Article summary
Explain actual Viber Registration and Activity fields, serial/parallel composition, timing and reconciliation—and the absence of avatar or high-net-worth fields.
AIPUSH Viber Registration and Activity can be combined in one business workflow, but the two tasks do not form an imaginary “full-profile” export. Registration returns phone and registration result. Activity returns phone, active days, mid and user offline time. Current Viber tasks have no avatar, age, gender, skin-tone or high-net-worth field.
Put both schemas on the table
| Task | Export fields | Question |
|---|---|---|
| Viber Registration | Phone, registration result | Was platform registration observed? |
| Viber Activity | Phone, active days, mid, user offline time | What identifier and activity were observed? |
A serial design can reduce the Activity cohort
If Activity has higher processing or review cost, run Registration on a well-sourced set first. Run Activity only on the subset with registration observed, continuing purpose and no opt-out. Give the two runs different batch_id and task versions. A record excluded from stage two is not “inactive.”
A parallel design measures coverage
When the research question needs unknown and coverage rates for each task, run both against the same frozen sample. Join results in staging through normalized phone and crosswalk. Different row order, duplicates and missing mid become explicit exceptions.
The combined-state matrix
| Registration | Activity | Interpretation |
|---|---|---|
| Observed | Time returned | Two timestamped platform observations |
| Observed | Unknown | Not “inactive” |
| Not observed | Historic return | Review dates and identity collision |
| Unknown | Error | Preserve uncertainty and inspect process |
mid and phone are not the same key
Keep observed Viber mid separate from CRM contact_id and phone. One mid joining several contacts, several mids for one contact or change across dates belongs in a collision queue. Never merge silently.
What the absence of an avatar field means
The current Viber export cannot establish whether a profile has an avatar, what it contains, age, gender or image type. Nickname and active days are not substitutes. For creative segmentation, use voluntary choice, first-party orders or aggregate content experiments.
“High-net-worth user” is not a platform field
Active days and offline time describe platform activity, not income, assets, spending power or intent. High-net-worth classification is a sensitive, high-risk inference and cannot come from Viber number checking. B2B priority should derive from real contracts, needs and customer relationship.
Reconciling TXT and Excel
Generate a one-phone-per-line TXT and hash separately for each task, preserving internal row_key. Accept each returned Excel for schema, types, unknowns, errors and reconciliation before combining observations. Do not manually create a wide table that discards source time.
Activity output needs expiry
Store checked_at and refresh_after for user offline time and active days. Stop ranking use after expiry while retaining necessary audit history. Registration and activity can have different lifetimes and cannot share one permanent valid flag.
A rule for choosing composition
Run Registration alone for platform state. Run Activity for an established cohort that needs activity observation. Use serial execution to reduce stage two, and a same-source parallel sample for coverage research. None of these paths creates avatar or wealth fields.
