BTA / Delivery desk
Missing Alerts and the Delivery Denominator
You cannot calculate a success rate from successful receipts alone.
Establish what should have existed
A receiver log can count observed messages. It cannot by itself establish the sender's complete attempted stream. A defensible delivery denominator needs source-side event identity or another reliable expected set. Preserve the observation window, inclusion rules and original expected IDs. When those artifacts are unavailable, report observed receipt counts without pretending they measure coverage.
Distinguish a known missing message from an unknown source interval. A source event with no matching receipt is a reconciliation gap. An unavailable source log means the expected count may itself be unknown. These conditions need different labels; merging them into a success percentage hides the actual evidentiary limitation.
Retain exclusions and repeated attempts
A retry is another attempt attached to an original, not automatically another expected instruction. A cancelled instruction can still have generated a legitimate transport event. Count the object defined by the question: original instructions, attempts or screen displays. Publish that definition with any totals rather than changing it when the number looks better.
Consecutive windows help reduce convenient selection, but they do not guarantee a representative market sample. Preserve outages, permission changes and missing log intervals. Do not silently exclude periods because the route failed. If exclusions are justified, state them before summarising and retain the excluded evidence.
A synthetic set only tests a declared set
BTA's example kit has a finite expected population so a test design is inspectable. The files contain invented messages, not observed delivery outcomes. They can teach reconciliation questions and support a safe isolated run; they cannot provide a real vendor delivery rate without actual run logs.
For a documentary product review, identify whether the selected workflow exposes source and destination evidence. A system may document a route while leaving completeness difficult to verify. Report that limitation directly. A precise percentage with no expected-set basis is less useful than a clear explanation of which count remains unknown.
Original Evidence Packet
| Artifact | Record to retain |
|---|---|
| Expected set | Source population and inclusion rule |
| Observed set | Receipt IDs and unavailable intervals |
| Count unit | Original, attempt or display; exclusions retained |
Continue the Investigation
Source Boundaries
Official references describe published mechanisms. BTA's evidence packets and proposed test procedures do not claim a completed paid-feed or account trial.