A typical mid-sized company runs an ERP, an accounting package, a webshop, a POS, and at least one important Excel file. Each system knows part of the truth. To answer “how is the business doing?”, someone has to query all five and add up the answers by hand. The good news: merging them into one view is a solved problem. The bad news: the hurdles sit elsewhere than most people expect, and three of the four biggest ones are not technical.

Hurdle 1: The same thing has a different name everywhere

The customer “Müller GmbH” is “Müller GmbH” in the ERP, “Mueller GmbH, 1050 Vienna” in accounting, and has three accounts in the shop because they ordered three times with different email addresses. To a human, obviously the same customer. To a program, five different ones.

This mapping problem, matching in the trade jargon, is the single biggest line item in almost every consolidation project. It is solved with shared keys such as VAT or customer numbers where they are maintained, plus mapping rules and a review list for the cases no rule catches. Important for planning: this is diligent work full of decisions only your team can make. A vendor can prepare it, but cannot finish it alone.

Hurdle 2: The systems count differently

The shop reports €84,000 for October, accounting says €71,000. Both numbers are correct. The shop counts gross orders by order date, accounting counts invoiced amounts net of tax by invoice date, and the gap consists of VAT, undelivered orders, and two cancellations.

So before anything is merged, each metric needs a ruling on which view wins: which date counts, gross or net, how credit notes, deposits, and cancellations are handled. This sounds dry, but it is the point where five contradictory numbers become one that everyone trusts. Write these definitions down, or the debate starts over in six months.

Hurdle 3: Not every system wants to hand over its data

This is where the technology comes in, and it is more predictable than its reputation. Roughly, source systems come in three tiers:

Source system	Connection	Data freshness	Effort
Modern cloud system with a documented API	done in days	continuously current	low
Older software with an export function, no real API	automated exports, somewhat fragile	usually once a day	medium
Legacy system without an export function	direct database access or report files	depends on the access path	noticeably higher

A word on Excel as a data source

The important Excel file is usually the trickiest source, not for technical reasons but because it is maintained by hand. One shifted column, one typo in a date, and the automated import trips up or, worse, silently reads wrong values.

Two approaches have proven themselves. Either the file gets fixed rules, a protected input sheet with defined columns that the import validates and complains about loudly. Or, often the better answer, the Excel file is replaced by a small input form, which makes the source clean in the first place. What does not work: automatically ingesting a freely growing spreadsheet forever and hoping.

Hurdle 4: The project unearths data corpses, and someone has to decide

As soon as systems sit side by side, inconsistencies surface that nobody noticed before: articles active in the ERP but deleted in the shop, customers with contradictory payment terms, orders stuck on “in progress” for two years. This is not the project failing; it is one of its returns. Such legacy debris costs money even while nobody is looking: Gartner put the cost of poor data quality (2020) at an average of 12.9 million US dollars per organization per year – an average across large enterprises, but the mechanism is the same in mid-sized companies, just with smaller amounts. The cleanup does create work, though, and that work is often missing from the schedule.

Our advice: plan for this cleanup phase from the start and name one person in-house who is allowed to settle disputes. Reporting projects that slip rarely hang on the programming. They hang on nobody deciding for weeks which payment term is the right one.

The sequence that works: two sources first

The tempting plan is to connect all five systems in one big push. The better one is to start with the two sources that together answer the most urgent question, frequently ERP and accounting for revenue and open invoices. That way matching, definitions, and cleanup are rehearsed once at small scale, and the pattern is established.

Every further source becomes cheaper and faster afterwards, because the fundamental questions are settled. Day-to-day operation stays manageable: the source systems keep running unchanged, the consolidation only reads along. So this is not a migration project but a layer on top, one that grows piece by piece. And even the first step puts you ahead of most: according to Statistik Austria (2025), only 26 percent of Austrian companies do any data analytics at all.