By Yair Knijn · December 10, 2025

One policy, four legal entities: the intercompany allocation that failed the audit

The controller of a four-company group put every van, lease car, and pool vehicle on one master fleet policy. One premium, one renewal, one broker contact. At month-end she recharged the cost to the four operating entities by headcount, because headcount was the key already sitting in the cost-allocation model. Clean, defensible, done. That was the wrong assumption: headcount was never the thing the insurer priced.

The external auditor did not argue about the headcount split in the abstract. He asked one question that the single policy could not answer: which entity bore which risk, and did each entity pay for the risk it actually carried? There was no per-entity trace of vehicles, premium, or claims. The allocation was a number with nothing underneath it.

Why group fleets pool cover and why auditors push back on the allocation

Pooling makes sense. One policy across the group buys a better rate, removes duplicate admin, and lets you negotiate as one fleet instead of four small ones. The premium discount is real and the controller was right to chase it. None of that is the finding.

The finding is the recharge. The moment a cost paid by one legal entity is pushed onto three others, you are inside intercompany territory, and the auditor stops treating it as a procurement line and starts treating it as a related-party transaction. He wants to see that each entity paid for the cover it consumed, not a flat per-head average that happens to balance to zero across the group.

The arm's-length problem: premium and claims that don't follow the risk

The OECD and the IRS apply the same arm's-length standard to insurance cost shared inside a group that they apply to any related-party charge: the entities that benefit from the coverage should pay for it, and the allocation has to reflect actual benefit and risk, not administrative convenience. A common transfer-pricing position is that once a premium is set on an arm's-length basis, it still has to be allocated among the group members on an arm's-length basis. If premium is allocated by a key that ignores the risk, the deduction each entity took for its share can be challenged, and in the harder cases disallowed.

Headcount fails this on contact. The entity with twelve office staff and two cars was paying more premium than the entity with four staff running thirty delivery vans through city centres at 3 a.m. The vans crashed; the office cars did not. Premium flowed away from the risk and claims flowed toward it, and nothing in the recharge connected the two.

Transfer-pricing and intercompany exposure most fleet teams never model

Fleet teams think in vehicles and renewals. The exposure here is a tax and audit one, and it compounds quietly:

• Premium deductibility, per entity. A recharge that does not track risk lets the tax inspector argue an entity over- or under-paid, and reprice it.
• Claims attribution. A €40k bodily-injury claim sits in the loss ratio of whichever entity actually owned and operated the vehicle. If your records cannot say which entity that was, the claim lands wherever the spreadsheet guesses.
• The documentation burden. The standard the auditor and the inspector both want is a contemporaneous trace showing how the arm's-length share was derived. Reconstructing it a year later, from one undifferentiated policy, is the part that fails.

Per-entity segregation that survives both the auditor and the tax inspector

The fix is not a fifth tab on the allocation model. It is keeping the policy pooled while the data underneath it is segregated by legal entity from the day each vehicle goes on cover. Every vehicle carries its owning entity. Premium is allocated on a key that reflects exposure, vehicle count, vehicle type, and usage, not headcount. Claims post back to the entity that operated the vehicle. When the auditor asks which entity bore which risk, the answer is a report, not a forensic exercise.

In FleetLedger, each legal entity is its own segment inside one fleet program, so the group keeps the pooled rate while premium and claims stay traceable per entity, with the registration data behind every allocation. The audit question stops being a finding and becomes a query. See how the per-entity fleet program works.