Article summary
A step-by-step WhatsApp registration-check workflow covering normalization, deduplication, TXT preparation, Excel acceptance and safe CRM rejoining.
A WS Registration Check is successful when every result can be traced back to a correctly prepared phone and interpreted as a dated registration observation. The job is not simply “upload and download”; input quality and result acceptance determine whether the Excel output is usable.
Define the question and cohort
Write down why registration status is needed, which records qualify and when the result expires. Exclude unknown-source, opted-out or irrelevant records before file preparation. More input rows do not improve a poorly defined purpose.
Preserve the original phone
Keep phone_raw as text and create phone_normalized under a current country rule. Use explicit country evidence for local notation. Put country conflicts, extensions, impossible lengths and suspected truncation in an exception queue.
Run a pre-upload checklist
| Check | Pass condition | Failure action |
|---|---|---|
| Format | International candidate validates | Exception |
| Duplicate | One normalized phone | Retain cross-reference |
| Source | Documented provenance | Quarantine |
| File | One phone per TXT line | Regenerate |
Create a separate crosswalk
Map batch_row and normalized phone to contact_id inside your environment. Do not upload name, company, order or permission with the TXT. The crosswalk is what lets Excel return safely to the source system.
Select WS Registration
AIPUSH WS Registration returns phone and whether registration was observed. It does not return activity, avatar, Gender and Age or conversations. If the business question changes, create a new approved task rather than silently widening this one.
Accept the Excel file in staging
Preserve the original workbook, import phone columns as text, validate expected headers and add received_at. Count returned, not registered, unknown, missing and duplicate separately. Unknown must not be forced into “no.”
Rejoin without overwriting
Use the batch crosswalk, not row number. A clean one-to-one match can update a dated observation field; collisions enter review. Registration status cannot change consent, suppression, identity or the CRM primary phone.
Calculate a defensible rate
The denominator is unique, format-valid, successfully checked phones in the intended cohort. Report formatting exceptions and task unknowns outside it. This prevents a high “registration rate” produced by dropping difficult records invisibly.
Close the batch
Record task version, counts, operator, checked_at, allowed use and delete_after. Remove temporary TXT and restrict the result workbook. Recheck only when a current business decision requires it—not on an automatic whole-database schedule.
