Reference · ArbeitszeitCheck
What our DATEV export actually produces
Payroll offices ask for "the DATEV file" as if it were a single thing. It is not. Here is the file ArbeitszeitCheck produces, byte for byte, and what your payroll office still has to do with it.
The file
The export is a plain text file. One line per employee and wage type, pipe-separated:
Personalnummer|Datum|Lohnart|Menge|Einheit|Text
00000042|31.10.2026|1234|160.00|Stunden|Arbeitszeit Oktober
00000042|31.10.2026|5601|8.50|Stunden|Sonntagszuschlag
That is the whole file. No XML envelope, no certificate exchange, no API handshake, this is the import format DATEV Lohn und Gehalt has accepted for years, and it is the format your payroll office already knows how to feed into their system.
Where the numbers come from
Each employee needs a Personalnummer, the eight-digit, zero-padded number DATEV uses to route hours to the right payroll record. You set it once per user in the user settings. Employees without one are skipped in the export, and the export tells you who was skipped; nobody silently disappears.
The Lohnart is yours to decide. Regular hours go to whichever four-digit Lohnart your payroll setup uses (1-8999). Premiums, Sunday, night, holiday surcharges, are a separate matter: the admin maps each premium bucket to the Lohnart your payroll office assigned for it. Nothing exports under a Lohnart you did not explicitly map. If a bucket has no mapping, those hours simply do not appear, which is deliberate: a missing line in a review is cheaper than a wrong Lohnart on a payslip.
The part people underestimate: closed months
When a month is closed in ArbeitszeitCheck, the premium calculation is frozen into a snapshot. The DATEV export for a closed month reads that snapshot, not the live rules. If you change your surcharge policy in December, the October export still reports October under October's rules. A sealed month stays sealed. Your payroll office gets the same file whether they ask for it in November or in March.
This also means the correction path is real: reopen the month, fix the entry, close it again, re-export. The file reflects the corrected closure, not the first attempt.
What the file does not do
It does not transmit itself. There is no DATEV data-center connection, the file downloads to your machine and travels to your payroll office however you normally exchange files with them (email, DATEV's own upload channels, your tax advisor's workflow). It also does not carry master data: no addresses, no tax classes, no bank details. It carries hours and wage-type codes, which is exactly the slice of the problem a time-tracking app should carry.
And it is not a substitute for the conversation with your payroll office about which Lohnarten they actually want. That conversation happens once; the mapping screen exists so you only have it once.
The version for procurement
If your payroll office asked whether your time tracking "supports DATEV": yes, as an import-ready export with per-user Personalnummer, admin-mapped Lohnarten, and sealed-month snapshots. If they asked whether it talks to DATEV directly: no, and neither do most tools at this size. The file is boring by design, boring files are the ones that survive audits.