Plain English · Verifiable answers · Runs locally

Ask your CSV a real question.

Ask for a total, ranking, trend, filter or quality check. Verify the evidence, build a report, then reuse its questions with your next file.

Ask CSV turns common plain-English questions into deterministic calculations inside your browser. It can total or average a metric, rank categories, build a time trend, filter rows, and inspect data quality. Every answer shows its matched columns, formula, and evidence. Pin verified findings into a report, then save its questions as a reusable recipe for next week's file—without storing the CSV or any answer.

01 · How it works

Question in. Evidence out.

There is no model prompt and no hidden server. The tool matches your wording to the file's real columns, runs a defined calculation, then shows its work.

i. drop

Load your file

Drop CSV or Excel, paste CSV text, open a local workspace file, or use the built-in sales sample.

ii. ask

Name the result you need

Ask for a total, average, top or bottom ranking, breakdown, time trend, filter, row count, missing values or duplicates.

iii. repeat

Report once, rerun later

Pin verified findings, export the brief, then save its ordered questions as a recipe for the next compatible file.

02 · What it answers

Useful questions, not magic.

The scope is deliberately concrete. Supported questions return reproducible calculations; unsupported questions return relevant starters instead of an invented number.

  • 01

    Totals and averages

    “What is total Revenue?” or “Average Order Value where Region is North.”

  • 02

    Rankings and breakdowns

    “Top 5 Products by Revenue” or “Average Revenue by Region.”

  • 03

    Trends over time

    “Revenue over time,” “monthly Sessions,” or “average Cost by week.”

  • 04

    Data quality

    “Show missing values,” “find duplicate rows,” or “how many rows are in this file?”

  • 05

    Reusable reports

    Pin exact answers, export their evidence, and save the ordered questions as a data-free recipe for the next file.

"The answer is only useful if I can see exactly how it was calculated."
— the design rule
Good to know

Deterministic by design.

Ask CSV is intentionally narrower than a general chatbot. That constraint makes its answers fast, private and auditable.

  • Column names matter

    Use the metric and grouping names shown beside the question. Common aliases such as sales/revenue and product/item are recognized.

  • Filters are explicit

    Use patterns such as “where Region is North,” “where Status contains pending,” or “where Amount > 100.”

  • No generated narrative

    The engine produces calculations, charts and rows. It will not invent causes, forecasts or business recommendations that are absent from the file.

  • Findings stay tab-local

    Answers, formulas, rows, evidence, and filenames are never persisted. A recipe saves only the name, report framing, and ordered question text you explicitly choose.

03 · FAQ

Ask CSV questions.

Can I ask a CSV questions in plain English?
Yes. Ask CSV recognizes common analysis questions about totals, averages, minimums, maximums, row counts, grouped rankings, time trends, filters, missing values and duplicate rows.
No. It uses a deterministic question parser and calculation engine in your browser. Your file is not sent to an AI service or to csvtodashboard.
Every answer shows the operation, matched metric, grouping and filter, the calculation formula, and the result rows used to support it.
Examples include total Revenue, top 5 Products by Revenue, average Revenue by Region, Revenue over time, total Revenue where Region is North, show missing values, find duplicates, and how many rows.
Yes. Drop an XLSX or XLS workbook and choose a sheet when the workbook contains more than one. The workbook is read locally in your browser.
Yes. Pin verified answers to an analysis brief, reorder them, add analyst context, and export a self-contained HTML file or Markdown document. The report is assembled locally in the current tab.
Yes. Save the brief as an analysis recipe, then choose it before loading the next compatible file. The recipe stores only its name, report title, context, and ordered questions. It does not store filenames, CSV rows, answers, formulas, or evidence.