Moving from spreadsheets: how to check your data

Map the sources

Data rarely lives in one spreadsheet. Contacts may belong to a sales team, payments to an accounting system, contracts to a folder and statuses to message threads. Assign an owner to each source and decide what must be migrated. The entire historical archive does not necessarily belong in the new product’s active database.

Agree on field meanings

Amount might mean order value, money received or outstanding balance. Date might refer to record creation or service commencement. Document the meaning, format and units of important columns. Discuss missing values separately. An unknown phone number must not become a fabricated number merely to satisfy an import validation rule.

Set rules for merging duplicates

Matching names do not necessarily identify the same customer. Define matching criteria and cases that require human review. Merging records must preserve related orders, documents and interaction history. Prepare a small collection of difficult examples and approve the expected result before applying the rules to the entire dataset.

Run a trial migration

Import a copy into a test environment and check more than row counts. Open customer records, related documents and reports for a selected period. Reconcile amounts and statuses against the original sources. Record discrepancies separately from corrections so that you can distinguish migration changes from errors that already existed.

Plan the cutover

If staff keep editing the old spreadsheets, new data will appear between the trial and final import. Agree how to capture it: a short editing pause, importing subsequent changes or another defined process. Assign someone to approve the switch. Preserve an original copy and define the conditions for returning to the previous workflow.

Define completion

Migration is complete when responsible staff approve the agreed checks and can perform core tasks. Give them field definitions, known limitations and a way to report problems. Plan support for the first working days separately: real operations can expose details that are difficult to spot by browsing sample records in a test environment.