Why the System Doesn't Talk to the Other System

“The system doesn’t talk to the other system.” You’ve heard it from Finance, from Sales, from the intern trying to reconcile two spreadsheets at 6 p.m. It’s the number-one phrase this whole course has been building toward — and after seven modules, you finally have every piece needed to understand it.

Here’s the root cause: data silos. Each system of record you met in modules 2–5 grew up separately — the CRM bought by Sales, the ERP run by Finance, the HRIS owned by HR, the ITSM tool living in IT — each from a different vendor, each with its own idea of what a “customer” or an “employee” even is. Nobody designed the mess; it accreted, one reasonable purchase at a time. So the same customer lives three times: “ACME Corp” in the CRM, “Acme Corporation” in the ERP, “ACME GmbH” in the billing tool. To software, those are three unrelated strings. Which is why the Sales dashboard says 1,040 customers, the Finance report says 812, and both teams are certain the other one is wrong.

Master data and the politics of truth

The fix has a name: master data — the core shared records (customers, products, suppliers, employees) that every system needs a consistent copy of — and MDM (master data management), the project that decides which record wins. Sounds simple. It never is, because MDM is politics wearing a technical costume. Deciding that the ERP’s customer record beats the CRM’s isn’t a database setting — it’s telling the Sales department that Finance’s version of their customers is now the official one. Whose duplicate gets deleted? Whose naming convention survives? Which team must change a process it’s run for a decade? That’s why MDM projects take years and burn political capital, and why “we’ll just clean up the data” is the second most naive sentence after “can’t we just connect them?”

Sync gaps, swivel chairs, and the price of pipes

Even where the pipes from module 7 exist, they leak in predictable ways. Most run as overnight batch jobs, so there’s always a timing gap: Monday morning, the sales dashboard has the weekend’s deals but the finance dashboard won’t get them until tonight’s run — the numbers will agree, just not yet. Where pipes don’t exist at all, humans become the integration: swivel-chair work, named for literally swiveling between two screens re-typing what one system knows into the other. It’s slow, soul-crushing, and error-prone — which is exactly why it’s the number-one target for RPA and the automations you met in the AI & Automation course.

So why not just build every pipe? Count them. Every pair of systems is a separate integration, so N systems means up to N×(N−1)/2 pipes: 4 systems is 6 pipes, 8 systems is 28, 20 systems is 190 — each with its own mappings, error handling, and version upkeep. Nobody can afford that, which is why companies build a hub instead: one place, like a data warehouse or data platform, that every system feeds once, and every report reads from. That hub-versus-spaghetti trade-off is the entire reason the Azure Data for Non-Engineers course exists.

When two systems disagree, do this: ask which one is the system of record for that specific field (module 1’s spine concept — it’s been the answer all along), trust that copy, and report the mismatch to whoever owns the integration or the data. What you should not do is hand-edit the copy you happen to have access to. That “fix” gets overwritten by tonight’s sync at best — and at worst it survives, and now there’s a fourth version of the truth.

Spot it: what’s really going on?

Why it matters

You now hold the complete map this course promised. Module 1 gave you the spine — system of record versus system of engagement. Modules 2–5 introduced the four big records: the CRM for customers, the ERP for money and materials, the HRIS for people, the ITSM tool for IT work. Module 6 explained how you get in; module 7, how the systems reach each other; and this module, why the seams still show anyway. So the next time someone sighs “the system doesn’t talk to the other system,” you can translate it precisely: separate systems of record, missing or broken pipes, duplicated master data, and a batch job that runs at midnight. Better yet, you know the professional move when the numbers disagree — find the system of record, trust it, report the mismatch. That one habit alone puts you ahead of most people who’ve worked with these systems for years.