Bulk classification
Classify spreadsheet rows with AI: Excel and CSV
Short answer
To classify spreadsheet data with AI, choose the column to judge, define the categories in plain language, and run one decision per row. Each row gets a label and a confidence score in new columns. Sort by confidence to review only the uncertain rows, instead of reading thousands of survey answers, reviews or transactions by hand.
Spreadsheets are where a lot of unstructured text ends up: survey answers, exported reviews, support ticket dumps, bank transaction descriptions, lead lists. Categorizing them by hand is slow, and formulas with SEARCH() and nested IFs break on the first phrasing nobody anticipated.
Running a decision per row turns a free-text column into clean categories you can pivot, chart and filter.
Get Early Access to ClassifierHub: 2× credits in your first paid month.
The workflow
- Pick the input: one text column, or several columns combined (for example title plus review body).
- Define outputs: a category, a yes/no flag such as "mentions a competitor", or a score such as sentiment.
- Test on 20 rows and compare with your own judgment. Adjust descriptions where they disagree.
- Run the full sheet. Each output becomes a new column, plus a confidence column.
- Sort by confidence and review the bottom few percent by hand.
Without code: the Excel & CSV runner
ClassifierHub's Excel & CSV runner (opening to Early Access members first) takes an .xlsx, .xls or .csv file, lets you pick the column and the decision, and gives you the file back with result columns such as category and category_confidence. The file is parsed in your browser and only the selected column values are sent.
Doing it through the API
For files that live outside a spreadsheet tool, or for recurring jobs, submit rows to the batch endpoint. Credits are reserved up front and refunded for items that fail, and results come back in input order with your row ids.
curl https://classifierhub.com/v1/batch \
-H "Authorization: Bearer $CLASSIFIERHUB_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"config": {
"outputs": [{
"key": "category", "type": "choice",
"instructions": "Which category does this product review belong to?",
"options": {
"quality": "Build quality, defects, durability",
"shipping": "Delivery time, packaging, damage in transit",
"price": "Value for money, discounts",
"service": "Customer support experience",
"other": "Anything else"
}
}]
},
"items": [
{ "id": "row-2", "input": "Arrived two weeks late and the box was crushed." },
{ "id": "row-3", "input": "Stitching came apart after one wash." }
],
"webhook_url": "https://example.com/hooks/classifierhub"
}'
# 202 -> { "job_id": "...", "status": "queued", "total_items": 2, "results_url": "/v1/batch/.../results" }The ClassifierHub API opens to Early Access members first.
Good use cases
- Survey and NPS comments: themes, sentiment and whether the respondent asks for follow-up.
- Product reviews: quality, shipping, price or service complaints.
- Transactions: expense categories from bank descriptions.
- Lead lists: fit score against your ideal customer profile.
- Catalog cleanup: product category and whether a listing breaks your rules.
Why not ask a chatbot to do it?
Pasting a sheet into a chat assistant works for a few dozen rows. Beyond that, answers get truncated, labels drift between batches and there is no per-row confidence. A per-row decision is consistent across the whole file and tells you which rows to double-check.
Frequently asked questions
Related guides
Last updated . ClassifierHub is an independent product built on top of the Jev decision model, accessed through OpenRouter. It is not affiliated with or endorsed by TypeSafe or OpenRouter.