«Blast done, 500 पर "Delivered" - antispam pass» - 3 घंटे बाद number dead। Send और ban के बीच gap experienced operators को भी confuse करता है: software delivery झूठ नहीं बोलता, पर delivery final check नहीं। Meta exact ban mechanics publish नहीं करता; forums की «inner workings» observation reconstruction है, real algorithm leak नहीं।
Article के बाद: confirmed antispam signals vs «trust»/«scoring» hypotheses; successful delivery hours बाद fate नहीं बताती।
Framing: exact «dying» mechanics - reports कैसे जमा, reputation कैसे घटे, block कब decide।
Practice: Meta internal trust formula, factor weights, exact ban algorithm publish नहीं। Forums का «Trust Score», «sliding complaint window», «behavioral scoring» = market reconstruction - official docs नहीं। Honest article: Meta-confirmed vs plausible unverified hypothesis अलग।
User reports/blocks - main confirmed trigger। «Report»/«Block» पर Meta को last 5 messages का contextual log - documented, antispam filter का हिस्सा।
Content main factor नहीं। Neutral text भी - mass «Block» → automation reacts। Stop-words sole trigger नहीं।
Filled profile protection नहीं। Avatar, site, Business description superficial check - negative audience reaction offset नहीं।
WABA transparent। Official API: complaints >0.1–0.2% deliveries critical → template «Low» - Meta documented, WABA thresholds।
500 numbers blast। Software OK, all «Delivered», send के दौरान no block। 2h बाद recipients opened → report cascade। Ban ~3h after blast ended।
Dangerous myth destroyed: software delivery = antispam stage 1, not final। Main wave = read + react time, not send time।
«Trust Score». Account age, in/out ratio, IP, client type - plausible, Meta term/formula publish नहीं।
Complaint thresholds gray numbers. Fresh unwarmed: 3–5 reports/10–15 min critical; old warmed: 20–40/day spread - practice estimate, not official limit।
Preventive ban share 75–85% - disputed, weak source।
Reply <300–500 ms = no human read → lower reputation - observation, not proven।
Verbal «Stop»/«Enough» without report button - maybe counted, no proof।
Device ban list debate - fingerprint/IMEI blacklist vs bad SIM + cache/IP reset। Open - infrastructure risks।
150 msg/day month, warm base। 1–2 reports/day। Day 31 standard msg to regular client → instant block।
Cumulative effect: no day looked risky; steady small negativity crossed invisible line। Mechanism = hypothesis; pattern fits many cases।
| Gray | WABA | |
|---|---|---|
| Complaint overrun | Permanent ban | Flagged/Restricted |
| Threshold | Unpublished | 0.1–0.2% documented |
| Recovery | Usually none | Quality rating recovery |
Gray → «Number blocked»। WABA → intermediate correctable states - structural difference।
Delivery without ban ≠ antispam passed। 100–200/day ≠ official safe limit। Warmup ≠ full immunity। Randomization ≠ identical text only - not audience consent।
Risk: delivery only, ignore post-open reaction। Delivery report today ≠ fate hours/weeks later।
Key metric: complaint/block share - only officially confirmed signal, gray + WABA (different consequences)।
Regular blasting → track refusal replies, complaint growth first hours - closer to real risk than delivery alone।
Practical rule:
Software shows message went out, not accepted - number owner pays, not script developer।
