Article summary
Use country evidence and numbering plans to separate Guinea, Guinea-Bissau and Equatorial Guinea before interpreting WhatsApp avatar blanks.
A string match on “Guinea” can place three different countries in one batch. If admission is wrong, avatar blank rates and market conclusions for Guinea +224 are distorted before checking begins.
Normalize country names first
| Source value | Canonical field | Handling |
|---|---|---|
| Guinea | Explicit country ID | Verify provenance |
| Guinea-Bissau | Separate country ID | Do not include |
| Equatorial Guinea | Separate country ID | Do not include |
| Ambiguous translation | country_unknown | Human review |
Do not filter by “contains Guinea”
Use a controlled country ID, user selection or order address rather than a name substring or folder.
Use numbering plan as cross-evidence
Preserve phone_raw and original international code. +224 may support a Guinea candidate but does not prove residence or identity alone.
Remove country conflicts from the avatar denominator
Report country_conflict and format_exception separately. They are not no-avatar outcomes.
Keep avatar states limited
Separate observed, not_observed, visibility_limited and task_unknown. An image cannot infer country, ethnicity or language.
Generate TXT by country evidence
Each batch contains one phone per line and a country_cohort_id. Returned Excel goes only to its source queue.
Classify mapped phones independently
A mapped phone retains its own country evidence and does not inherit the submitted label or merge CRM across countries automatically.
Report name-matching errors
Count string-rule admissions and human corrections so the upstream dropdown and translation can be fixed.
The distinct Guinea answer
Prove which Guinea first, then discuss avatar visibility. Country disambiguation is the real first step.
