Visualize CSV Online

The picture comes first, the setup comes second

The usual online chart maker opens with a blank canvas and a list of axes to assign. That works if you already know what you want. It works badly in the far more common situation, which is that somebody sent you a file, you have not looked at it yet, and the first thing you want is to see anything at all. So the widget above reads the file first and draws what the file appears to be about, then hands you the controls to argue with it.

The reading step is where most of the useful work happens. Each column is profiled once and assigned a role, which is a stronger claim than a data type. A column of ISO dates is time, so it becomes an axis with a sensible bucket size rather than a list of strings. A column of values with currency symbols is money, so it prints as $16.2k on the axis, in the labels, in the tooltip and in the exported header, all from one decision. A column of short repeating words is a category. A numeric-looking column of identifiers is an identifier, so it never gets offered as something to sum. Roles are read from the values before the header name, which matters: a column called amount full of the words pending and settled is a status.

From there the first chart is a short, boring cascade. Measure plus date gives that measure over time. Measure plus category gives a breakdown. A measure with nothing to group by gives a distribution. Categories with no measure give counts. The aggregate starts at Sum whenever there is a real measure to sum, because opening on a count of rows produces a chart that is technically valid and answers nobody's question.

Every visual tool, one sentence each

No marketing adjectives in this list. Each line says what the tool does and, where it matters, what it refuses to do.

• csv-to-chart · the general chart maker with the bar shapes listed first, covering bar, column, line, area, donut, scatter, histogram and boxplot from one parse.
• csv-to-graph · the same engine with Line, Area and Scatter at the top of the dropdown, for trends, rates of change and relationships between two measures.
• csv-to-heatmap · two columns cross-tabulated into a shaded grid, ordered by total, with combinations that never occurred left blank rather than painted as zero.
• csv-to-pivot · the same cross-tab as a table, with every total recomputed from the source records so an average total is not an average of averages.
• csv-to-treemap · nested rectangles sized by value, for composition with more categories than a donut can survive.
• csv-to-sankey · flow between a from column and a to column, with ribbon width as volume and only positive flows drawn.
• csv-to-timeline · dated events laid out along a time axis in lanes, for release histories, incidents and project milestones.
• csv-to-map · rows plotted geographically when the file carries a latitude and a longitude, or place names it can resolve.
• csv-to-waterfall · the bridge from an opening total to a closing one, with rises and falls colored by direction and nothing else.
• csv-to-funnel · ordered stages with the drop-off between them, kept in file order because the sequence is the meaning.
• excel-to-chart · the same chart maker with an XLSX front door, reading the workbook in the browser and flattening a sheet to the same rectangle.
• json-to-chart · nested JSON flattened into rows and then charted exactly as a CSV would be.
• csv-to-embed · a chart published as an iframe whose data and configuration ride compressed inside the URL, so there is no server and nothing stored.
• embed-gallery · live embeds you can look at before building one, useful for seeing how a shape behaves inside somebody else's page.
• csv-story · one file narrated as a sequence of scenes, each with a chart and a written takeaway, for when the point is an argument rather than a picture.
• templates · ready-made CSV layouts you can fill in, which is the fastest route when you have the numbers but not the file.
• examples · finished charts built from real sample files, so you can see the output before committing your own data.
• dashboards · a KPI strip, several panels and generated insights from a single file, for when one chart is not the whole answer.

Your file does not leave the tab

This is a structural fact about how the site is built rather than a policy someone wrote down. Reading the CSV, inferring the column roles, grouping the rows, computing the aggregates and painting the picture all run inside your browser tab, in code that finished downloading when the page loaded. There is no upload endpoint behind any of these tools, so there is no request that carries your rows, no queue they sit in and no bucket they land in.

Two consequences follow that are easy to check yourself. First, open the network panel in your browser's developer tools and drop a file: no request appears, because there is nothing to request. Second, disconnect from the network entirely once this page has loaded, then keep working. Charts still draw, aggregates still compute, exports still download. A tool that quietly needed a server could not do that.

The one thing worth knowing before you publish is that an embed is different by design. When you build one, the data and the configuration are compressed into the iframe URL so the chart can render without a server behind it. That is what makes an embed permanent and dependency-free, and it also means the numbers travel inside the link. Embed the sales summary; think twice before embedding the payroll.

Frequently Asked Questions