Article summary
Check whether name script, transliteration and acquisition source distort the usable demographic sample in a Chad +235 list.
When a Chad +235 list comes from different scripts, partners and historical sheets, profile-usable rows may naturally favour one source. Age and gender percentages can then describe the pipeline rather than the customers.
Build a source-by-script cross-table
| Source | Original script kept | Transliterated | Profile usable | Unknown |
|---|---|---|---|---|
| Website | Count | Count | Count | Count |
| Partner | Count | Count | Count | Count |
| Legacy CRM | Count | Count | Count | Count |
Never generate a profile from name script
Neither the original nor transliterated name may populate age or gender. Spelling changes amplify errors and disguise cultural assumptions as data.
Preserve raw phones and source labels
Do not overwrite phone_raw, source_label or country_evidence. Normalisation creates candidate values while retaining the reason for change.
+235 carries no population meaning
The international code supports numbering plans but proves no region, language, community or current holder. Cross-border use and reassignment need independent evidence.
Report unknown by stratum
If one script or source has a materially higher unknown rate, disclose the bias and limit use. Do not delete those rows and then call the remainder representative.
Keep names out of TXT
AppShai accepts one phone per line. Name script and customer details remain in the private crosswalk; row_key reconnects the returned Excel workbook.
Blind the quality sample
A reviewer can inspect states and reconnection with names and avatars hidden, reducing the influence of script and appearance.
Ask separately for language
Multilingual communication follows user choice and confirmed conversation, not script, age or gender inference.
The Chad quality line
Profile analysis becomes honest only when every source and script group can see its missing rate. Otherwise repair the pipeline before interpreting the audience.
