The network for creativity
Join 1.25M professional creatives like you
Connect with clients, get discovered, and run your business 100% commission-free
Creatives on Contra have earned over $150M and we are just getting started
Back to feedPost
A clean CSV should come with a record of what changed.
For recurring exports, five checks make the next batch easier to trust:
Keep IDs as text. 0001 and 1 may identify different records.
Agree the date format before converting it. Flag ambiguous dates rather than guessing.
Define a duplicate. Exact complete-row equality is different from deciding two rows describe the same customer.
Keep blank or incomplete rows for review instead of silently dropping them.
Re-run the same rules. The cleaned CSV should remain unchanged; the second audit should describe the second run.
I built a small Python demonstration with invented inventory data: 7 input records, 6 output records, 1 exact duplicate removed, 6 field changes and 15 review flags. Its 16 synthetic tests pass. This is demonstration evidence, not client or production results.
The source, before/after CSVs, audit and tests are public:
https://gitlab.com/shivanshjain456/csv-cleanup-evidence-lab
Formula-like cells are flagged, not made safe for spreadsheet execution. Import every column as text. Treat the audit as carefully as the source, since it can contain original values.
For a scoped cleanup, start with a redacted, non-sensitive 10–20-row sample:
https://contra.com/s/eOFrEcq2-csv-cleanup-with-a-change-log
gitlab.com
Shivansh / CSV Cleanup Evidence Lab · GitLab
GitLab.com
The network for creativity
Join 1.25M professional creatives like you
Connect with clients, get discovered, and run your business 100% commission-free
Creatives on Contra have earned over $150M and we are just getting started
Related posts
Same data. Two users. Completely different priorities.
https://contra.com/p/s2mFPUPZ-alvio-healthcare-management-web-and-mobile-app
I'm testing a $79 report for Shopify merchants who want to see exactly what changed between their original product CSV and a proposed price-update file. It separates intended price changes from changes to other fields, so you can review both before deciding what to do next.
Here's a simple synthetic example: a mug's price changes from $20 to $22 as intended, but the proposed CSV also blanks its Vendor and changes its Status from active to draft. The report lists those differences separately from the price change. This is an invented demonstration, not client work or a customer result.
The proposed $79 report covers up to 1,000 matching products, with one row and one default variant per product, using supported columns only. It uses deterministic comparison with AI-assisted explanations.
This is a read-only comparison. It does not access your store, execute an import, correct files or certify that an import is safe.
Have both CSV files ready for an upcoming price update? Message me on Contra with your product count and intended change, and let me know whether you would request the report at $79 if scope, availability and terms can be confirmed. Start with a description only; don't send confidential files or store access. An inquiry does not place an order or request payment.
Catching accidental changes outside the price column is the part a plain spreadsheet compare never shows clearly. How are you testing the $79 price, DMs to merchants or a landing page?
"Engineered an automated data-cleansing framework to audit, normalize, and restructure high-error transactional records. By resolving systemic data anomalies—including string irregularities, duplicate entity entries, schema mismatches, and null variances—this pipeline elevated raw, unreliable logs into clean, decision-ready assets primed for business intelligence modeling."
