Build a Dashboard From a Published Table

~30 minutes · a published table, a chart, a shared dashboard

You have a clean table published from a workflow. This tutorial turns it into something people can read at a glance: a chart, then a dashboard built around it, then a dashboard shared with the right people. No SQL is required to get a first chart on screen — you point, click, and query.

If you followed Load, Transform, and Publish Data, the sales_final table you published there is exactly the starting point this tutorial assumes. Any published table works just as well.

What You’ll Build

A single dashboard with two charts over one published table — a Big Number showing total revenue and a bar chart breaking revenue down by a grouping column — formatted, saved, and visible to your workspace.

One published table carries all the way through to a shared dashboard.

Prerequisites

• A PlaidCloud workspace, and a project you can open.
• A published table in that project. If you don’t have one, follow Load, Transform, and Publish Data first, or publish any existing table — right-click it in the Tables tab and choose Set Published Table Reporting Name. See Publishing Tables.
• Nothing to install in the Dashboards area — your dashboard account is created for you the first time you open a dashboard from PlaidCloud.

Only published tables reach dashboards. A dashboard reads a table by its published reporting name, not its internal table name. If your table isn’t showing up as a dataset later, the usual cause is that it was never published, or was published under a name you don’t recognize — check the Published As column in the Tables tab.

Step 1: Open the Dashboards Area

1. From your project in PlaidCloud, open the Dashboards area.
2. The first time you do this, your dashboard account is set up automatically — there is nothing to create by hand. What you can see follows from your workspace membership and security groups.
3. Your published table is now available as a dataset, because it belongs to a project you can access.

For how dashboard visibility maps to workspace membership, see Using Dashboards.

Step 2: Prepare the Dataset

Before charting, tell the dashboard how each column should behave. This is a one-time setup per table that makes the point-and-click explorer work well.

1. Click the edit icon on your dataset.
2. Open the List Columns tab.
3. For each column, set the roles that apply, then Save:

Setting	Turn on when
Groupable	You’ll want to break metrics out by this field (region, product, month)
Filterable	You’ll want to filter on this field
Sum / Min / Max	It’s a numeric measure you’ll total or summarize
Count Distinct	You’ll count unique values of this field
Is temporal	It’s a date or time field

Mark your date column temporal. Time-based filters and the “Since / Until” controls in the explorer only work on a column flagged Is temporal. It’s the setting most easily missed.

Step 3: Build Your First Chart

Start in the explorer, which opens as a plain Table View, and shape it into a Big Number for total revenue.

Use your own measure column. This tutorial charts a revenue column, but the prerequisite is only “any published table” — wherever you see revenue, substitute the numeric column you actually want to total (amount, sales, quantity, and so on).

1. Click the dataset to open it. By default you get a Table View.
2. In the metrics section, add the measure you want — for a running total, type it in and pick your summed column (for example a SUM of revenue), or use COUNT(*) to count rows.
3. Add any Group By fields you want to break the number down by. For a single headline number, leave Group By empty.
4. Click Query to run it and see the result.
5. Scroll to Chart Options (or the visualization type selector) and switch the chart type to Big Number. Charts can be Tables, Big Numbers, Bar Charts, and more — each displays the same queried result differently.
6. Click Query again to redraw.

Now make a second chart that breaks the same measure down:

1. Start a fresh exploration on the same dataset.
2. Keep the SUM of revenue as the metric, and add a Group By on a groupable column such as region or product.
3. Set the chart type to Bar Chart and Query.
4. If you only want the top rows, open Options and set a Row Limit — for example 10 for a top-ten bar chart.

Prefer display formatting over changing the data. You can round or reformat values two ways: transform them in the table, or format them on display. Formatting on display is the common choice — the underlying values keep full precision and every viewer sees a clean number. That’s the next step.

Step 4: Format the Numbers

Raw measures come back unformatted. Fix that on the chart, not in the data.

• Big Number: edit the chart, open the Customize tab, and set Number Format.
• Table: edit the chart, open the Customize tab, select a number column (shown with a #) under Customize Column, and set its D3 Format.

The default is Adaptive formatting, which adjusts to the values. To pin a format, choose one from the dropdown or type a D3 format string:

Format	Input	Shows
,d	12345.67	12,346
$,d	12345.67	$12,346
$,.2f	12345.67	$12,345.67
,.2%	13.215	1,321.50%

The full format grammar — currency symbols, padding, precision, SI prefixes — is in Formatting Numbers and Other Data Types.

Rounding to a whole dollar? Use ,d or $,d, not ,.0f. The ,.0f form is a known source of broken number formatting; ,d (decimal, rounded to integer) is the reliable way to drop the decimals.

Step 5: Add a Calculated Measure (Optional)

If the number you want isn’t a plain column — a ratio, a margin percent, a conditional sum — write it as a metric. Metrics and calculated columns are written in PostgreSQL-flavored SQL on the dataset.

SUM("revenue") / (COUNT(*) + 0.00001)

The + 0.00001 avoids a divide-by-zero when a group has no rows. For more patterns — CASE logic, COALESCE for nulls, conditional sums — see Example Metrics and Example Calculated Columns.

Step 6: Assemble the Dashboard

A dashboard is a layout of saved charts. Associate each chart you built with the same dashboard so they land together.

1. Open Edit Chart on your Big Number.
2. In the Dashboards section, enter the name of the dashboard it should appear on. If the dashboard doesn’t exist yet, this is where you name it.
3. Save, and repeat for your bar chart, pointing it at the same dashboard.
4. Open the dashboard to see both charts. From the dashboard’s own edit mode you can arrange the tiles, resize them, and add filters.

For the full range of layout, filter, and interaction options, see Learning About Dashboards.

Step 7: Share It

Who sees the dashboard follows from PlaidCloud access first, then any dashboard-service roles on top.

• By default, it’s visible to the project’s people. The dashboard list in PlaidCloud shows the published dashboards belonging to projects a person can access — so anyone who can open the project can find your dashboard, with no extra step.
• Add chart owners as you go. When you save a chart, add everyone who needs it to the chart’s Owners — otherwise a colleague may hit a “requires the datasource… permission” error on a chart they can’t reach. Adding owners at save time is easier than retrofitting access later.
• Restrict to specific roles only if you need to. A dashboard can be limited to particular roles in the dashboard service, set on the dashboard’s own properties. A restricted dashboard is hidden from the list for anyone without a matching role — so if you restrict one, tell the people who should have it. Project access in PlaidCloud still applies on top: a role grant never hands out a dashboard for a project someone can’t reach.

See Using Dashboards for the full visibility rules.

Keeping the Dashboard Current

You don’t refresh dashboards by hand. When a workflow republishes the table behind a chart, every chart reading it drops its cached results, and the next person to open the dashboard sees the new numbers.

A stale-looking dashboard is almost never the cache. If numbers look old after a workflow run, check the workflow — that the run finished and republished the table you expect — not the cache. Republish under the same published name, too: charts stay attached to a name, so a table rebuilt under a new name leaves the old charts pointing at old data.

If you ever do need to force it, open the menu beside Edit dashboard and choose Force refresh dashboard. Full details, including how long results are held, are in Dashboard Caching.

Where to Go Next

• Learning About Dashboards — chart types, layout, filters, and troubleshooting
• Example Metrics and Example Calculated Columns — SQL patterns for custom measures
• Formatting Numbers and Other Data Types — the full D3 format reference
• Dashboard Caching — how refresh and cache lifetime work
• Load, Transform, and Publish Data — the tutorial that produces the published table this one starts from