The work elsewhere
What the source systems say we should have
20 comparisons have been set up between a count this platform can produce and the same count taken from the system that owns the records. 7 agree to the record, 7 differ for a reason somebody wrote down, 4 differ and nobody has explained why, and 2 have never run at all. A check that has never run is not agreement.
Every comparison, worst first
Platform count against the source system’s own total for the same measure over the same period. The residual is what is left after the part somebody can explain is taken out.
Systems with no comparison at all
These are not systems that agree. They are systems nobody has checked, and the coverage statement treats them accordingly.
How often each one runs
Cadence is what was agreed with the system owner, not what a scheduler enforces. Nothing on this page runs itself.
Actions
What is waiting on a person
A kernel that never checks itself against the world drifts and nobody notices until an auditor does. These are the checks that did not come back clean.
differences explained and left in place
An explained difference is not a reconciled one. The extract and the source count different things on purpose, the reason is written down, and the numbers still do not match. Hiding that would be worse than showing it.
source systems with no drift check at all
Reporting Semantic Layer · Marketing Automation · Digital Analytics · Product Lifecycle Management · Workforce Management · Planning and Consolidation. No comparison has been defined against these, so the platform has no way of knowing whether its counts match theirs.
checks where the difference has no explanation
25,250 records are unaccounted for across them, the largest being Configuration Management Database at 13,200 items. Nobody is claiming the platform is wrong or the source is wrong. The difference is stated and left open.
checks that have never run
Bank Connectivity Gateway · Research Notebook and Experiment Data. Both sides read zero because neither side has been read. This is reported as never run, not as agreement, and the distinction matters more here than anywhere else on the page.
Operations
What this desk is allowed to start
A surface that only reports is not operable. This is the work this page can set in motion, and the bound it runs into.
Trigger and bound
This surface compares a platform count against a source system’s own total on a stated cadence and publishes the difference. It cannot correct either side. Where a difference is unexplained it stays unexplained until the system owner accounts for it.
Live observability
What the record shows right now
All 20 drift checks by outcome, 4 states in use, covering 20 of the 26 source systems in the estate.
Current distribution
20 checks
Is policy and strategy coming to fruition
Whether the written intent is holding here
Yes on the checking. No on the closing.
Not holding on the record
The intent was that the platform compares itself against the outside and states the difference. 20 checks now run against 20 systems and every difference is published, including the 4 nobody can explain. What is not yet true is that the checks are complete: 6 systems have no check defined, and 2 of the checks that do exist have never run because the system on the other side is not connected.