"Off-market" is a term practitioners use constantly and define rarely. It implies that a clause deviates from some baseline of normal commercial practice, but that baseline is almost never made explicit. This post tries to make it explicit.
When we built Clausebeam's deviation-scoring model, we had to operationalize what "off-market" means. The intuitive definition, something like "a clause that is unusual enough to warrant attention," is not precise enough to build a model around. We needed to distinguish between clauses that are unusual because they are genuinely risky, clauses that are unusual because they reflect a legitimately negotiated position, and clauses that look unusual only because the reviewer has seen a limited sample of agreements in this particular category.
The framework we developed is also useful for practitioners doing manual review, as a way of structuring the judgment call that "off-market" actually requires.
Four categories of unusual clauses
When a clause deviates from what you would typically expect, it usually falls into one of four categories. Getting the categorization right changes what you do about it.
The first category is genuinely off-market: the clause creates obligations or risks that experienced practitioners in this contract type would consistently consider unusual and potentially problematic. A liability cap set at 10% of contract value in a SaaS MSA where the prevailing market range is 1x to 3x annual fees falls here. An indemnification clause with no cure period and no cap on the indemnitor's obligations falls here. These are clauses where deviation from market reflects risk allocation that you should consciously examine before accepting.
The second category is sector-specific standard: the clause looks unusual from a general commercial perspective but is standard within a particular industry or contract type. A clause giving a government entity unilateral termination rights without any payment obligation looks alarming in a commercial SaaS context but is entirely standard in federal government contracts where the Termination for Convenience of the Government clause is a regulatory requirement. Before flagging something as off-market, you need to know whether the contract type has its own standard language that differs from general commercial norms.
The third category is negotiated-and-accepted: the clause deviates from standard language because someone negotiated it into this specific agreement, and both parties accepted it knowingly. A liability cap at 3x annual fees instead of 1x is not "off-market" in the sense of being problematic. It is the result of a negotiation that produced a customer-favorable outcome. Deviation in the customer's favor is different from deviation that increases the customer's risk.
The fourth category is drafting variation: the clause looks different from typical formulations but says essentially the same thing. Legal drafting has significant stylistic variation across drafters, firms, and jurisdictions. A clause that achieves the same functional result through different sentence structure or vocabulary is not off-market; it is just differently drafted.
The baseline problem
The fundamental challenge with "off-market" analysis is that any deviation assessment requires a baseline. If you have only reviewed SaaS agreements drafted by mid-size enterprise vendors, your sense of what is "market" for indemnification caps may not reflect what private equity firms see in M&A contexts, or what a government contractor's paper looks like, or what has shifted in the past 18 months of contract practice following increased scrutiny of data breach exposure.
This baseline problem is one reason why volume review is genuinely valuable, not just as a time-saver but as a calibration tool. An attorney who has reviewed 200 NDAs across a range of counterparties and deal sizes has a much more reliable internal baseline for "unusual" than an attorney who has reviewed 20. Clausebeam builds its deviation scoring from a reviewed dataset that spans multiple contract types and counterparty profiles, which is why we can offer a cross-market perspective rather than just a "compared to our firm's standard form" comparison.
The one-sidedness test
One practical heuristic we have found useful for distinguishing genuinely off-market clauses from unusual-but-legitimate ones is the one-sidedness test: does the clause's deviation from standard language systematically favor one party over the other, and does that favoring track with the party who drafted the agreement?
A deviation that favors the drafting party in a way that shifts material risk onto the non-drafting party is much more likely to be an intentional off-market push than a deviation that creates mutual benefit or that deviates from standard in a risk-neutral way. This is not a perfect test. Sophisticated buyers sometimes push for off-market terms that favor them, too. But as a first-pass filter, "who does this deviate in favor of, and is that the drafter?" is a useful question.
Consider a residuals clause in an NDA. Residuals clauses are not standard in bilateral commercial NDAs between parties of comparable sophistication. They are a known carve-out that benefits the disclosing party in one direction (usually the vendor or service provider). When a vendor's form NDA includes a residuals clause, that is a deliberate, vendor-favorable deviation from bilateral NDA norms, and it should be flagged and explained to the client rather than treated as just a "stylistic difference."
Frequency does not equal acceptability
We want to address a conflation that appears in some legal AI tools and in some attorney reasoning: treating "common in vendor paper" as equivalent to "acceptable." The fact that many vendors include a particular provision does not mean that provision is market-standard in the sense of being a reasonable risk allocation. It may mean that vendors have systematically pushed an aggressive position for years and gotten away with it because most recipients either did not read carefully or did not have leverage to push back.
The best example is consequential damages exclusions in SaaS agreements. Many SaaS vendors exclude consequential damages entirely in their standard paper. These exclusions are common. They are also, in many arrangements, a significant one-sided risk transfer: they eliminate the category of damages the customer is most likely to actually suffer from a service failure, which tends to consist primarily of business disruption and lost revenue. Common does not mean fair, and common does not mean acceptable for your specific deal.
Off-market analysis should ask whether a clause creates a risk allocation that is proportionate and commercially reasonable, not just whether you have seen the same provision many times before.
How Clausebeam applies this
Our deviation scoring does not reduce to a single "market/off-market" binary. For each flagged clause, we indicate what the typical formulation looks like for the contract type, what specific element of the reviewed clause deviates from that typical formulation, and in which direction the deviation runs (drafting-party-favorable vs. non-drafting-party-favorable). We also distinguish high-frequency deviations (commonly seen but still worth noting) from low-frequency deviations (genuinely unusual in the review dataset).
This gives reviewers the information to apply the framework above: is this a sector-specific standard I should know about, a successfully-negotiated favorable term, a drafting variation that means the same thing, or a genuinely problematic off-market push? The model provides the structural analysis. The practitioner makes the judgment call. That division of labor is, we think, where legal AI adds value without replacing attorney judgment.
The goal is not to produce a risk score that tells you what to do. It is to make the analysis fast enough that practitioners can afford to actually do it on every clause in a deal room, rather than the ones that happen to get their attention on a busy Thursday afternoon.
