How to analyze customer reviews csv
This page is for teams who already have reviews in a spreadsheet: a survey export, a helpdesk dump, a Shopify product-reviews app, a Typeform file, or a sheet an intern updates on Fridays. You do not need a special template from us. You need a header row, a rating column, and a review text column.
There is no Chrome extension for "CSV." If your reviews live on Google Maps, Amazon, or Trustpilot, use the exporters we actually built, then come back with that file. For every other source, use the platform's own export. Survey tools usually have CSV in the results menu. Shopify review apps have CSV in their settings. Zendesk and Intercom exports can work if you include a score and the comment body. If the platform has no export, copy rating and text into Excel by hand for a sample, then decide whether a proper export is worth setting up.
Prepare the file before you upload. Put headers in row 1. Do not merge cells. Save as CSV (UTF-8) or .xlsx. Keep star ratings as numbers (1 to 5, or 1 to 10 if that is your scale). Put the full comment in one column, not split across cells. Remove total rows and charts that sit under the data. If a comment contains commas, use a real CSV from Excel or Google Sheets rather than concatenating text in Notepad.
Required columns: rating and review text. Optional columns we can map: title, date, author, verified flag, and product variant. Dates should be real dates, not "two days ago." If your tool only has NPS (0 to 10), you can still map that to rating. Themes will still cluster, though the star distribution chart will follow your scale, not a classic 5-star storefront.
Insights depend on the business behind the CSV. Support tickets tagged as "reviews" often hide refund and wait-time language. Product surveys hide packaging and instructions. Employee-collected Google-sheet reviews hide location. Start with low scores, read the repeated phrases, then search the review list for those phrases to see the raw quotes. If two products share a file, keep a variant or product column so you do not "fix" the wrong SKU.
Privacy: this analyzer never uploads the CSV. That matters when comments include names, order IDs, or emails. Still, strip personal data you do not need before you share a screenshot of the dashboard. The export summary you can download from the tool is a theme table, not a replacement for the original file.
If you are not sure the mapping is right, use Adjust columns after a run, or start with one of the sample files on the main analyzer to see the shape of a clean result. Then upload your real sheet.
Columns to include
- Rating / score (required)
- Review or comment text (required)
- Title
- Date
- Author
- Verified flag
- Product, location, or variant