Build a Tableau Dashboard

What this covers

A complete walk from a blank Tableau Desktop workbook to a small, polished dashboard on the Tessallite demo model: a sales map-by-country table, a product leaderboard, a monthly trend, and a filter that ties them together — on a live connection, so the dashboard reads current governed data every time it opens.

Before you start

• Tableau Desktop.
• Your Tessallite sign-in, the workspace slug, and the gateway address. On the local demo: admin@acme-demo.com / acme-demo, slug acme-demo, gateway localhost, port 5433.
• The connection mechanics are the same as any PostgreSQL source; the reference is Connect a BI Tool via JDBC. This page starts once you can see the model in Tableau.

Step 1 — Connect, and stay live

1. Open Tableau Desktop. In the Connect pane, choose PostgreSQL.
2. Server: localhost. Port: 5433. Database: acme-demo (or acme-demo/modelx to scope to the model this walkthrough uses).
3. Username: your Tessallite email. Password: your Tessallite password. Click Sign In.
4. On the data source page, drag the modelx table onto the canvas.
5. At the top right, choose Live (not Extract).

Why live matters here. An extract freezes a copy of the data into the workbook. It cannot be accelerated by new summaries, it drifts stale until refreshed, and — most quietly dangerous — it is a copy of governed data sitting outside the reach of the rules. A live connection asks Tessallite a fresh, security-checked, accelerated question every time someone opens or filters the dashboard.

Step 2 — Meet the model in the Data pane

Look at the left-hand Data pane. Everything in it comes from the semantic model:

• Dimensions — the ways to slice: country_code, product_name, customer_segment, channel_code, business_date, and friends.
• Measures — the governed numbers: net_sales, gross_margin, transaction_count, Revenue.

The measures arrive with their definitions already attached. net_sales is not "whatever sum someone typed into this workbook" — it is the model owner's definition, the same one Excel and Power BI see. Tableau aggregates them as you drag, and Tessallite computes the aggregation in the database.

Tip: let Tableau aggregate, never pre-sum. Drag the raw measure and let Tableau wrap it (SUM(net_sales)). Do not build a custom SQL fragment that sums first — the model's measures are defined to be aggregated at query time, and pre-summing breaks exactly the cases (ratios, distinct counts) the definitions protect.

Step 3 — Sheet one: the country view

1. Double-click country_code. Tableau reads it as geography and draws a map (if it lands as text instead, drag country_code to Rows — a tidy table is just as good for this walkthrough).
2. Drag net_sales onto Colour (on the map) or onto Text (in the table). Darker or bigger means more sales.
3. Rename the sheet Sales by Country (double-click the tab).

That double-click was a query: Tessallite grouped sales by country and returned only the result — one value per country — not the hundred thousand rows behind it.

Step 4 — Sheet two: the product leaderboard

1. New sheet (Worksheet > New Worksheet), named Top Products.
2. Drag product_name to Rows, net_sales to Columns. Bars appear.
3. Click the Sort descending button in the toolbar — longest bar on top.
4. To keep it to the leaders: drag product_name to Filters, open the Top tab, choose By field > Top 10 by SUM(net_sales).

The Top 10 is computed over the full set of products in the database, then the ten winners are drawn — the genuine leaderboard, not the ten biggest of whatever was on screen.

Step 5 — Sheet three: the monthly trend

1. New sheet, named Monthly Trend.
2. Drag business_date to Columns. Tableau defaults to YEAR; click the little + on the pill (or right-click it and choose Month) to walk down to months.
3. Drag net_sales to Rows. The trend line draws itself.

Step 6 — Assemble the dashboard and wire the filter

1. Dashboard > New Dashboard. Drag the three sheets onto the canvas: country view top-left, leaderboard top-right, trend along the bottom.
2. On the Sales by Country sheet (inside the dashboard), click the small funnel icon — Use as Filter. Now clicking a country re-queries the other two sheets for that country alone.
3. Add a quick filter for a second slice if you like: on any sheet, right-click channel_code > Show Filter, and the tick-box panel appears on the dashboard.

Every click on the dashboard is a fresh live query. Tessallite answers each one through the same routing as every other tool — summary when one fits, governed source path otherwise — and applies your security on the way.

Step 7 — The number-checking habit

Before you circulate the dashboard, do the five-second check Maya does: glance at the grand total of net_sales on the leaderboard sheet and compare it with the same figure in the Tessallite web app's query panel (or ask the agent). They match — they must, because both come from the one model — and being able to say "this ties to the governed number" is what separates a dashboard people trust from one people debate.

Common trap: blending in a second, ungoverned source. Tableau makes it easy to drag in a spreadsheet of "adjustments" and blend it with the model. The moment you do, the dashboard shows numbers nobody else can reproduce and the trust check above fails. If an adjustment is real, it belongs in the model — one definition, every tool. See Why Your Numbers Match.

Troubleshooting

Related

• Connect a BI Tool via JDBC
• JDBC Connection Guide
• Build a Power BI Report
• Choose Your Connection