Before Joule goes live: data quality in multi-country SAP

Where the drift comes from

In most multi-country SAP setups, material classes, customer classifications, and tax codes were defined locally — per country, per implementation wave, sometimes per consultant. Each rollout had its own deadlines, its own integrator, its own historical compromise on the customizing tables. None of them were wrong locally. None of them were ever harmonised globally, because nobody queried across countries at the same time.

The reports were always local. The dashboards were per entity. The data worked fine — for local questions. Joule asks cross-country questions. That's when decades of drift surface in a single prompt.

A concrete example: tax code "V1"

Tax code V1 in an Austrian entity might be standard domestic VAT. In a German entity, the same code might be reverse-charge B2B. In a Swiss entity, it might not exist at all, or might point to a Mehrwertsteuer bracket that doesn't map cleanly to either.

Joule aggregates by the field ID. "V1 across DACH" returns a number. The number is the sum of three different things. Nobody on the call will catch it, because nobody on the call has reviewed T007A across all three entities in the last decade.

The same pattern shows up in customer classifications (KNVV-KDGRP, KLASS), material classes (MARA-MATKL, MARC), and document types (TVAK). The fields look identical. The semantics drifted years ago.

The numbers behind the pattern

The state of multi-country SAP master data is well-measured inside the community, even if the consequences for AI weren't:

• Up to 94% inactive vendors carried in master data records.
• 54% missing VAT numbers across cross-border vendor masters.
• 43% missing bank details in payment-relevant records.
• 70–85% of AI project failures trace back to data quality (Gartner, IDC, RAND).

SAPinsider's 2026 framing is plain: "Data quality is a prerequisite for Joule, not a byproduct." The framing is correct. The infrastructure to act on it across countries is what most SAP shops don't yet have.

SAP knows — but the awareness gap is real

SAP acquired Reltio in May 2026, specifically for cross-system master-data matching. SAP Business Data Cloud is positioned as the layer where this gets consolidated. Both are real moves.

But 83% of DSAG members reported in 2026 that they hadn't yet evaluated Business Data Cloud. The tooling is coming. The awareness is lagging by 12–24 months. Joule pilots are going live in that gap.

Five questions before the first cross-country pilot

Asking five short questions to the country leads — about material classes, customer classifications, tax codes, document types, and validation strategy — tells you whether the data is AI-ready. Every answer that starts with "we define it locally" is a drift signal.

We've turned the five questions into a one-page diagnostic, designed to be filled in by the country leads in 15 minutes per country. The output is a clean go / no-go signal for the cross-country Joule pilot — before the CFO sees a confident wrong answer.