Article summary
Preserve legacy Swaziland and current Eswatini source labels, create a canonical country key, then review WhatsApp avatar visibility.
When one list contains “斯威士兰,” Swaziland and Eswatini, the first issue is master-data normalisation, not three types of customer. Create one country key before avatar checking while preserving every row’s source spelling.
Keep three columns during normalisation
| Field | Example | Purpose |
|---|---|---|
| country_source | Swaziland | Trace the raw value |
| country_canonical | Eswatini | Consistent reporting |
| normalised_at | Date | Version audit |
Do not erase history with string replacement
A global overwrite loses provenance and can damage orders, contracts or legacy joins. Add a canonical key rather than deleting the original field.
Let the numbering plan use its own evidence
Preserve phone_raw and country_evidence. A country-name field may help provenance review, but cannot assign a prefix to an unsupported national string automatically.
Keep avatars out of country-name decisions
Photos, flags, visible words and clothing cannot select a country label. They also cannot establish citizenship, community or identity.
Check duplicate entities before merging
Legacy and current names may place the same phone in two groups. Use an internal row key and controlled deduplication for relationships, not visual identification.
Submit deduplicated phones only
TXT contains one phone per line and no country label or customer detail. Excel lands in staging and reconnects through the crosswalk to the canonical country key.
Timestamp avatar states
Record avatar_observed, not_observed, visibility_limited and unknown separately. Country-name normalisation does not turn an old result into a current fact.
Monitor reporting changes after normalisation
Compare record, duplicate and unknown counts before and after merging so changes reflect classification repair rather than lost rows.
The value of the Eswatini page
Normalise different names for the same country into a traceable dimension before discussing profile visibility. This supports both legacy search terms and current data governance.
