1. The operating brief
Customer, lead, vendor, product, order, or project information is spread across forms, inboxes, spreadsheets, exports, and an underused CRM. Staff cannot trust the records, but an unrestricted cleanup could overwrite valid history or create new duplicates.
MY-VA can provide a data entry virtual assistant or managed data queue. The work begins with a source map and data dictionary, then moves through controlled batches with validation, exception review, and a record of each approved change.
This page describes a controlled data workflow. It does not promise perfect source data, automatic deduplication, or a specific completion rate without reviewing the records and systems.
2. Define every destination field and source of truth
The data dictionary identifies each destination field, required format, allowed value, source priority, validation rule, and owner. It distinguishes blank, unknown, not applicable, and intentionally withheld values so missing information is not disguised as complete.
When several systems contain the same field, the client names the source of truth. The assistant does not decide whether an invoice platform, CRM, support desk, or spreadsheet should win when the business has not established that rule.
- Source files, systems, exports, and record owners.
- Field mapping and accepted formats.
- Required values and validation checks.
- Unique identifiers and duplicate criteria.
- Fields that require client approval before change.
3. Limit access and preserve a recoverable trail
The assistant receives only the access needed for the assigned records and actions. Where possible, work starts in a staging file, import sandbox, or restricted view before production records are changed. Sensitive fields stay excluded unless they are essential to the approved process.
Each batch keeps the original value, proposed value, source, action, date, and exception note. Import files are retained according to the client policy, and destructive merges or deletions remain with an authorized owner unless expressly approved and recoverable.
4. Standardize, deduplicate, and route exceptions
Routine actions can include formatting names and addresses, mapping categories, updating contact details from approved sources, attaching source references, completing required fields, and preparing import files. Duplicate review compares approved identifiers instead of relying only on similar names.
Ambiguous records enter an exception queue with a reason. Examples include conflicting account owners, incomplete addresses, several possible matches, invalid dates, missing consent fields, closed accounts, or values that fail system validation. The assistant does not force an uncertain record through the normal path.
5. Check the work before and after import
Quality review uses approved samples and high risk fields. Checks can compare source and destination values, confirm field formats, test duplicate rules, count rejected imports, and verify that required relationships remain intact.
After import, totals and spot checks confirm that the expected records changed, exceptions remained isolated, and no field shifted into the wrong column. Error patterns are converted into clearer rules or focused coaching before the next batch.
6. Move from cleanup project to record maintenance
Once the backlog is controlled, the same role can own a daily or weekly intake queue for approved updates. New forms, returned emails, sales notes, support outcomes, and vendor changes follow the same field rules and exception path instead of recreating the backlog.
Reporting can show records received, records updated, duplicates identified, exceptions opened, exceptions resolved, import failures, and reviewed errors. The client sees both throughput and the issues that still need a business decision.
A record can be traced to its source, every material change follows an approved rule, and uncertain data reaches a named owner instead of being guessed.