Consequential damages and carve-backs: a reviewer’s checklist

What does “consequential” actually exclude? Ask three reviewers and you will get three answers, two of them confident. The clause appears in almost every MSA, SPA and licence we see, usually in a form copied from a document that was itself copied, and it is frequently read as a blanket exclusion of lost profits. It is often not that.

This guide is for the reviewer with a draft open and twenty minutes: what the exclusion catches, what a careful counterparty carves back, how the clause interacts with the cap, and how Review flags each pattern.

What the exclusion actually catches

Under English law, and in jurisdictions that follow it, the courts have historically read “consequential” or “indirect” loss narrowly, as the second category of loss in the classic two-limb test: losses that do not arise naturally from the breach but were within the parties’ contemplation because of special circumstances known at the time of contracting. Loss that flows directly from the breach in the ordinary course of things is first-limb loss, and an exclusion of “consequential loss” does not touch it.

The practical consequence is that lost profit is often a direct loss. If a supplier fails to deliver and the buyer loses the margin it would have made on resale, that is, on the orthodox reading, direct. A clause that excludes “indirect or consequential loss” may leave it fully recoverable. Drafters who want to exclude lost profit have to say so expressly, and most modern forms do: “loss of profit, loss of revenue, loss of business, loss of goodwill, whether direct or indirect”.

Two caveats. Some courts have signalled a willingness to read these words in their ordinary commercial sense rather than through the two-limb gloss, so the position is less settled than it was. And US state law varies; several states treat consequential damages as a broader category. Review resolves this by reading the exclusion against the governing-law clause, not in isolation, and the flag text says which reading it applied.

The carve-back list

A well-drafted exclusion is followed by a list of things that are not excluded. These are the carve-backs, and they are where most of the negotiation lives. The list a reviewer should expect to see, and ask about if absent:

• Liability that cannot be excluded by law: death, personal injury caused by negligence, fraud, fraudulent misrepresentation. A clause that purports to exclude these is unenforceable in most jurisdictions and signals a careless draft.
• Breach of confidentiality. Loss from a confidentiality breach is almost always indirect in nature, so without a carve-back the exclusion swallows the remedy.
• Data protection obligations, especially where a DPA sits alongside the main agreement and imposes indemnities that the main exclusion would otherwise hollow out.
• Indemnities for third-party IP claims. Defence costs and settlement sums are typically consequential in form.
• Wilful default or gross negligence, where the governing law recognises the distinction.
• Payment obligations. A customer should not be able to characterise unpaid fees as the supplier’s consequential loss.

Each carve-back can be taken out of the exclusion only, or out of both the exclusion and the cap. That distinction matters more than the list itself.

Interaction with the cap

The exclusion says what kinds of loss are recoverable. The cap says how much. A carve-back from the exclusion that does not also carve back from the cap means the loss is recoverable but only up to the cap. A carve-back from both means it is recoverable without limit.

Reviewers miss this because the two provisions sit in different sub-clauses with slightly different carve-back language. A common pattern across our active accounts: confidentiality is carved back from the exclusion in 7.2 but not from the cap in 7.3, which the drafter may or may not have intended.

Playbooks handle this in one of three ways. Some firms require a single consolidated carve-back list that applies to both provisions. Some accept different lists but require the difference to be deliberate and recorded. Some set a “super-cap” for the carved-back categories, a higher ceiling rather than no ceiling. Review reads whichever rule the playbook encodes and flags deviations at the severity the playbook assigns.

How Review flags each pattern

Since Review 4.1, the limitation-of-liability module reads the exclusion, the cap and the carve-backs as one structure. The flags a reviewer will see, with default severities before playbook tuning:

• Exclusion with no express reference to lost profit, where the playbook expects one: Medium, with a note on the governing-law reading applied.
• Carve-back list missing a category the playbook treats as mandatory: High for fraud, death and personal injury; Medium for confidentiality and data protection; Note for the rest.
• Carve-back present in the exclusion but absent from the cap, or vice versa: Medium, with both sub-clauses cited side by side in the trace.
• Exclusion that purports to exclude non-excludable liability: High, with the proposed redline deleting the offending words rather than the whole clause.
• Governing law for which the exclusion language is ambiguous: Note, suggesting express wording.

Each flag carries a citation to the playbook rule and, where relevant, to the firm’s own precedent in Recall. Since Review 4.2, where the clause matches the playbook position exactly, Review says nothing.

The checklist

Before closing the document:

1. Read the exclusion against the governing law. Does “consequential” mean what the drafter thinks it means there?
2. Is lost profit addressed expressly, and in which direction?
3. List the carve-backs. Compare the list to the playbook’s mandatory set.
4. Check whether each carve-back applies to the exclusion, the cap, or both. Confirm that any difference is intended.
5. Confirm nothing non-excludable is purportedly excluded.
6. If a DPA or separate indemnity sits beside the agreement, check which document’s limitation regime prevails.

Six questions, five minutes, and the clause most likely to be litigated in the whole agreement has been read rather than recognised.