Article summary
Prepare Senegal +221 phones and distinguish avatars, business-account results and mappings; a logo is not personal identity and a blank is not an invalid lead.
The useful outcome of a Senegal WhatsApp avatar check is not “face recognition.” It is a review that separates picture state, business-account result and phone mapping. A brand logo may belong to a genuine merchant, a portrait may not depict the holder, and a blank avatar does not make a lead invalid.
Preserve provenance before +221 normalization
Store phone_raw, country_evidence and contact_id first. Remove display punctuation and validate +221 under the current Senegal plan. Country or length conflicts enter an exception queue instead of being padded by guesswork.
| Record | Handling | Do not infer |
|---|---|---|
| Reliable +221 structure | Candidate for TXT | Current physical location |
| Country unknown | Return to provenance | Country from a person’s name |
| Duplicate phone | Retain the deduplication crosswalk | Delete the customer relationship |
The actual WS Avatar schema
The task returns phone, business-account result, avatar and WhatsApp mapped phone. It does not return activity time, age or gender. A picture-only question does not justify the wider Full Format task.
Record logos and portraits without inventing identity
| Observation | May record | Must not assert |
|---|---|---|
| Brand logo | Avatar exists / type clue | Legally registered company |
| Portrait | Image is visible | Holder or decision-maker |
| Group / landscape | Original task state | Family, profession or interests |
| Blank | Unknown class and date | Fake account |
TXT still contains phones only
AIPUSH accepts one normalized phone per line as TXT. Company name, industry, picture interpretation and permission remain internal. Reconnect Excel through contact_id after return.
A business account is not a credit score
The field does not establish company size, legal registration, ability to pay or likelihood to purchase. B2B verification needs independent corporate, contract or payment evidence rather than avatar and account type.
Split blank avatars into meaningful states
Not returned, not visible, processing error and explicit no-image states remain separate with checked_at. Calling every blank “no avatar” corrupts repeatability and quality analysis.
Mapped phones belong in a relationship table
One-to-many or many-to-one collisions pause automatic updates. A mapping never replaces the customer master directly; retain source phone, batch and accountable reviewer.
Never infer sensitive attributes from the picture
Do not guess ethnicity, religion, health, age, gender or income from an avatar, and never use it to determine price, service or marketing priority. Restrict image-field access and define a deletion date.
How to accept the Senegal batch
Report +221 format pass rate, avatar coverage, business-account returns, missing-state categories and mapping conflicts. Avatar checking improves data explainability; it is not an automated judgment about customer authenticity.
