Trust begins before the chart
A dashboard can be visually clear and still be wrong, incomplete or differently defined from another report. The reliability of a result depends on the source population, mapping, validation, duplicate handling, transformation rules, calculation logic and reconciliation that happened before the number appeared.
A governed data path
Source file or system is identified.
File/record structure is profiled.
Fields are mapped to governed definitions.
Required data-quality rules are applied.
Duplicate or repeated-processing risks are controlled.
Records are transformed into the analytical model.
Counts and financial totals are reconciled.
Business rules calculate measures and classifications.
Results preserve lineage to contributing records.
What reconciliation proves
Reconciliation checks whether records and values moving through a process can be accounted for. It can compare source row counts to accepted, rejected or held records, and financial totals to downstream totals. A passing reconciliation does not prove every field is correct, but it provides evidence that the defined population moved through the process without unexplained loss or duplication.
Why two reports can show different numbers
- Different date fields are used.
- One report includes only final claims while another includes all submitted claims.
- One uses claim count and another uses line count.
- Different facilities or payer groups are included.
- Different definitions of denial, payment or A/R are applied.
- One report is refreshed later.
- One source includes adjustments or reversals that another excludes.
What lineage means in plain English
Lineage is the ability to move backward from a reported result to the records, source and rules that produced it. If an executive sees $2.4 million of A/R over 90 days, good lineage makes it possible to identify the underlying balances, their source records and the calculation that placed them in that total.
Evidence makes a result reviewable
- Source identifier and import/run history.
- Validation results.
- Mapping version.
- Rule or calculation version.
- Reconciliation result.
- Affected record identifiers.
- User actions and approvals where relevant.
- Documents or payer evidence supporting final disposition.
A control question for every important KPI
Ask: Can we define this number, reproduce it for the same source population, explain why it changed, and trace material components back to evidence? If not, the organization may have reporting but not full auditability.