Discord Delivery Coverage Worksheet
Reconcile unique originals, received items, late arrivals and cancellations. Duplicate events stay outside the denominator.
Formula and Default Example
Received coverage is unique received divided by original instructions. Usable coverage subtracts disjoint late and cancelled subsets first. The fictional defaults yield 90% received coverage and 75% usable coverage. They are not measurements of a real room.
Classification Comes First
If one original is both late and cancelled, apply a predeclared priority so it enters only one subtraction bucket. Preserve both event notes in the raw journal. A duplicate event does not increase received originals or published originals. The worksheet cannot verify your classifications.
What the Output Cannot Say
A complete count still says nothing about win rate, signal quality, member-wide delivery or executable prices. The JSON export records your entered scenario only. The form runs locally in the browser with no network submission.
Primary Sources
The new context boundary
The missing-context input concerns only unique received originals already counted as timely and not cancelled. It must not exceed that subset. Keep a reason, original ID and required context object in the observation journal. The original delivery and usable-coverage fields remain available; the additional context-complete output is a stricter information finding. These input rules do not implement Discord effective permissions or discover missing messages automatically. The user supplies independently reconciled counts, and the JSON export remains a scenario rather than observed provider performance.
Completed fictional packet
The new filled packet enters eight published originals, six unique received, one late, one cancelled, two duplicates and two missing-context items. It gives two missing originals, four timely uncancelled and two context complete: 75%, 50% and 25% respectively. No win rate follows. The original twenty/eighteen/two/one/three default gives its earlier 90% and 75% results when missing context is zero. CSV and JSON retain the invented classifications, so neither scenario becomes a vendor claim simply because its arithmetic is exact.
A zero population remains invalid
Coverage requires a positive original population under a declared interval and identity rule. Zero published items cannot produce a meaningful delivery fraction. Negative, fractional and unsafe integer counts are also invalid, as are subsets exceeding their parents. A rejected scenario must not export a stale earlier result. Reset and reload restore illustrative defaults without connecting an account or storing a private journal. A real review additionally needs consented source evidence and actual receiving clocks; the worksheet cannot supply those from numbers alone.
A completed fictional room packet
Invented observer R, originals O1-O8 and two repeated O1 events. No real Discord server, member identity, paid subscription, price, broker fill or financial return.

| Object | Record | Bounded finding |
|---|---|---|
| O1 | Required original and updates present | Received true; context complete true |
| O2 | Required original and updates present | Received true; context complete true |
| O3 | Linked parent unavailable | Received true; context complete false |
| O4 | Required attachment unavailable | Received true; context complete false |
| O5 | Received after declared validity | Received true; context complete false |
| O6 | Linked cancellation retained | Received true; context complete false |
| O7 | No receipt in invented window | Received false; context complete false |
| O8 | No receipt in invented window | Received false; context complete false |