4 · Methodology

Owner: Steven Uggowitzer + Simon Adebola · Editors: Pallavi Dani, Stephanie Berrada Effort: ~3.5 person-days (heaviest complementary lift in D1) · Due: v0.5 EOD Fri 2026-09-04 · v0.8 Fri 2026-09-11 · final Wed 2026-09-17

What WHO expects here

The single most-scrutinised chapter of the Inception. WHO’s evaluation office reads it twice. It must (a) defend the mixed-methods design, (b) show how the scoring aggregates without hand-waving, (c) name the AI-usage discipline transparently, (d) address ethics + consent + confidentiality, (e) include plain-language explainer boxes so a lay reader can follow the reasoning.

Rewrite cycle expected.

Ultimate contents

Overview: three discovery streams triangulated

Stream 1 — Documentation ingest (official). WHO/AFRO documentation, IT register, cluster-level materials, AFR-RC74-7 lineage.
Stream 2 — HQ + partner intelligence. HQ Modernization Task Force spreadsheet, HISP SA design docs (via Steven’s channel), Fjelltopp CKAN docs (via Droti), prior AFRO evaluations.
Stream 3 — Shadow-IT DNS + certificate-transparency sweep. SSLMate CertSpotter enumeration of AFRO-adjacent domains + WHOIS + TLS cert issuer chains. Ethics and provenance transparency: publicly resolvable DNS + public CT logs are standard reconnaissance and break no WHO IMT policy. The methodology annex names this stream openly.

The three streams triangulate. A platform surfaced by only one stream is flagged as low-confidence; a platform surfaced by two or three streams is high-confidence.

KIIs — sampling, protocol, coding

13–18 semi-structured KIIs (upper end targeted). Sampling frame in Chapter 6. Protocol in Chapter 8 (both EN and FR versions). Coding schema: thematic + inductive; two-analyst consensus per interview; Simon leads coding.

Platform walkthroughs

6–8 hands-on sessions per the bid commitment. Walkthrough checklist (T3 in the master plan Part E): workflow trace, sample data record, screen recording (with permission), export formats, error handling, admin surface, user roles, monitoring/logging surfacing.

Scorecard — how the eight dimensions aggregate

Full anchor definitions per level (0–4) in Chapter 9. This chapter shows how the eight dimensions aggregate to a per-platform maturity read, and how those per-platform reads aggregate to a per-AFR-RC74-7-pillar view. Worked example: DHIS2-AFRO scored end-to-end so a reviewer can trace one platform’s score from evidence to pillar rollup.

AI-usage discipline (with forward pointer to Chapter 12)

Named tools (Claude Opus for reasoning, transcription APIs for KII coding acceleration, afro-terminology for consistent FR rendering). Named guardrails: no AI-generated claims without human evidence-check; no AI-generated citations without lookup; all AI-authored output attributed and version-tracked. This chapter carries the transparency; Chapter 12 carries the substance.

Ethics posture per interview type. Consent workflow (T6 in the master plan). Data-retention commitment (per Annex B of the RFP). Anonymisation policy for the client-facing risk register and annexes.

Method limits

Honest paragraph on what the method cannot see: platforms behind country-specific firewalls; platforms not on public DNS; personal-scale reporting workflows below the “above-personal-desk” definition threshold.

Plain-language explainer boxes

Two or three call-out boxes in this chapter aimed at the lay reader. WHO reviewers appreciate accessibility signals; skipping them is a common failure mode.

Evidence to draw on

RFP/AFRO/013/2026 (methodology requirements)
IQVIA Technical Approach §Methodology
WHO SCORE assessment methodology (as a reference standard)
workspace/plan/00_MASTER_PLAN.md Parts C, D, E, F
Chapter 12 (AI methodology annex) drafts

Complementary effort triggered

Mixed-methods design justification. Scoring-aggregation reasoning. Plain-language explainer boxes. Two-analyst rewrite cycle expected. Simon co-owns; Steven pressure-tests.