A source-system change should preserve the definitions and evidence needed to compare historical reporting. Keep original exports, document the transformation and validate an overlapping scope before treating the new view as continuous with the old one.
The new system may use different property codes, account hierarchies, date fields or units. A familiar looking LOS can hide those differences if the transition only matches visible labels.
Preserve the old record
Retain the source exports and the reporting definitions used before the change. Record which periods and properties they cover. Keep the version needed to explain previously accepted reviews.
Continuity is harder to reconstruct after files, mapping decisions and knowledgeable reviewers have moved on.
Build the translation explicitly
Map property identities, account categories, date rules and measure units. Identify transformations that change aggregation or signs. Explain unresolved items separately from accepted mappings.
Do not assume the same account name represents the same activity in both systems. Review representative records where the definition is material.
Validate an overlapping slice
Choose periods and properties that can be reviewed in both structures. Reconcile totals and test category level examples. Identify differences in timing, population or source coverage before interpreting the result as an operating movement.
For a fictional field, an accepted transition might show matching expense totals but a different category split. That is not yet a fully comparable category trend. The team needs to review the mapping and document the decision.
Make the boundary visible
Identify the transition point, accepted scope and known limitations in the recurring review. Where historical comparisons have been normalized, preserve the route back to the original source.
Use guided activation to connect the old record with the first useful new question. Upstream’s data continuity approach should be assessed against that concrete workflow rather than a promise that every source will behave identically.
The outcome is a reporting record the team can explain across the transition, with changes in operations separated from changes in reporting.
Research and references
These sources offer background for the review question. The worked examples and review methods above are Upstream’s practical guidance.
- A Survey of Data Quality Measurement and Monitoring ToolsEhrlinger & Wöß · Frontiers in Big Data · 2022 · Peer-reviewed systematic review · Open access
Further reading on profiling, quality measurement and monitoring.
- Beyond Accuracy: What Data Quality Means to Data ConsumersWang & Strong · Journal of Management Information Systems · 1996 · Peer-reviewed research
Research perspective on evaluating data quality beyond accuracy.
Common questions
Should the old source exports be retained?
Yes. They support explanations of prior reviews and validation of later mapping or adjustment decisions.
What does an overlapping acceptance period test?
Whether both structures represent the same accepted population, measures and periods, including category level differences.
Put the question to work
Explore the connected workflow for data continuity or bring your well data question to a working session.