In Depth
Most liability policies carry two limits that work together. The per-claim limit (sometimes called the per-occurrence limit) caps what the insurer pays on any single claim. The aggregate limit caps what it pays for the whole policy period, added up across every claim. When total paid losses hit the aggregate, the coverage is spent for that period, no matter how much room any individual claim still had under the per-claim figure.
The two limits answer different questions. The per-claim limit asks how bad a single claim can get. The aggregate asks how bad an entire year can get. For a risk that tends to produce one large, rare loss, the per-claim limit is what matters. For a risk that can produce many smaller losses in the same period, the aggregate is the binding constraint, because a cluster of ordinary claims can add up to the ceiling even when none of them is large on its own.
Two details often catch buyers off guard. First, defense costs can erode the aggregate. If the policy pays legal fees from the limit rather than on top of it, every dollar spent defending a claim is a dollar less available to pay the next one. Second, a run of related claims from a single underlying problem can march the aggregate toward zero while each individual claim stays well under the per-claim cap.
What It Looks Like
Say a policy has a $1M per-claim limit and a $3M aggregate (illustrative). Over the year the insured has four separate claims that each settle for $900K. Every one fits comfortably under the $1M per-claim limit, but together they come to $3.6M. The aggregate caps total payments at $3M, so the fourth claim is only partly covered and anything past $3M for the year falls back on the insured.
The per-claim limit was not too low here. Each claim sat within it. What piled up was the number of claims, and capping that total is the job the aggregate does.
Why It Matters For AI Vendors
It is easy to shop for coverage on the per-claim limit alone, since that is the headline number on a certificate. But the aggregate is what governs a bad year. Any policyholder whose losses can come in bunches rather than one at a time needs to size the aggregate against the realistic total, not just check that a single claim would be covered. Ignoring it is how a buyer ends up underinsured despite a healthy-looking per-claim limit.