An intercompany transaction is 1 event that most finance systems record twice and then spend a week arguing with itself about.
Entity A invoices Entity B. Entity A books revenue. Entity B books cost. At the group level neither happened, so somebody has to find both halves, confirm they agree, and write an elimination that cancels them out. When the 2 entities sit on separate ledgers, finding both halves is genuine detective work: different account codes, different timing, different currencies, and 2 people who each recorded their side correctly according to their own books.
None of that difficulty comes from the accounting. It comes from having recorded 1 event in 2 places that cannot see each other.
countries Alva Labs eliminates across on 1 ledger: Sweden, Norway and the UK
of manual consolidation at Oper Credits across 3 entities, now posting as transactions happen
trading entities at Tillo on a single ledger, closing in 5 days across 23 currencies
The month-end reconciliation nobody defends
Ask a group controller what their intercompany process is and the honest version usually involves a spreadsheet with 1 tab per entity, a column for what each side thinks it recorded, and a difference column that is never zero on the first pass.
The differences are almost always mundane. A recharge posted in the wrong month. A management fee booked at 1 rate on one side and a slightly different rate on the other. A cost allocation someone changed in Q2 that nobody propagated. Each takes minutes to find and seconds to fix, and they arrive in bulk.
Intercompany accounting software, as a category, makes that reconciliation faster. It standardises the templates, chases the counterparty, flags the differences earlier. Useful, and entirely downstream of the actual issue, which is that the 2 sides were ever allowed to diverge.
When both sides are the same record
At Alva Labs, consolidation and intercompany elimination run across the Swedish, Norwegian and UK entities on 1 ledger. There is no counterparty confirmation step because there is no counterparty system. Both legs of the transaction are the same record viewed from 2 entities, so they cannot disagree about the amount, the date or the rate. The elimination drafts against the transaction as it posts.
Oper Credits describes the before state precisely. Consolidation across 3 entities took "half a week, a little more," in the finance manager's words. It now posts as transactions happen, with no month-end assembly. The company has gone from 8 systems to 3 in the process.
Tillo runs 4 trading entities across 23 countries and 23 currencies on a single ledger, closing in 5 days against a previous 12. Group entity management, in Harriet Stewart's words, "has saved every member of the team hours and hours."
"Group entity management has saved every member of the team hours and hours."
Harriet Stewart, Head of Business Systems, Tillo
Old model
Record twice, reconcile later
Each entity books its own leg on its own ledger. At month-end someone matches the 2 sides, chases the differences and writes the elimination.
Light's model
Record once, eliminate on post
Both legs are the same transaction seen from 2 entities. The elimination drafts as it posts. There is nothing to reconcile because nothing diverged.
What changes
The difference column disappears
No counterparty confirmation, no rate mismatches, no allocations propagated by hand. The group view is continuously true rather than true once a month.
What this does to the audit
Intercompany is where auditors look hardest, because it is where the group's numbers are most easily overstated and where the evidence is traditionally weakest. A reconciliation assembled in a spreadsheet is evidence of the reconciliation, not evidence of the transactions.
On an immutable ledger the position is different. Every entry, including every elimination, is attributable and timestamped, corrections are new entries referencing the original rather than edits in place, and the auditor pulls 100% of the journal population through the API instead of sampling. Customers with $500M ARR have completed audits on this basis. The intercompany testing stops being an exercise in reconstructing what the group did and becomes an inspection of what it recorded.
The practical read
If a company has 2 entities and 1 recharge a quarter, a spreadsheet is fine and this is not a problem worth solving. The moment there are 4 entities, a couple of currencies and monthly allocations moving between them, intercompany becomes a standing tax on the close: a fixed number of days spent every month proving that 2 systems, which both recorded the truth, recorded the same truth.
The teams that stopped paying that tax did not buy better reconciliation. They stopped generating the discrepancy. Everything downstream, the confirmations, the difference columns, the chase, was infrastructure built to manage a problem that a single ledger does not create.
See how consolidation posts continuously across entities in Light, or book a demo.