Article summary
Use Zambia’s 2024 introduction of the 57 mobile resource to expose stale range rules and build versioned cleanup and Full-Format workbook acceptance.
When a Zambia WhatsApp list contains phones of the correct length but a validator rejects them, the fault may be a stale rule rather than a bad number. In December 2024, Zambia’s regulator announced the introduction of the 57 mobile resource through the ITU. That change creates a practical test for every cleanup pipeline: does it consult a current plan, or does it treat a years-old prefix table as permanent truth?
A +260 national significant number has nine digits
The ZICTA plan published in ITU Operational Bulletin No. 1309 gives both the minimum and maximum national significant number length as nine digits. It also separates fixed, mobile, reserved and other resources. Nine digits after +260 is therefore an initial structural test, not proof that the value is a mobile phone suitable for a platform check.
The 57 resource exposes systematic stale-rule bias
The bulletin records 57 as a non-geographic mobile-services resource introduced on 23 December 2024 and assigned to Airtel Zambia. A validation library frozen before that event can send a concentrated block of new phones to invalid. That pattern is not random noise; it is a version failure. Repair the rule before deciding that the records should be removed.
Use rejection patterns to diagnose the pipeline
| Observed rejection | Likely cause | Investigation |
|---|---|---|
| 57-leading values fail together | Outdated range table | Compare rule date with the ITU update |
| All national-format values fail | Domestic trunk handling or missing country context | Inspect source_country and raw value |
| Apparently random digit loss | Spreadsheet conversion or truncation | Return to the original export; do not invent digits |
| Fixed and mobile values mix | Length-only validation | Route by numbering use |
Make the numbering-rule version a batch field
Every cleanup run should retain rule_source, rule_published_at, rule_version and processed_at. When a new resource appears, the team can replay only affected phones rather than the whole CRM. An old batch with no rule version cannot support a permanent “invalid phone” conclusion.
Run Full Format after numbering classification
AIPUSH WS Full Format may include phone, activity time, active days, gender, age, avatar, skin tone, avatar type, business status and mapped WhatsApp phone. Separate +260 mobile candidates, fixed phones, short codes, malformed values and other country codes first. Otherwise a null result mixes formatting failure, no returned observation and temporary invisibility into one unexplained cell.
Prefer state columns to a true-or-false verdict
Keep normalization_state, range_state, screening_state, mapping_state and review_reason beside the returned data. A phone can be consistent with a current mobile range while a profile field remains unknown. Another phone may have unresolved formatting and therefore never enter screening. These states are more useful than a single valid=true flag.
A mapped phone must not silently overwrite CRM
If the mapped WhatsApp phone differs from the submitted value, create a mapping-conflict queue and retain both values with the check time. Notation conversion, number change and other account relationships are possible. Update the primary contact only after separate customer confirmation or reliable business evidence.
Profile columns are not substitutes for business facts
Gender, age, skin tone and avatar type need an unknown state, retention period and use restriction. They cannot establish income, ethnicity, health or political belief, and they should not replace language, location and need explicitly provided by the person. They are unsuitable as the sole basis for consequential decisions.
Design TXT batches for future range updates
The TXT contains one phone per line while country_hint and contact_id remain in an internal crosswalk. Include source and date in the batch name, such as crm_legacy_2026q3. A later ZICTA or ITU change can then identify and reassess the relevant prefix cohort without uploading names, orders or other context again.
Challenge the validator with the new range
A release test should include the 57 resource, established mobile ranges, fixed phones, reserved ranges, short and long values, and foreign country codes. Success does not mean passing every sample. It means routing every class correctly and returning an explainable reason for rejection. That test demonstrates maintainability better than a few familiar numbers that happen to pass.
The Zambia lesson is specific and reusable: numbering plans evolve, and stale validators can label legitimate new resources as bad data. Source dates, use-based routing and exception queues are prerequisites for a traceable Full-Format project.
