Compare two CSV exports by one agreed key and get a small, auditable difference report. You can also request a report of changes in selected columns, such as catalogue prices or stock.
Pilot enquiries: indicative A$40 total for up to 1,000 combined data rows. The final quote confirms scope, availability, turnaround and payment method before an order is accepted. One correction within the agreed comparison rules is included. This is an enquiry offer.
Request a quote by email. Start with the two row counts, column headers, key, and any value columns you want compared. Please wait for agreement before sending files.
- Rows with a unique key found only in the first file.
- Rows with a unique key found only in the second file.
- A separate review file for blank keys and every row affected by a duplicated key.
- A shared-key list, row counts, source hashes and output hashes.
- When agreed, a changed-cell report for selected columns on unique shared keys.
The default key comparison is exact and case-sensitive. Trimming or Unicode case-folding can be agreed explicitly. A shared key does not prove two records describe the same entity. Non-key values are compared only when the columns are agreed in advance, as exact text: 20.00 and 20.0 differ. No numeric, date or currency interpretation is included.
The report is emailed as an inline JSON bundle containing the CSV text and file checksums. The free offline report unpacker saves its files together as a ZIP:
- Download
report-unpacker.htmland open the saved file in Chrome. - Paste the plain-text report message, or open a saved
.txtor.jsoncopy, then select Check bundle. - Select Download ZIP after every file passes its length and checksum checks.
The single HTML file runs locally without uploading your report. Load synthetic example lets you try it before receiving a report. File contents, UTF-8 byte-order marks and line endings are preserved. The helper accepts the five standard report files and the optional changed-cell report, up to 10 MiB per message.
The unpacker was exercised in desktop Chrome on macOS, including a 390-pixel viewport. Other browsers, phone use and other email providers have not been verified. The quote still confirms that the actual delivery format and report size work for you before an order is agreed. Verification and source references.
Use synthetic or non-sensitive catalogue-style CSV data. This pilot is unsuitable for confidential customer records, credentials, regulated data or high-stakes decisions. Files and results pass through email and an AI-assisted processing workflow. Keep your original files.
The pilot covers two UTF-8 CSV files with headers, one key column, at most 1,000 data rows combined and 200 KB combined. Live account access, scraping, enrichment, fuzzy identity matching and automatic changes to a source system are outside the scope.
The example inputs contain fictional names and .example.test email addresses. They were processed by the existing Local Row Check 1.0.0 command-line tool. The generated reports are actual tool output, not a customer result or a claimed sale.
| Source | Input rows | Only in that source | Shared unique key | Needs review |
|---|---|---|---|---|
| Left | 5 | 1 | 1 | 3 |
| Right | 3 | 1 | 1 | 1 |
For each source: input = only + shared + review. Every duplicated-key group goes to review, including the matching row from the other source.
For the optional value report, inspect the Local Row Check 1.1.0 catalogue example.
It has three unique shared keys, two with changes across three selected cells. A title change is ignored because only Price and Stock were selected. Duplicates still go to review. These are synthetic records and actual tool output.
See the method and output guide before requesting a quote. The sample was executed and its hashes and row counts checked. Processing uses AI-assisted tools.