Document delivery reconciliation: fictional batch B-01 Guide: https://yenra.com/content-analysis/ Updated October 4, 2026. All records are invented. These static CSVs do not send messages or access a service. Open as UTF-8 comma-separated files. Import IDs and UTC ISO timestamps as text to preserve their exact values. FILES AND FIELDS expected-documents.csv: one row per intended document-recipient-version. batch_id: delivery population identifier. document_id: stable document identity. version: required revision (integer; all sample values are 1). recipient_ref: opaque recipient reference, not an email address. acknowledgment_required: 1 means required by the sample cutoff; 0 means no. available-files.csv: one row per available file, not per expected document. file_id: unique physical-file identifier. batch_id, document_id, version, recipient_ref: expected-document join key. file_path: fictional file location; files themselves are not supplied. delivery-events.csv: one row per distinct event. event_id: unique event identity used to remove duplicate log records. batch_id, document_id, version, recipient_ref: expected-document join key. event_type: dispatched, delivered, bounced, accessed or acknowledged. event_time_utc: ISO 8601 timestamp ending in Z (UTC). evidence_ref: fictional pointer to the event evidence, not a real URL. ASSUMPTIONS AND COUNTING Observation cutoff: 2026-10-04T12:00:00Z, inclusive. All six expected keys require acknowledgment by that cutoff for this exercise. In operational use, set the rule and deadline with the service owner. All sample events precede it. Full join key: batch_id + document_id + version + recipient_ref. Expected keys must be unique. Match file rows on the full key. Count 0 as missing, 1 as exactly one file and more than 1 as duplicate-file exceptions. Deduplicate events on event_id, then count distinct expected keys per type. Do not deduplicate genuine retry events that have different event IDs. Flag unknown keys, wrong versions and events after cutoff separately. This sample contains none of those three extra exception types. EVENT MEANINGS Dispatched: service accepted or attempted a send operation. Delivered: receiving mail server accepted the message in this email example. Bounced: a failure was reported. Accessed: a request/tracking event exists; human reading remains unproven. Acknowledged: the sample's designated acknowledgment action was recorded. An operational system must specify identity binding and the exact action. Created is explained in the guide, but is not an event type in this sample. Missing events can reflect incomplete logs or ingestion delay: investigate. INDEPENDENTLY CHECKED EXPECTED RESULTS 6 unique expected keys; 6 file rows; 5 keys with at least one matching file. Availability = 5 / 6 = 83.333...%, shown as 83.3%. 4 keys with exactly one file. Missing: D-05. Duplicate-file key: D-04 (2 files). 12 event rows with 12 unique event IDs. Dispatched: D-01,D-02,D-03,D-06 (4). Delivered: D-01,D-03,D-06 (3). Bounced: D-02 (1). Accessed: D-01,D-06 (2). Acknowledged: D-01 (1). Missing required acknowledgment: D-02,D-03,D-04,D-05,D-06 (5). Event measures overlap; do not sum them as exclusive outcomes. Duplicate-file resolution is required before releasing D-04. Reconcile the missing D-05, delivery failure D-02 and acknowledgment gaps separately. EXCEPTION WORKSHEET (copy for each exception) Run ID / cutoff / log-completeness evidence / service owner: Exception ID / full expected key / type / expected result / observed result: Source evidence reference / data completeness or timing uncertainty: Action / owner / due date / corrected-run ID / retest / closure evidence: ADAPTATION Freeze your authorized manifest, keep original observations and record later corrections in a new run. Protect recipient mappings and follow the applicable records policy. Technical logs alone do not establish legal proof of receipt. Event semantics reference: https://docs.aws.amazon.com/ses/latest/dg/monitor-sending-activity.html Open tracking reference: https://docs.aws.amazon.com/ses/latest/dg/faqs-metrics.html