Dashboard examples
Most tools make you upload a file before they will show you what they can do. This gallery is the opposite. Ten sample datasets, each with a preview drawn from its own rows, so you can see the output for a shape of data you recognize and then decide whether to go looking for your own export.
The previews are the data
A gallery of stock illustrations would have been quicker to build and would have told you nothing. Every sparkline on this page is drawn when the page loads, from the same sample file the card links to, using the same grouping code the chart tools use. If a dataset trends downward in the second half of the year, its sparkline slopes down, because that is what its rows do.
The choice of sparkline is itself a small demonstration of how the tools behave. A file with a date column gets a line of its main measure over time, bucketed at whatever grain suits the span. A file with no date column gets a row of small bars, one per category, sorted biggest first. That is exactly the fork the auto-pick cascade takes on the chart pages: a measure with a date is a trend, a measure with a category is a breakdown. You can watch the rule work here before you meet it there.
The line under each blurb is generated too. It names the leading measure, its total in the right units, and either the date range or the number of distinct categories. Money reads as money because the column was inferred as currency; a count reads as a count. One inference, applied on the card, in the chart, in the axis and in the export.
What each example is here to show
- Sales is the default shape: a date, two categories, a money column. If your file looks like anything, it looks like this one. Start here.
- Marketing has two money columns pulling against each other, spend and revenue, which is the case where picking the measure by column order rather than by size gets it wrong. Here the larger total wins, which is the one you meant.
- Support has a long right tail on resolution time. It is the argument for having a median at all, and the reason a mean resolution time describes a ticket nobody filed.
- People carries salaries, which is the file where a distribution is honest and an average is not.
- E-commerce has a status column with returns and cancellations in it, so the same rows answer two different questions depending on what you group by.
- SaaS mixes churned and active accounts, which makes the split panel worth reading rather than decorative.
- Web has a percentage column. Summing a column of rates produces a number with no referent, and the role inference is what stops that happening.
- Finance is categories with a workflow attached, and the shape that naturally wants a bridge rather than a bar chart.
- Inventory has no date column at all. Watch what the layout does when there is nothing to trend: it leads with breakdowns and a distribution instead of drawing a line through nothing.
- Geo carries real latitude and longitude. It is on the list because a column of North, South, East and West is not geography, and a map drawn from one is a chart that implies a place.
Frequently Asked Questions
Are the sparklines on the cards real?
Yes. Each card fetches its own sample CSV when the page loads, aggregates it the same way the tools do, and draws the result as inline SVG. A dataset with a date column gets a line of its measure over time; one without gets a row of bars by its main category, because a line drawn through unordered categories implies a sequence that does not exist. The row and column counts printed above each title come from the same read, so nothing on the card is decorative.
Why show examples at all instead of just asking for a file?
Because most tools make you hand over a file before they show you anything, which is a strange bargain when you have not yet decided whether the tool is any good. This gallery runs the other way around: pick a shape of data you recognize, see exactly what comes out, then decide whether to go and find your own export. It is also the fastest way to check a specific behavior, such as what happens to a file with no date column.
What is in each sample file?
Between 16 and 180 rows of deterministically generated data with realistic column names, roles and distributions for its domain. They are generated by a seeded script that produces byte identical files every run, so a screenshot taken today still matches next year. Every one is downloadable from its card if you want to look at the raw rows or use it as a fixture in your own tests.
Where does View dashboard take me?
Into the app with that sample loaded and the dashboard view open. Nothing is uploaded on the way: the app fetches the sample file from this site, which is a public static file, and builds the dashboard in your browser. From there you can clear it and drop your own file into the same layout.
Can I use these files in my own work?
Yes. They are synthetic, contain no real people or companies, and you are welcome to use them as demo data, as test fixtures, or as a shape to model your own export on. The names are invented and the numbers are pseudorandom from a fixed seed.
Which example should I look at first?
The sales one, if your file has a date, a couple of categories and a money column, which covers most business exports. Look at the inventory example if your data has no date at all, because it shows how the layout reorganizes itself around breakdowns and a distribution when there is nothing to trend. Look at the web analytics example if you have a percentage column, since averaging rather than summing a rate is the single most common way a dashboard gets a number wrong.
Do the examples cover every chart type?
Between them they exercise trends, breakdowns, splits, distributions, correlations, geography and bridges. The geo example carries real latitude and longitude on purpose, because a column of North, South, East and West is not geography and a map drawn from one is a chart that implies a place. If you want to see a specific shape, the Chart it link on any card opens that sample in the chart tool where you can switch types freely.
Is anything stored when I open an example?
No. The samples are ordinary static files served from this site, and every calculation happens in your browser. There is no session, no account and nothing written down about which example you looked at beyond ordinary web server logs.