Fabric – including its Lakehouse and Warehouse – does not enforce primary key constraints. It explicitly requires 'NOT ENFORCED' on constraint creation. This means duplicate rows from pipeline retries, overlapping incremental windows, fan-out joins, or schema errors are silently absorbed. Zap lands data in SQL Database for Fabric with enforced primary keys. If a duplicate tries to load, it's rejected immediately. Across hundreds of tables and hundreds of loads, this isn't a nice-to-have – it's the only way to guarantee your totals are correct.

Most incremental pipelines only reprocess rows where the source timestamp changed. But what about the thousands of existing rows that look up a value from a dimension table – like a customer’s region or a product’s category – that also changed? Those rows are now stale and nobody reprocessed them. Zap checks timestamps on all looked-up columns to build the true minimal incremental set. You get the performance benefit of incremental without the silent staleness.

Zap transforms each source change as two transactions: one to reverse the original value, one to add the new value – similar to a ledger adjustment. When synchronized to Power BI via XMLA, this enables add-only processing rather than full partition reloads. Power BI processes additions much faster than updates – the result is 5× faster refreshes. Faster refreshes mean less Fabric capacity consumed and lower bills.

Microsoft actually tried this with “Default Semantic Models” – and sunset the feature in September 2025. The semantic model is a separate artifact with its own relationships, measures, and hierarchies. When your warehouse schema changes, you have to manually update the semantic model definition – or write C# against XMLA endpoints. Zap’s model drives both layers. Publish once, both update. No scripting, no drift.

No. Zap extracts from your ERP and transforms inside Fabric using ELT – your data never leaves the platform. Zap governs the star schema, enforces keys, and synchronizes the semantic layer. Fabric provides the infrastructure Zap runs on, plus notebooks, data science, and AI when you’re ready.

Most customers are running trusted financial reports in days, not months. Pre-built models for GL, AR, AP, and inventory eliminate the bulk of implementation time. But these aren’t mandatory – you can build fully custom models on the same platform with the same enforced keys, smart incremental, and synchronized publishing. The pre-built just means you don’t have to reverse-engineer what we’ve already validated hundreds of times.

Zap Live connects to the same data model as your Power BI dashboards, ensuring a single source of truth while preserving financial statement formatting during hierarchical navigation. Finance teams can create their own calculations without IT assistance and refresh ERP data on-demand – true self-service for financial reporting.