Finance & accounting: five digits separate a SIREN from a SIRET
A French SIRET number has 14 digits, a SIREN has 9, and both identify the same company. You need to know which one you're searching for.
Exact match, never approximate
In a ledger of accounting records or a business directory, the user is looking for an exact identifier — not something approximate. SIRET, IBAN, VAT number: each has its own structure (the detail of each identifier). You want the right invoice, not one that merely resembles it.
Each identifier, its own structure
The finance rule pack activates when you pick this industry, at indexing time. It produces these annotations, each becoming a usable filter:
| Annotation | What it isolates |
|---|---|
SIRET_55210055400021 | the SIRET, with or without spaces |
SIREN_552100554 | the SIREN, never confused with a SIRET |
IBAN_fr7630006000011234567890189 | the French IBAN, its 27 characters in one block |
TVA_fr40552100554 | the intra-EU VAT number |
DOC_FACTURE | the document type — invoice, credit note, quotation, purchase order |
This pack recognizes nothing approximate, and that is the point: every identifier is recognized at its exact length — 14 digits for a SIRET, 9 for a SIREN, 27 characters for a French IBAN, 13 for an intra-EU VAT number. One digit too many or too few is not a variant to be matched loosely, it is another company or a typing error. On top of those: three document types, recognized in French as well as in English — invoice and credit note, quotation, purchase order.
Three accounting records, as they are
| Indexed reference | What the engine extracts |
|---|---|
| Invoice SIRET 552 100 554 00021 | SIRET_55210055400021 · DOC_FACTURE |
| Quote SIREN 493118373 | SIREN_493118373 · DOC_DEVIS |
| Transfer FR76 3000 6000 0112 3456 7890 189 | IBAN_fr7630006000011234567890189 |
Two spellings, one same company
The same SIRET is written “552 100 554 00021” on one document and “55210055400021” in your accounting export. Both must lead to the same record — and the first 9 digits must never be mistaken for a separate SIREN.
Reliable even when miscopied
The finance pack recognizes each identifier by its own structure, regardless of formatting. A generalist engine would treat these numbers as plain text, even though exact-identifier search is the only use case that matters in this field.
A term specific to your catalog that the pack still misses (an internal document type, for example)? Add your own Custom Rule from the console, simply describing what it should recognize — no regex required.
Index your accounting records
The getting started guide covers the full setup, from sign-up to your first indexed catalog. The documentation details every endpoint.
Watch the breakdown happen. The table above is the engine's output, not hand-written. The same mechanism runs on the home page, with no signup — on a hardware or fashion catalog, not accounting documents. What you see there is the annotation mechanism itself, the one the finance pack applies to your SIRET numbers and document types.