

Enumerator Selection & Briefing: The Non-Negotiable Starting Point

Every defensible TPM design begins with the enumerators. For a Somali assignment, that means working with field leads who already operate inside the relevant districts — not parachuting in. Briefing is structured, not anecdotal:

A signed code of conduct, with the protected-disclosure mechanism written into it.
A district-by-district security brief sized to the specific routes the enumerator will follow.
A do-no-harm checklist tailored to the activity being verified.
Per-instrument briefing on the verification questions and the GPS / photographic evidence the enumerator must collect.
A reconciliation slot at the end of each enumeration cycle to walk back through anomalies before the report is finalised.

Where an enumerator cannot personally visit a site, the brief specifies the area-based monitoring alternative — including the named local proxy reporter, what they are authorised to verify, and what they are not.

Sampling for Verification, Not Coverage

The single biggest design error we have seen in TPM elsewhere is confusing sampling for coverage with sampling for verification. They are different problems. Coverage asks: did the activity reach the planned beneficiaries? Verification asks: can we trust the implementing partner’s reported numbers?

Our sampling rule for verification is risk-weighted, not random-uniform. High-risk districts, high-value transfers, and partners with recent reconciliation issues get a higher sampling fraction. Lower-risk activities are sampled at a baseline floor. The rule is documented in writing; donors can audit why a given site was or was not visited.

Remote and area-based monitoring — Area-based monitoring fills the gap where direct enumeration is restricted. The methodology defines what counts as evidence, who is authorised to collect it, and how it is reconciled against the implementing partner’s reports.

Chain-of-Custody for Evidence

Every photograph, every GPS coordinate, every signed beneficiary list has to carry a chain of custody — from the enumerator’s device through to the reviewer’s desk. We use:

Time- and location-stamped photography on KoBo / ODK forms, with the location buffer documented per district.
Enumerator-signed digital declarations attached to each completed submission.
A reviewer who did not collect the data flagging anomalies before reconciliation with the implementing partner.
A locked archival copy of every submission, time-stamped at upload, available for donor audit on request.

None of these is novel in isolation. The discipline is in applying all four on every submission, with no exceptions for difficult districts.

Reconciliation, Not Accusation

A finding of variance against an implementing partner’s report does not, in itself, mean fraud. It means a reconciliation conversation has to happen. Our methodology builds in a structured reconciliation step: the partner sees the variance, has a defined window to respond with evidence, and the TPM report records both the variance and the partner’s rebuttal. Donors see the full record.

That structure protects implementing partners against unfair findings, and protects donors against laundered ones. It is the part of the methodology that most rewards being written down in advance.

What We Are Deliberately Not Claiming

We are not the only TPM provider in Somalia, and we do not claim to have solved every access-constrained district. Some districts genuinely cannot be verified directly today; the methodology has to be honest about that and document it as a limitation rather than paper over it. A TPM report that admits where it could not see is more useful than one that pretends otherwise.

What Comes Next

We are currently piloting a more structured area-based monitoring protocol for one of our donor counterparts, with reconciliation rules built directly into the digital form rather than handled at desk-review stage. If you are commissioning TPM in Somalia and want to compare notes on methodology, get in touch via the contact page.