Free CSV tools for data handover
The free CSV workspace helps developers, freelancers and everyday spreadsheet users clean exports, compare records and produce CSV, JSON or JSONL. It reads CSV values as strings, which keeps identifiers such as 001 intact during the browser workflow. No account or payment is required, and data is not uploaded to a conversion server. Good results still depend on understanding the file's headers and keys.
For freelancers: define the cleaning rule first
Cleaning can trim cell whitespace and remove duplicates. With no matching key header, duplicate removal compares whole rows. With a matching key, the first row for that key is retained. That distinction matters when two product records share an ID but disagree about price. The tool cannot decide which business record is authoritative.
For example, 001,Notebook,12 and 001,Notebook,14 are different complete rows. Whole-row deduplication keeps both. Key-based deduplication using sku keeps the first, even though the later price differs. Agree on the intended rule before exporting a client's dataset. Keep the original so you can explain which records were removed and why.
For developers: compare exports by a stable key
Place the older CSV in the first input and the newer file in the second. Enter the exact header name, such as sku, and choose Compare by key. The header must exist in both files. Blank or repeated comparison keys cause an error rather than silently overwriting entries. Header spelling, capitalization and surrounding spaces matter.
Suppose the older data contains 001 at price 12 and 002 at price 2. The new data contains 001 at price 14 and a new product 003. Comparison reports a changed price for 001, an added 003 and a removed 002. The report uses status, key, column, before and after fields. It also considers columns found in only one input.
Compare parses the inputs as supplied; it does not automatically apply the cleaning options first. If you want to compare cleaned data, clean and download it, then use that result as a comparison input. There is no column-mapping interface or automatic matching by similar names in this release.
For everyday users: choose the right export
A preview shows the first 100 rows. The download contains the full processed result, not just that preview. CSV is suitable when the receiving application expects a table. JSON maps headers to property names, while JSONL writes one object per line. CSV values stay text in these exports instead of being guessed as dates, numbers or booleans.
The JSON-to-table option expects a non-empty array of objects in the first input. It uses the union of keys; missing properties become empty cells and nested values become JSON strings. Decimal values follow JavaScript precision, and unsafe parsed integer values are rejected. Large identifiers should be quoted strings in JSON.
Understand spreadsheet formula protection
Spreadsheet-safe CSV export is enabled by default. Formula-like cells receive an apostrophe prefix to reduce execution risks in spreadsheet software. This intentionally changes those CSV cell values. It is not applied to JSON or JSONL exports. Disable the protection only for trusted data when you understand what the receiving application will do.
Limits and useful feedback
Each header must be unique and non-empty, and rows must have consistent column counts. Quoted delimiters and multiline fields need proper CSV syntax. The trim option affects values, not headers. Large files are held in memory and can slow the tab; there is no streaming or background-worker workflow.
If you find an issue, select CSV workspace on Contact. State whether parsing, cleaning, comparing or exporting failed. A small invented example with headers and the expected output is more useful than sending a private customer list. The linked guide provides the full sequence and explains why an input may be rejected.
Free table work across many records
The CSV workspace applies an operation across the supplied table, which is useful for a larger export as well as a small example. It is not a folder watcher or a background pipeline. Before a full cleanup, test a few rows that include spaces, leading-zero IDs, a repeated key and a quoted comma. Decide whether duplicate means an identical row or the same business key. A freelancer should ask the client which record should survive when values disagree, because keeping the first row is a rule, not a determination of truth.
For a concrete handover, use a small table with sku and price headers. Preserve 001 as an identifier rather than assuming it means the number 1. After cleaning or comparing, download the result and reopen it in the receiving application. Some spreadsheet applications infer types when opening a CSV, so local string preservation does not control every later import. Use the application's text-column import settings when leading zeros matter.
Developer and everyday-user checks
A developer can use an invented export to understand added, removed and changed records before applying the workflow to authorized data. A spreadsheet user can trim values and review the preview, remembering that it shows only the first 100 rows. Check the full downloaded file, confirm the intended headers and inspect a few records beyond the preview. No payment or account is required for these operations. Device memory and input validity still determine whether a job can finish, so break very large work into appropriate source-side exports rather than assuming an unlimited bulk import.
