UK Audit File (SAF-T)

Transaction Tax

Transaction-level sales, purchase and ledger data for a tax inspection of a UK entity, built on Oracle’s SAF-T extract in its Generic format, in Excel or XML. Oracle’s extract offers Generic, Norway and Romania formats; there is no HMRC-specific one (Oracle, status October 2026).

Related Consolidates the former Excel and XML variants — choose the output format at run time.

Run note · Performance The subledger (XLA) join — especially Projects-to-XLA — is a known BI Publisher performance trap on large ledgers. Pre-join it in the warehouse, or apply tight ledger, period, and source filters before running online.

Run note · Data latency Several tax reports read the Tax Reporting Ledger extract rather than live transaction tables — run the extract first or the report returns stale or empty results.

Sample build of the UK Audit File (SAF-T) — reconciled, and rendered tool-neutral so it runs in Power BI, ThoughtSpot, or Tableau.

UK Audit File (SAF-T)

Sample build · illustrative

Filters

Period

FEB-26

Ledger

US Primary

Currency

USD

18,400

Transactions

Periods

99.6%

Complete

Document Type	Transaction	Date	Tax Code	Net	Tax	Account
Standard	Sample	2026-04-30	Sample	Sample	Sample	1000-2100-000
Corporate	—	2026-03-31	—	—	—	1000-5400-000
Standard	Sample	2026-02-28	Sample	Sample	Sample	1000-1410-000
Default	—	2026-01-31	—	—	—	2000-2100-000
Standard	Sample	2025-12-31	Sample	Sample	Sample	1000-6300-000
Standard	Sample	2026-04-30	Sample	Sample	Sample	1000-2100-000

AI Analyst · active

reading

The report assembles sales, purchase, and ledger detail with tax codes in Oracle’s Generic SAF-T layout.

flag

A small set of transactions carry no tax code, so they'd appear in the audit file as untaxed and prompt a query from the tax inspector.

root cause & next step

Assign tax codes to the untaxed transactions before generating the file; an untaxed line is the first thing an inspector flags.

Illustrative data. The live interactive version — drill-through, filters, export, and the AI Analyst — runs on your warehouse. See it live →

This is the report's BI Publisher data model — the SQL data set BI Publisher runs against Oracle tables to produce the output. The same SQL becomes a dbt model in your warehouse, so one definition drives both the formatted report and the analytics layer.

Data sources

How it interconnects: this data set reads the physical tables above. Those same tables surface in OTBI as subject areas and in BICC as PVOs — three lenses on one source. Open any table to trace its subject areas and View Objects.

The SQL data set is authored to this report's exact spec during the build and ships as the BI Publisher data model plus a matching dbt model — one definition, both layers.

The data-warehouse model — one fact surrounded by conformed dimensions (what you slice by) and measures (what you aggregate), expressed as dbt so it migrates with you. Grain: one row per source transaction.

●— fact → dimension join

Element	Type	Definition
AP_INVOICES_ALL	dimension	dimension
XLA_AE_LINES	dimension	dimension
ZX_LINES	dimension	dimension
GL_CODE_COMBINATIONS	dimension	dimension
Net	measure	measure

Runs on your cloud warehouse — Snowflake, BigQuery, Redshift, or Synapse on AWS, Google Cloud, Azure, or any provider. Reconciled to the source control total — 0% variance by design. You own the code, the model, and the data.

How the data gets here: a BICC bulk extract of the source tables above, on the same pattern for every report. See the extraction pattern & data flow →

See the complete model

How this report's fact and dimensions fit the full picture, via conformed keys.

Transaction Tax data model →Enterprise model →

Every source object behind this report. Each linked table has its own page with full column descriptions, drawn from the Oracle BICC lineage and articulated for practitioners.

Table	Reporting columns	Subject areas
RA_CUSTOMER_TRX_ALL	58	16
AP_INVOICES_ALL	63	15
XLA_AE_LINES	23	17
ZX_LINES	No page in this catalog’s data dictionary yet
GL_CODE_COMBINATIONS	7	61

Reporting columns = fields the report selects that are exposed as analytics attributes; subject areas = the OTBI subject areas the table appears in. Setup and configuration tables (master data, ledger and book setup, lookups) are referenced by the report's joins but aren't exposed as analytics columns or subject areas — that's expected, not a gap.