Article summary
Separate Turkey +90 ten-digit phone normalization from avatar acceptance, preserving timestamps, nulls and evidence without inferring identity, religion or spending power.
A Turkey WhatsApp Profile Picture check produces a snapshot of whether an avatar was observable, not a verified photograph of a customer. Phone format, platform mapping, avatar reference and human review are four separate evidence layers. Compressing them into a “real user” label distorts both marketing and risk decisions.
Normalize +90 under a current numbering rule
A Turkish regulator notification published through the ITU’s Operational Bulletin identifies country code +90 and a ten-digit national significant number. National dialling notation is not retained in international form. Conversion still requires Turkey provenance; do not prepend +90 to an arbitrary column of ten digits.
Keep four evidence layers separate
| Layer | Record | Does not prove |
|---|---|---|
| Phone | Raw, normalized value, rule version | Current holder |
| Mapping | Input and mapped WhatsApp phone | Same natural person |
| Avatar | Reference/null and observed time | Authentic photograph |
| Review | Purpose, judgment and reviewer | Official identity verification |
Quarantine national notation and mixed countries first
Keep phone_raw as text. Explicit +90 and 0090 forms can be parsed under a documented rule. Convert national notation only when CRM country, form choice or contracted market supports Turkey. Wrong length, country conflict, extension and notes enter an exception queue; digits are never dropped or added to manufacture a pass.
What WS Profile Picture actually returns
AIPUSH WS Profile Picture may export phone, business-account observation, avatar and mapped WhatsApp phone. A null avatar can reflect no picture, privacy settings, temporary visibility or the observation boundary and must not be collapsed into “fake phone.” A business-account observation is not an endorsement of the business.
Use an avatar state machine
| State | Meaning | Next action |
|---|---|---|
| observed | An avatar was observable at check time | Apply a short lifetime |
| not_observed | No avatar was returned | Retain unknown cause |
| mapping_conflict | Mapped phone differs from input | Block automatic merge |
| expired | Business-defined lifetime passed | Stop use or delete |
Manage TXT separately from image observations
The upload is TXT with one phone per line and no name, order, region, religion or permission. An internal crosswalk retains batch_row and contact_id, while Excel lands in staging. If there is no continuing purpose, do not turn an image reference into a permanent CRM avatar, download it again or run face recognition.
A Turkish name and avatar are not sensitive-attribute evidence
An avatar, name, clothing and background cannot reliably establish religion, ethnicity, political opinion, health or economic capacity. Marketing should not create individual sensitive labels from them. Localize content with the user’s stated language and market information, not visual guesswork.
The appropriate use is data quality, not identity judgment
Low-risk uses include measuring observable-avatar coverage, identifying batch anomalies and manually reviewing mapping collisions. The result is unsuitable for credit, service eligibility, account trust tiers or deciding whether a person deserves contact. Registration, activity and permission remain independent questions.
Acceptance reports denominators and nulls
The batch record includes unique inputs, format exceptions, mapping collisions, avatar observed/not_observed, business-account coverage, checked time, field lifetime, access roles and deletion date. Stratified null reporting distinguishes a data problem from an observation boundary instead of rewarding a visually impressive hit rate.
