Skip to content

Weekly Checks

Applies to: All subscriptions

Purpose

Catch what a daily check cannot: trends, drift, and patterns across many documents. The daily routine asks "is anything stuck?"; the weekly routine asks "is anything getting worse?"

Audience

The operations owner, with the IT team for the infrastructure items.

Prerequisites

  • Daily Checks being performed
  • Access to monitoring history covering at least four weeks

Steps

1. Rejection patterns
Verify Validation and government rejections for the week, grouped by cause
Expect A low rate, with no single cause dominating
Warning signs One field or one code accounting for a large share; a rate rising week on week; a new cause appearing
Act Find the systemic cause. A dominant single cause is nearly always a mapping rule, a master data fault, or an ERP change — not a series of user errors

This is the highest-value check in the weekly routine. A pattern visible across a week is invisible in any single day.
Verify Documents processed per day, per registration, per ERP
Expect The normal weekly shape
Warning signs A sustained fall from one source; growth outrunning capacity planning
Act A fall means an integration problem or a business change — establish which. Growth means revisiting capacity before it becomes urgent

3. Capacity
Verify CPU, memory, disk free, database growth against forecast
Expect Within the healthy ranges
Warning signs Any metric trending towards its threshold; growth exceeding forecast by more than a quarter
Act Plan capacity now rather than under pressure. Re-forecast if growth diverges from the plan

4. Backups
Verify Every backup in the past week ran and verified
Expect All successful and verified
Warning signs Any failure; any unverified backup; growth in backup duration
Act Investigate immediately. A week of failed backups discovered at restore time is the worst outcome in this section

5. Expiring credentials and certificates
Verify Days remaining on TLS certificates and provider credentials
Expect More than 30 days on everything
Warning signs Anything under 30 days; anything with no named owner
Act Start renewal. Provider credentials involve a third party and can take weeks — 30 days is a floor, not a comfort

6. Job and integration health
Verify Job completion times, queue behaviour, write-back delivery statistics
Expect Stable week on week
Warning signs Completion times rising at constant volume; recurring delivery failures; dead-lettered events
Act Rising completion times at constant volume usually mean database maintenance is overdue. Recurring delivery failures mean an endpoint problem that will eventually open the circuit

7. Provider call volume
Verify Calls made against the contracted allowance
Expect Tracking to plan
Warning signs Running ahead of plan; a rise not matched by document volume
Act A rise without matching volume means something is polling a read that could be served locally. Review the calling pattern

8. Open items and alert quality
Verify Items carried forward from daily checks; alerts that fired without action
Expect Few, each with an owner and a date
Warning signs A growing carry-forward list; an alert firing repeatedly with no action
Act Close, reassign, or explicitly accept. An alert firing weekly with no action should be fixed or deleted

Validation

Check Pass condition
The routine was performed Logged with date and owner
Trends compared against history Not just this week in isolation
Systemic causes identified Any dominant rejection cause has an owner and a plan
Backups verified Not just "ran" — verified
Nothing expires within 30 days Certificates and credentials both
Carry-forward list is short Not growing week on week

Troubleshooting

Symptom Cause Action
The same rejection cause every week Systemic, treated as instances Fix the mapping or master data
Volume from one ERP falling gradually A partial integration failure, or a business change Establish which before assuming either
Job times rising at constant volume Database maintenance overdue Run index and statistics maintenance before considering capacity
Backups failing silently No verification step Verify every backup; a restore test quarterly
A certificate expired despite weekly checks The check was performed, nobody owned the renewal Assign an owner, not just a check
Provider volume above plan A polling read Review the calling pattern
Carry-forward list growing Findings recorded without owners Assign an owner and date to every finding
The weekly routine is skipped when busy It competes with daily work Schedule it as a fixed appointment; it is the routine that prevents the crises