Where Data Comes From — and Why the Source Decides Everything

What you'll learn

Name the main places business data comes from — and why the source changes how much to trust it
Explain why two systems can hold different versions of the same fact
Tell a source system, a database, a data warehouse, a dashboard, and a report apart

Marketing says the company has 48,000 customers. Finance says 42,000. Both pulled a real number from a real system this morning. Who’s right? Possibly both — they may be counting different things (“anyone who ever signed up” vs “anyone currently paying”) from different systems that update on different schedules. Before you can trust any number, you have to know where it was born and what happened to it on the way to your screen. This lesson is the tour of those origins. It pairs naturally with the Systems Map lesson elsewhere in this bundle — that lesson names the systems; this one follows the data out of them.

Ten places data is born

Almost every number you’ll ever see started life in one of a familiar handful of places — and each place has its own reliability personality.

Business data comes from operational systems (the tools that run the work), customer systems like a CRM, financial systems like an ERP, HR systems (an HRIS), surveys, website and app activity, sensors or machines, humble spreadsheets, external data bought or pulled from outside, and plain manually entered data. You don’t need to know how any of these work under the hood. You just need one instinct: ask where a number came from, because the source tells you how much to trust it. A figure auto-recorded by a payment system is a different animal from a figure someone typed into a spreadsheet at 5pm on a Friday.

The words that matter

Source system: The place data is first created — the CRM when a deal is logged, the ERP when an invoice is paid, the website when someone clicks.
Database: The organized store where a system keeps its data, in tables of rows and columns. Every source system has one behind it.
Data warehouse: A central store that copies data out of many source systems so it can be combined and analyzed together. On Azure and similar clouds, this is where company-wide reporting lives.
Dashboard: The screen — often Power BI or similar — that turns warehouse data into charts you actually look at.
Report: A fixed document (a PDF, a weekly email, a spreadsheet) built from that data for a specific moment or audience.
Manual data: Anything a human typed in by hand — the most flexible source, and the easiest to get wrong.

Why two systems disagree about one fact

Data rarely travels in a straight line, and every hop is a chance for it to drift. The same “fact” can differ across systems because it’s defined differently (marketing’s “customer” includes free trials; finance’s doesn’t), timed differently (the CRM updated at noon, the warehouse copies overnight, so the dashboard is a day behind), or simply entered differently (one team writes “USA,” another “United States,” and now they look like two markets). On top of that, data can be delayed, duplicated (the same customer entered twice), incomplete (regions that haven’t reported yet), or flatly incorrect (a typo in a key field). None of this means the data is useless — it means “the number” always deserves the follow-up “from which system, defined how, as of when?”

Your dashboard is the last stop on a journey that begins in several separate systems, each on its own clock.

Text description of this diagram

On the left are three source systems — CRM (customers), ERP (money and orders), and HRIS (people) — each the birthplace of its own kind of data. Dashed arrows carry moving dots from all three into a central data warehouse (a cloud platform such as Azure) that copies their data into one place so it can be combined. From the warehouse, data flows onward to a dashboard (Power BI or similar), and from there to you, the business user. The amber banner names the catch: the dashboard is only as fresh as its slowest feed — if one source updated last night and another updates weekly, the single tidy screen quietly blends data of different ages. The neat final chart hides a messy, multi-speed supply chain behind it.

Match the source to its data

Every number has a home system. Drag each fact to the source that owns it — or tap a fact, then tap a source.

CRMcustomer system

ERPfinancial system

HRISpeople system

Web / app activitybehavior data

Tip: drag with a mouse, or tap an item then tap a source on touch screens. Get one wrong and the answer key appears.

Common misunderstanding

“All our numbers come from one place, so they should match.” They almost never come from one place. A single dashboard usually stitches together several source systems, each with its own definitions and update schedule, then adds a warehouse copy step on top. When two reports disagree, the cause is rarely that someone lied — it’s that they drew from different systems, or the same system at different moments, or counted subtly different things. “Which source, as of when?” resolves most of these fights before they start.

Try this at work

When a number matters, ask three things about its origins: which system is it from, how is the thing being counted defined, and how current is it? You’ll often find the “disagreement” between two teams dissolves the moment you line up those three answers. Reflect: pick a number you report regularly — do you actually know which source system it’s born in, or just which dashboard you copy it from?

The bottom line

Data is born in a handful of familiar systems — CRM, ERP, HRIS, surveys, web activity, sensors, spreadsheets, external feeds, and manual entry — then copied into a warehouse and drawn onto a dashboard. The source, its definition, and its freshness decide how much a number is worth, which is why two systems can honestly disagree about one fact.

Why it matters

You’ll spend your career reading the last screen in a long supply chain — and the quality of your decision inherits the quality of that chain.

Knowing where data comes from turns “the numbers don’t match” from a mystery into a checklist. It also sets up the next lesson: once you know a number’s source, the very next question is what kind of number it is — a raw metric, a target, or a true KPI — because those are not the same thing, and confusing them is its own expensive mistake.

Quick check

1. Marketing counts 48,000 customers; Finance counts 42,000. The most likely cause is…

2. What does a data warehouse do?

3. Which source generally deserves the most caution about accuracy?

Answers explained

C is correct — different definitions (trials in or out?), different systems, and different update times routinely produce two honest, non-matching counts. (If you picked A or B: fabrication and breakage are rare; definition-and-timing gaps are the everyday cause.)
A is correct — a warehouse copies data from many source systems into one central store built for combining and analysis. (If you picked B: it doesn’t replace the sources — it reads from them. If you picked C: warehouses hold digital data, not paper.)
B is correct — hand-typed data is the most flexible source and the easiest to get wrong, so it earns the most scrutiny. (If you picked A or C: automatically captured data still has issues, but a human typing by hand is the classic weak point.)