CSV profiler
Column types, nulls, uniques, distributions and correlations from a CSV, in one pass.
/analyze
Free · no account · nothing uploaded
CSV analytics builder
Turn a raw export into an answer: pick the columns that matter, filter to the rows you mean, group by a category, aggregate a metric, and chart the result.
Analytics builder
Say it the way you would say it to a person. Your words are matched to one of 18 tools by readable rules, and the tool runs right here in your browser. No upload, no account.
This tool runs in your browser and needs JavaScript turned on.
esc
How to use the analytics builder
- Drop a CSV into “Your files”, or try the sample sales table.
- Describe the analysis: “average salary by department”, “count rows by region” or “sum revenue where region is North”.
- Check the grouped table and its bar chart. The answer names the leading group and how far ahead it is.
- Download the result as a CSV, or open the settings form to set the columns, filter, grouping and aggregation by hand.
Things you can type: group by region and sum revenue average price by category
The four operations that answer most questions
Select, filter, group, aggregate. Almost every question anyone asks of a table is some arrangement of those four, and a surprising amount of analytical work is just applying them carefully in the right order.
This tool makes each one an explicit step rather than a formula buried in a cell, so the analysis you built is legible afterwards — including to the person who has to check it.
What you can build
- Column selection, to cut a forty-column export down to the six that matter.
- A row filter on any column: contains, equals, or a comparison such as greater than.
- Grouping by a categorical column, with counts, sums, averages, minima and maxima per group.
- A chart of the grouped result, and a column-quality panel so you know what you are charting.
Local, and repeatable
The file never leaves your machine, and each run leaves its configuration and its result in the transcript as a block you can read later or clone into a variation. That is the same contract the FlexContext workspace uses: a tool run is a record, not a disposable message.
Where a grouped analysis goes wrong
Grouping is where most honest mistakes happen. An average over a group hides its spread, so two groups with identical means can behave nothing alike. A total over a filtered set quietly answers a different question than the one you asked, because the filter removed rows you had not accounted for.
The classic trap is worse than either: a trend that holds within every group and reverses when the groups are combined. It has a name — Simpson’s paradox — and the only defence is to look at the groups and the whole, which is two clicks here rather than two spreadsheets.
The column-quality panel exists for the same reason. A group of four rows and a group of forty thousand look identical in a bar chart, and only one of them means anything.
Questions
Analytics builder: common questions
Do I need a spreadsheet application?
No. The whole thing runs in the browser tab, including the chart. Nothing is installed and nothing is uploaded.
Can I export the result?
Yes. Every run puts its table in Results as a CSV you can download, or hand to the format converter for JSON. The chart is drawn in the page.
Does it use AI to analyse the data?
Not here. These are deterministic operations you configure yourself — which is what makes the result reproducible. AI-assisted analysis over a persistent library of sources is what the FlexContext product adds.