A lot of AI tools read a contract and describe it as concerning. LiabilityScore™ produces a number. That difference is deliberate, and understanding how the number gets calculated clarifies what it means.
Most contract review tools, AI-powered or otherwise, return qualitative output: a list of clauses flagged as risky, color-coded highlights, or a summary of things to watch out for. That output is useful. It is also hard to compare, because it does not indicate how risky one contract is relative to other contracts of the same type, or how the risk level compares to the consideration the contract provides.
A score does something different. It compresses a complex set of clause-level signals into a single number on a 0–100 scale, where higher means more risk. The number is comparable across contracts. It is trackable across versions during negotiation. It is shareable with a business partner who does not want to read a 40-page lease. And it is benchmarkable: a score of 72 on a commercial lease means something specific, not just "there are some issues."
The FICO analogy isn't accidental. FICO didn't make lenders fair — it made borrower risk legible, which created competition, transparency, and eventually access. Before credit scoring, people were denied credit based on hunches. Scoring replaced hunches with calculations. LiabilityScore™ does the same thing for contract risk. When enough signers start flagging the same predatory clauses, drafters will remove them preemptively — not out of goodwill, but because friction costs money. That's how markets normalize. Information equality precedes behavioral change.
Before a single clause gets scored, the engine identifies what kind of contract has been uploaded. This matters because a 90-day vendor agreement and a 10-year commercial lease present risk in fundamentally different ways. An indemnification clause in a software license has different implications than the same clause in a personal guaranty.
Type detection runs first, using a fast classification pass that identifies the contract category — commercial lease, residential lease, employment agreement, service contract, NDA, and others. That classification determines which scoring playbook gets applied. Each playbook is calibrated to the risk profile and clause patterns specific to that contract type.
If the engine can't confidently classify the contract, it defaults to a universal analysis that flags the most common high-risk clause patterns across all contract types, and notes the classification uncertainty in the report.
Once the contract type is identified, the full analysis runs. The engine reads the entire document and extracts individual clauses — not just the ones with obvious headings, but embedded provisions, cross-references, and language buried in definitions sections that functions as a substantive clause even if it isn't labeled as one.
For each clause, the analysis answers three questions. First: is this clause present? Second: how is it written — tenant-favorable, landlord-favorable, or balanced? Third: what is the realistic financial or legal exposure this clause creates?
That third question is where most tools stop at "flagged" and LiabilityScore™ goes further. A holdover clause at 150% creates different exposure than one at 200% with "any portion thereof" language and a consequential damages provision. The engine scores the specific version of the clause in the analyzed contract, not the generic existence of the clause type.
Each identified clause contributes to the overall score based on two factors: severity and weight.
Severity measures how risky the specific clause language is on a spectrum from favorable to neutral to concerning to high risk to extreme risk. A holdover clause capped at 110% with a landlord notice requirement scores in the favorable range. The same clause at 200% with "any portion thereof" and no notice requirement scores extreme. The difference between those two versions can represent tens of thousands of dollars in real exposure — the scoring reflects that.
Weight measures how much that clause type matters within the overall contract. A holdover clause in a 5-year commercial lease is weighted more heavily than a holdover clause in a month-to-month residential agreement, because the potential exposure is orders of magnitude larger. Personal guaranty provisions are weighted very heavily in commercial leases because their downside is uncapped personal liability. An arbitration clause is weighted more moderately — it constrains a signer's procedural options, but it does not create open-ended financial obligation.
The weighted severity scores across all identified clauses roll up into a raw score, which is then normalized to the 0–100 scale. That normalization is calibrated against thousands of real contracts to ensure that the score distribution is meaningful: a 30 is genuinely low risk, a 75 is genuinely high risk, and a 90 sits in the band where contracts are commonly reviewed by counsel or substantially renegotiated before signing.
Scores map to five risk bands in the detailed report.
Every contract type has a clause library — the set of provisions the engine looks for, both risky ones that are present and protective ones that are absent. For commercial leases, the full library covers holdover provisions, personal guaranty terms, CAM caps and exclusions, restoration obligations, assignment and subletting rights, force majeure applicability, default and cure periods, rent escalation mechanics, co-tenancy protections, and early termination rights.
Absent protections are scored too. A commercial lease with no co-tenancy clause, no CAM cap, and no audit right isn't missing neutral provisions — it's missing tenant protections that a balanced lease would include. The absence is scored as a risk factor, not a neutral gap.
The score is the headline. The report is the substance.
For every flagged clause, the report includes a plain English explanation of what the clause means, what the realistic worst-case exposure is in dollar terms where calculable, and the negotiated alternatives that are typical for this contract type. Not general suggestions — specific language patterns, specific thresholds, specific dollar caps that negotiated versions of this contract type commonly include.
The goal is that someone who has never read a commercial lease before can look at a LiabilityScore™ report and understand which clauses carry the largest exposure and how a more balanced version of each clause is typically written. A 200% holdover clause with "any portion thereof" language sits well outside the negotiated norm for 5-year commercial leases, and the report says so plainly without requiring a law degree to parse.
LiabilityScore™ provides legal information, not legal advice. It identifies what a contract says and where the risk is. The legal judgment about what to do with that information — whether to sign, whether a specific clause is enforceable in a given jurisdiction, or how a court would rule in a dispute — is the reader's, generally informed by an attorney with knowledge of the specific situation and local law.
For contracts above a certain risk threshold, or where the financial stakes are high enough to warrant it, the report notes that an attorney review is commonly worth the investment. A LiabilityScore™ scan and an attorney review are not substitutes for each other — the scan identifies where the highest-exposure clauses sit in the document, which is the kind of pre-work that commonly makes an hour of billable attorney time more productive.
The free tier includes three scans per month and a full report on each. Upload a contract to see the score, the plain English breakdown, and the negotiated alternatives that are typical for that contract type. It takes 60 seconds.
Related: commercial lease analysis.
Before you sign, get a score.
Upload any contract to LiabilityScore™ and get a 0–100 risk score with a plain-English breakdown of every risky clause — in under 60 seconds.
Scan your contract free →Important
This article is for educational purposes only and does not constitute legal advice. LiabilityScore™ identifies potentially risky contract terms — it is not a substitute for review by a licensed attorney. Always consult qualified legal counsel for advice specific to your situation.