Tableau supply chain dashboard: files, joins and checks

A Tableau supply chain dashboard needs reconciled input files, complete relationship keys and consistent date filters before it needs charts. Start with the order, shipment and inventory exports already used in the business review. Prove their totals outside the presentation, then check the same figures in each dashboard view.

This guide concentrates on preparing and validating those files. The broader supply chain dashboard guide covers measure selection, while the control tower dashboard guide covers cross-site exceptions and accountable owners.

Define what each row represents

For a US parts distributor, the service review covers commitments due October 20 through October 22, 2026, assessed as of October 23. The order export has one row per order line. Its completion date is populated only when the full ordered quantity has shipped; otherwise it remains blank.

Columns A through G on an Orders sheet contain the following source values. Column H is a calculated on-time flag.

The Shipments sheet has a different row definition: one shipment event for an order line. Partial shipments create multiple rows.

The inventory export is a point-in-time balance, not activity during the service period. Its rows describe site, item and snapshot date together.

Do not remove the date when later snapshots are appended. A site-item key that is unique in this file will no longer be unique across multiple dates.

Establish spreadsheet control totals

In H2 of Orders, use this Excel formula and fill down:

=--AND(G2<>"",G2<=D2)

Dates must be valid and the due period already selected. A fully shipped line completed by its due date receives 1; a late or unfinished commitment receives 0. Missing completion data must be distinguished from a confirmed unfinished order before publishing.

Check shipped quantities against each order. On the order sheet, use:

=SUMIFS(Shipments!$B$2:$B$6,Shipments!$A$2:$A$6,A2)

The result must equal column E for each completed line in this example. Microsoft's SUMIFS documentation explains summing values that meet specified criteria. Investigate over-shipments, returns and canceled quantities separately rather than forcing a match.

The overall controls are:

=COUNTA(A2:A5)
=SUM(E2:E5)
=SUM(F2:F5)
=SUM(H2:H5)/COUNTA(A2:A5)

Percentages are shown to one decimal. Equal ordered and shipped quantities do not prove service: order B is fully shipped but still late. The inventory balance has its own control and is not reconciled by equating it with shipments.

Set relationships before building the views

Tableau recommends considering relationships first when combining tables. Its documentation explains that related logical tables remain separate and retain their individual levels of detail. In the data source view, add the required tables to the logical canvas and define the relationship fields explicitly.

Relate order lines to shipment events using the complete order-line identifier. If identifiers repeat across companies, include the company identifier. Do not assume that matching item codes identify an order.

For stock, establish which site-item snapshot belongs to the review before relating it to other measures. Test the result with the actual fields used in the view. A relationship does not settle the business meaning of an inventory date or allocation rule.

A flat join illustrates the danger. Order A appears beside both shipment events, so its $1,000 repeats. Summing order value after that join produces 1,000 + 1,000 + 400 + 800 + 200 = $3,400, instead of $2,400. If stock is then repeated beside the shipment rows, Dallas appears three times and Reno twice: 3 × 1,200 + 2 × 800 = $5,200, instead of $2,000.

Test filters with a small site reconciliation

Create separate views for order service, shipment activity and snapshot inventory. Label their date meanings. Tableau documents dragging dimensions, measures and dates to the Filters shelf; use the relevant date field in each view and verify the selected scope on the dashboard.

Filter the service view to Dallas. It must show order value of 1,000 + 400 = $1,400 and service of 1 / 2 = 50.0%. Reno must show 800 + 200 = $1,000 and 2 / 2 = 100.0%. Restoring all sites must return $2,400 and 75.0%, not an average of displayed site percentages.

These site checks help expose a filter applied to the wrong date or a repeated balance. For presentation choices after reconciliation, see Tableau dashboard design examples.

Where it goes wrong

• Shipment rows become the service denominator. A partial shipment is an event, not another customer commitment.
• Snapshot dates are ignored. Summing several dates counts the same inventory repeatedly across time.
• Incomplete keys create extra matches. Site, company or order-line identifiers may be necessary even when item codes look familiar.
• Filters silently change the population. A shipment-date filter can exclude overdue orders with no shipment. Compare due commitments at a stated cutoff.

Keep the signed-off control totals with the review. AI analytics for supply chain provides the broader business analysis context. Covirage is not affiliated with Tableau or Microsoft.

Check your presentation inputs in Covirage

Bring a sample of your own order, shipment or inventory export to Covirage before presenting the results. Covirage maps the columns, its tools compute and check the figures, and the external AI model explains the comparisons with every figure traced to its rows. Create a free Covirage account, then set up your business and bring your own files.

See Covirage for supply chain teams for the supporting industry context.