---
type: Website Page
title: "Before I Build a Dashboard, I Make the Numbers Agree | Pixeldust"
description: "My practical approach to manufacturing dashboards: reconcile orders and shipments, define useful metrics, and build reporting a Brazos Valley team can trust."
resource: "https://pixeldust.net/blog/manufacturing-dashboard-trustworthy-numbers-brazos-valley"
tags: [pixeldust, website]
status: stable
sources:
  - id: canonical-page
    resource: "https://pixeldust.net/blog/manufacturing-dashboard-trustworthy-numbers-brazos-valley"
    title: "Before I Build a Dashboard, I Make the Numbers Agree | Pixeldust"
---

# Before I Build a Dashboard, I Make the Numbers Agree | Pixeldust

> My practical approach to manufacturing dashboards: reconcile orders and shipments, define useful metrics, and build reporting a Brazos Valley team can trust.

## Source

- [Open the canonical human-readable page](https://pixeldust.net/blog/manufacturing-dashboard-trustworthy-numbers-brazos-valley)
- [Browse the Pixeldust knowledge index](https://pixeldust.net/okf/index.md)

## Page content

Manufacturing + data · Brazos Valley

# Before I Build a Dashboard, I Make the Numbers Agree

Here is how I turn an order-backlog question into reporting a team can verify, trust, and use.
[Mick Southerland By Mick Southerland Founder, Pixeldust Interactive](https://pixeldust.net/about) November 19, 2026 About 6 minutes
Conceptual illustration of manufacturing orders and shipment records connecting to an organized business dashboard Conceptual illustration
In this guide The local manufacturing question Agree on definitions Trace the records Show data freshness Choose the right tool Prove the result

When I plan a business dashboard, I want to know which decision it will help someone make on Monday morning. If the answer is “see everything in one place,” I keep asking. A screen full of charts can still leave the person running the business opening three spreadsheets to figure out which order needs attention.

For a manufacturer, I would start with something narrower: which customer commitments are at risk, and what is holding them up? That question gives the project a purpose. It also exposes the work that needs to happen before anyone chooses a chart.

## Why I would start with the order backlog

The [Bryan–College Station Chamber’s manufacturing directory](https://business.bcschamber.org/list/category/manufacturing-158) lists local businesses spanning cabinetry, pipe and tube, cleanrooms, and other production work. That is a useful reminder that our business community includes companies making and shipping physical products, with very different operating requirements.

I would not assume those companies use the same software or have the same reporting problems. The example below is illustrative: a small Brazos Valley manufacturer keeps customer orders in one system, production updates in a spreadsheet, and shipments in another. The owner wants to know what can ship this week. Each system answers a different part of that question.

Here is how I would approach that project through Pixeldust Interactive’s [Power BI and custom dashboard development service](https://pixeldust.net/services/power-bi-custom-dashboard-development-college-station) .

## I settle the meaning of “open” before counting orders

An accepted order, a completed production job, a shipment, and an invoice are different events. Calling all of them “sales” makes a report ambiguous before the first calculation runs.

I ask the owner and the people scheduling work to agree on a short definition sheet. Does open backlog exclude cancellations? Is an order late against the original promise or a revised date? Does a partial shipment leave the whole order open? Who can change a promise date, and is the previous date preserved?

For our example, I would track remaining quantity at the order-line level: ordered quantity minus cancellations and shipped quantity, with unusual adjustments flagged for review. I would show overdue lines, lines due soon, and missing promise dates separately. A missing date is an exception to investigate, not evidence that everything is on schedule.

## Then I follow one order all the way through

I would take a few representative orders and trace their identifiers through production and shipping. Customer names are unreliable matching keys: spelling changes, divisions share names, and someone eventually types an abbreviation. I want a stable order ID and line ID, with a documented way to connect shipments.

Here is a common modeling trap. One line orders ten units and has two shipments. Joining the order row directly to both shipment rows can repeat the original quantity and make ten look like twenty. I would aggregate shipments to the correct level before calculating the remaining balance, or keep separate related tables with carefully defined measures.

Microsoft’s [Power BI modeling guidance](https://learn.microsoft.com/en-us/power-bi/guidance/star-schema) explains why consistent record granularity and appropriate relationships matter. My practical test is simpler: can I point to the records behind the number and reproduce it without trusting the chart?

## I make stale information visibly stale

A dashboard can load successfully while yesterday’s shipping import has failed. I want the screen to show when each important source was last updated, whether its latest load succeeded, and which reporting cutoff the numbers represent.

I would agree on the required freshness with the team. A morning planning meeting might work well with a verified overnight update. A dispatcher making immediate shipping decisions may need something faster. I would not sell real-time infrastructure for a decision made once a day.

For Power BI, I check source credentials, refresh ownership, and any required gateway against [Microsoft’s refresh documentation](https://learn.microsoft.com/en-us/power-bi/connect-data/refresh-data) . My implementation plan includes a named recipient for failure alerts and a recovery procedure. A refresh schedule without someone responsible for it is an unattended promise.

## The workflow decides whether I recommend Power BI

I consider Power BI when the main job is filtering, comparing, and explaining data, especially if the business already uses it. I consider a custom dashboard when staff need purpose-built actions such as assigning an exception or requesting a revised shipping date. Reporting and changing operational records deserve separate permission and validation decisions.

Before choosing, I account for viewer access, licensing, source-system API limits, maintenance, and who will own the model and connections. A spreadsheet or the existing ERP’s report may be enough when the question is simple and the figures already reconcile. Custom software adds responsibilities as well as flexibility.

I also keep internal reporting private. Microsoft explicitly warns that [Power BI’s Publish to web feature exposes reports publicly](https://learn.microsoft.com/en-us/power-bi/collaborate-share/service-publish-to-web) . I would never use that shortcut for customer orders or confidential production data. I test access as each intended user role, including what happens when someone follows a shared link.

## What I would require before calling it finished

I would pilot one backlog view alongside the current process. Together with the team, I would check partial shipments, cancellations, changed dates, duplicate imports, missing identifiers, and a failed source update. Every unexplained difference goes on a reconciliation list with an owner.

Before hiring anyone, I recommend asking: who defines each metric; how can I inspect the underlying records; what happens when a source fails; what will ongoing access and support cost; and can another developer maintain this? Those questions reveal more than a polished demo.

Success means the figures reconcile, exceptions reach the right person, and the team spends less time assembling its morning answer. I would record the current preparation time and unresolved discrepancies before the pilot, then compare them afterward. I would also ask whether the view changed an actual scheduling decision.

This is how I approach dashboard work for College Station businesses: understand the decision, establish trustworthy records, and build the smallest view that helps someone act. If the numbers still disagree, my next step is to resolve the disagreement. Another chart can wait.

Have an operational question your reports cannot answer?
[Tell me which numbers disagree and what decision they delay](https://pixeldust.net/inquiry) . I can help identify whether the useful first step is a reporting fix, a connection between systems, or a focused dashboard.

By Mick Southerland
Founder, Pixeldust Interactive

[^canonical-page]: Content derived from the canonical Pixeldust Interactive page linked above.
