Contract Review: The Process and What It Catches
Reviewed 22 September 2026 against primary sources. The statistics come from 2 studies that can be read free in full, and are quoted with what they do not prove. The English law provisions are cited to the legislation. Where a process guide asserts a timing with no measurement behind it, this page says so rather than repeating it. How we research and correct.
Almost everything written about contract review describes the reading. The reading is not where it goes wrong. The published research on contracting says the failures sit either side of it: context that never arrived with the document, no standard to read it against, no agreed position to fall back to, and nobody owning the dates after signature. This page covers the process end to end, the evidence for where it breaks, and what the review can and cannot settle.
This is general information rather than legal advice. The process described is jurisdiction-neutral. The law is not, and sections 5 and 6 treat US-governed and English-law agreements separately, because the governing law decides a good deal of what your review concludes.
What is a contract review?
A contract review is a structured read of an agreement before you commit to it, aimed at finding what the document actually obliges you to do and what it will cost you if things go badly. For one person with one document, it is a read in the right order. Inside an organization it is a process that changes hands several times, and the difference between those two things explains most of the confusion in this subject.
A review that is doing its job reaches 6 things, in this order:
- What happens when it goes wrong. Liability caps, indemnities, any personal guarantee, governing law and where a dispute is heard.
- How it ends. Renewal, notice deadlines, early exit, and which obligations survive termination.
- Who the parties are. The exact legal entities, checked rather than assumed.
- What the money is. The price, what sits outside it, when it falls due, and whether it can be raised during the term.
- What each side has promised. Specific enough that you can tell when it has been done.
- What the governing law adds or removes. Some terms are unenforceable however clearly drafted, and some rights exist whether or not anybody wrote them down.
Seven stages, and the failure point at each. Four of the seven fail for reasons that have nothing to do with reading.
That order runs backwards through a typical agreement on purpose. The clauses that decide the cost of a bad outcome sit at the back, and by the time an ordinary reader reaches them their attention is gone. Reading the risk and exit clauses first is the single change that improves most people’s review, and it costs nothing.
1. The 7 stages, and who owns each one
The stage names in the figure above follow the shape used across published process guides, and they are broadly agreed. What those guides also carry, and what is worth treating with suspicion, is timings: a first pass in 24 to 48 hours, a detailed review in 3 to 5 business days, and so on. No published study measures those intervals. They are reasonable-sounding assertions, usually appearing on pages that sell software to shorten them, and this page does not repeat them as benchmarks.
End to end is a different matter, and there the measurement exists. APQC’s Open Standards Benchmarking puts the median cycle time to establish a contract with a supplier, from the opening of negotiation to signature, at 40 days across 3,081 organizations. That is a real number from a neutral benchmarking body. What nobody has published is how those 40 days divide between the stages, which is exactly the claim the process guides make.
Two stages are worth expanding, because they are the two most often missing entirely rather than merely done badly.
Stage 1, intake. A document arrives with a deadline and nothing else. Not what the deal is worth, not what was promised verbally in the meeting that produced it, not whether the counterparty is a new relationship or a renewal. A reviewer without that context cannot triage, and triage is the whole value of the review. Fixing intake means asking 4 questions before reading anything: what is this worth, what was promised that is not in writing, when does it have to be signed, and what happens to the business if we walk away.
Stage 7, dates and obligations. Automatic renewal with a long notice period is the most common way an organization ends up locked into another year of something it meant to leave. A clause requiring 90 days’ written notice before the end of a 12 month term puts the real decision point at month 9. That date belongs in a calendar on the day of signature, along with the obligations that survive termination, which commonly include confidentiality, intellectual property assignments and restrictions on soliciting staff or customers.
What to look for: whether stages 1 and 7 exist at all in your process. In most organizations they do not, and no amount of care at stage 3 compensates.
2. Where contract review actually breaks
The best evidence on this is the 2021 EY Law Survey, conducted with the Harvard Law School Center on the Legal Profession, which interviewed more than 2,000 business leaders across 22 countries and 17 industries and drew this report from 1,000 law department and contracting leaders. Harvard hosts the full report free. It is a survey of larger organizations rather than one-person businesses, which is a real limit on reading it across, but it is the largest neutral dataset on how contracting actually runs.
Its headline finding is not about lawyers being slow:
“More than 50% of organizations say inefficiencies in their contracting processes have cost them business.”
EY Law and the Harvard Law School Center on the Legal Profession, The General Counsel Imperative: How does contracting complexity hide clear profitability?, 2021
Six numbers with a method behind them, each with what it does not prove. The 69 and the 71 are the ones that matter.
Three numbers in that figure describe the same underlying problem. 69 percent of organizations do not require contracting staff to use a template or pre-approved model. 71 percent say they do not have the technology to monitor contracts for deviations from standard terms. 75 percent have no pre-approved fallback terms. Put together, that is a reviewer with no standard to read against, no way to tell when a contract has departed from one, and no agreed position to retreat to when the counterparty says no.
One note on that middle figure, because this page audits its own sourcing. The report states it two ways. Its narrative says 71 percent say they lack the technology to monitor for deviations; its infographic compresses that to “71% of contracts are not monitored”, which is a different and stronger claim. We use the narrative version. An organization can monitor without the technology to do it, and the number counts respondents rather than contracts.
That is why the ask gets escalated, why the escalation takes 3 days, and why the business concludes that legal is slow. The bottleneck is an absent standard, not an absent skill. Writing down your positions on a single page, and the fallback you will accept on each, is free, takes an afternoon, and moves more than any tool. Our contract checklist is a reasonable frame to start from, and contract management covers what happens to the document afterwards.
What to look for: whether your organization can produce its standard positions in writing. If it cannot, that is the project, not the review.
3. What the review is protecting
The commercial case for reviewing carefully is usually made with invented numbers. There is a real one. World Commerce & Contracting, working with Deloitte, published The ROI of Contracting Excellence in June 2023, drawing on workshops, interviews, roundtables and online surveys representing 1,236 organizations and collected between April 2021 and December 2022:
“We estimate that average value erosion now stands at 8.6%, with the best performers operating at a little over 3% and the worst more than 20%.”
World Commerce & Contracting with Deloitte, The ROI of Contracting Excellence, June 2023
Read that carefully, because it is routinely misquoted. It is a share of contract value, not of revenue. It is an estimate by the authors rather than an audited measurement. And the spread is the interesting part: the gap between a little over 3 percent and more than 20 percent is the size of the prize, and it is not explained by who has the better lawyers.
Where that value goes is partly predictable. World Commerce & Contracting’s Most Negotiated Terms survey of 21 October 2024 puts limitation of liability and price at the top of its global commercial ranking. We have not verified the full ordering against the report itself, which is available to members, and the sector cuts differ enough to be worth knowing: in the freely published US public sector edition, scope of work and price lead and limitation of liability sits eighth among suppliers. The usable point is narrower than the headline. A term that gets negotiated constantly is one the other side expects to discuss, and which terms those are depends on your sector.
4. Turning findings into asks
A review that produces a long list of comments has usually failed at its last step. A handful of comments reads as a difficult counterparty where 2 or 3 read as somebody who read the document and has a specific concern, and the credibility saved is what you spend on the term that decides whether you sign. That is a judgment about how these exchanges land rather than a measured finding, and nobody has published data on it.
That is a subject of its own, and this site covers it separately: how to review a contract and what to say takes the findings and works through which ones are worth raising, the exact wording to send, and the counter-offer rule that occasionally matters when a live offer is on the table.
What to look for: the one finding that would change whether you sign. Raise that, plus anything free like a typo or an inconsistent date, and let the rest go.
5. If the contract is governed by US law
Three things are worth knowing before the detail, because they change the shape of the review.
There is no single US contract law. Contract law is mostly state law, and the differences are not cosmetic. How long you have to sue, whether a liability exclusion holds, whether an oral agreement is enforceable, and how an automatic renewal must be disclosed all vary by state. A page that gives you one national answer on any of those is guessing. The main federal overlays worth knowing are the Federal Arbitration Act, which preempts contrary state law on the enforceability of arbitration clauses, and the CISG below.
Goods and services run on different tracks. Contracts for the sale of goods are generally governed by Article 2 of the Uniform Commercial Code as enacted in your state, which supplies its own rules on acceptance, warranties and modification. Louisiana is the exception and has never adopted Article 2. Services, real property and employment stay on the common law, and applying a UCC rule to a services agreement is one of the most common errors in general guidance. One more trap: where the buyer and seller have places of business in different countries that have both ratified it, the UN Convention on Contracts for the International Sale of Goods applies automatically as federal treaty law and displaces Article 2 unless the contract expressly opts out, even where it chooses the law of a US state.
The entire agreement clause does real work. Most commercial agreements say the written document is the whole of it, and the parol evidence rule generally keeps a prior assurance out. Whether that also defeats a claim for fraudulent inducement is exactly the kind of question that differs between states, so treat a confident national answer with suspicion.
The US-law material on this site sits in what is a contract, which covers what makes an agreement binding, and in the checklist. For anything where the amount at stake matters, a lawyer admitted in the governing state is the right call rather than a better checklist.
6. If the contract is governed by English law
This section covers England and Wales. Scotland and Northern Ireland have separate legal systems and several of the rules below differ.
Check the party at Companies House, not on the cover page. A group may have a dozen entities and the one on your contract may not be the one with the assets. Check the company number rather than the name, since two companies can have confusingly similar names and only one of them owes you anything. Note that Companies House does not register sole traders or ordinary partnerships. And signing for a company that has not yet been incorporated has a specific consequence under section 51 of the Companies Act 2006: the contract has effect, subject to any agreement to the contrary, as one made with the person who purported to act for the company, who is personally liable on it accordingly. Those five words are the whole of the practical defence, and they are the first thing a careful drafter uses.
A blanket exclusion of liability usually reaches further than the law allows. Under section 2 of the Unfair Contract Terms Act 1977, liability for death or personal injury resulting from negligence cannot be excluded at all, and other loss caused by negligence only so far as the term satisfies the requirement of reasonableness. Two limits are usually skipped. The Act governs business liability. And section 3 does separate work: where one party deals on the other’s written standard terms of business, it subjects to reasonableness any term excluding or restricting liability for breach, whether or not negligence is involved, and any term claiming a right to render substantially different performance or none at all. Since 1 October 2015 section 2 no longer covers consumer contracts, which are governed by the Consumer Rights Act 2015 instead, where the general test is fairness rather than reasonableness. Those are different tests and a good deal of older guidance has it the wrong way round, though the absolute bar survives the move: section 65 of the 2015 Act makes a term excluding liability for death or personal injury from negligence not binding on a consumer at all.
An entire agreement clause does not by itself defeat misrepresentation. To do that it has to go further and exclude liability for misrepresentation or include a non-reliance statement, and under section 3 of the Misrepresentation Act 1967 such a term has no effect unless it satisfies the requirement of reasonableness, with the burden on the party relying on it. Worth knowing that this was genuinely contested until recently: drafters argued for years that a non-reliance clause defined the basis of dealing rather than excluding liability, and so escaped section 3 altogether. The Court of Appeal closed that route in First Tower Trustees Ltd v CDS (Superstores International) Ltd [2018] EWCA Civ 1396. Note too that section 3 does not apply to consumer contracts. A verbal assurance that never made it into the document is not automatically worth nothing.
Deeds and deadlines. Under the Limitation Act 1980 you have 6 years to bring a claim founded on a simple contract and 12 years on a specialty, which includes a contract executed as a deed, unless a shorter period is prescribed elsewhere in the Act. The trap is that a document headed Deed is only a deed if it makes clear on its face that it is intended as one and is validly executed. For an individual, section 1 of the Law of Property (Miscellaneous Provisions) Act 1989 gives two routes: signing in the presence of a witness who attests the signature, or signing at your direction in your presence and that of 2 attesting witnesses. Either way it must then be delivered as a deed. A document called a deed but executed without a witness is not one. If it is otherwise a valid simple contract it carries 6 years rather than 12, and if the deed was used precisely because there was no consideration, it may not be enforceable at all.
Late payment. If you are supplying rather than buying, the Late Payment of Commercial Debts (Interest) Act 1998 entitles you to statutory interest, plus a fixed sum of £40 for a debt under £1,000, £70 for £1,000 to under £10,000 and £100 for £10,000 and above, with reasonable recovery costs above that sum. Two details that guidance usually drops. The Act does not set the rate: the current 8 percent above the Bank of England base rate comes from the Late Payment of Commercial Debts (Rate of Interest) (No. 3) Order 2002. And the base rate does not track changes day to day. It is fixed twice a year, by the rate in force on 30 June for debts falling due in the following 6 months and on 31 December for the 6 months after that. On a small invoice the fixed sum is often worth more than the interest.
7. Can software do the review?
It can do the reading, which was never the bottleneck. Look back at the figure: stages 1, 3, 4 and 7 are the failures, and only stage 3 is reading. Software helps most at 3, 4 and 7, and only where a standard already exists for it to apply.
Two cautions worth carrying. No vendor-independent benchmark has yet put dedicated contract review tools and general AI assistants on the same contracts: the 2 published head-to-head tests, by LegalOn in June 2026 and Ivo in April 2026, were each run by a vendor of one of the products tested. Independent evaluation of the dedicated tools does exist, and is worth reading: the Vals Legal AI Report of February 2025 blind-tested 4 legal AI products on tasks including redlining against a baseline of unaided lawyers, though it did not include a general assistant and discloses a customer relationship with at least one participant. And purpose-built legal AI is not error-free: a 2025 study in the Journal of Empirical Legal Studies found legal AI research tools hallucinating between 17 and 33 percent of the time, on products as they stood between March and May 2024.
If you are choosing, contract review software compared with just using an AI works through whether you need to buy at all, and contract review tools covers the 6 kinds of tool and the question each one answers.
8. What a contract review will not tell you
It will not tell you whether the deal is good, which is commercial rather than legal. It will not tell you how hard the other side will actually push. It will not tell you whether a clause is enforceable where you are, which turns on the governing law and the facts. It will not resolve genuinely ambiguous drafting, which is the point at which a lawyer stops being optional. And it will not make the decision. A review converts an unknown risk into a known one, which is the whole of its job and is worth doing for that reason alone.
The short version
A contract review is a structured read that finds what you are obliged to do and what a bad outcome costs, covering risk and exit first, then parties, money, promises and whatever the governing law adds. Inside an organization it is 7 stages rather than a read, and the research says 4 of them are where it fails: intake with no context, review with no standard, negotiation with no agreed fallback, and nobody owning the dates afterwards. The fix for 3 of those 4 is writing your positions down, which is free. Average value erosion across 1,236 organizations is estimated at 8.6 percent of contract value, and the spread between best and worst is where the argument for doing this properly actually lives.
If a contract is in front of you now, RateMyContract will read it back in plain English so you know what you are looking at.
Where RateMyContract fits in
RateMyContract is a free tool that reads one contract and explains it in plain English, flagging clauses people commonly overlook. On the figure above it sits at stage 3, and only at stage 3, for a reader working on a single document.
It does not do intake, holds no playbook of your positions, does not compare versions, keeps no record, does not diarise your renewal dates, does not negotiate, and does not tell you whether a clause is enforceable under your governing law. It gives no legal advice. It has not been independently benchmarked and publishes no accuracy figure, here or anywhere. If your problem is any of the other 6 stages, it is not the answer to it.
When to talk to a lawyer
Worth the fee where a personal guarantee or an uncapped indemnity is in the draft, where the amount at stake would genuinely hurt, where you are being asked to give up claims, where the drafting is ambiguous enough that two readings are both plausible, and on anything involving real property. Also worth it when you are writing the standard positions in the first place, because a playbook applied consistently applies a mistake consistently too.
Frequently asked questions about contract review
What is a contract review?
A structured read of an agreement before you commit to it, aimed at finding what the document obliges you to do and what it will cost if things go badly. For one person with one document it is a read in the right order, starting with the clauses that describe failure. Inside an organization it is a 7 stage process: intake, a first risk scan, the full clause review, redlining and negotiation, approval, a pre-signature check, and the dates and obligations that have to be diarized afterwards.
How long does a contract review take?
Nobody has published measured stage timings. The figures that circulate, such as a first pass in 24 to 48 hours and a detailed review in 3 to 5 business days, come from process guides rather than studies, usually from pages selling software to shorten them. End to end has been measured: APQC’s benchmarking database puts the median at 40 days from the opening of negotiation to signature on a supplier contract, across 3,081 organizations. What decides that elapsed time is not reading speed but whether context arrived with the document and whether there are agreed fallback positions. For one straightforward agreement read by one person, an hour in the right order gets you most of the way, which is our own judgment rather than a measured figure.
What should a contract review check?
Six things, in this order. What happens when it goes wrong: liability, indemnity, any personal guarantee, governing law and forum. How it ends: renewal, notice deadlines, early exit and what survives. Who the parties actually are as legal entities. The money, including what sits outside the quoted price and whether it can be raised. What each side has specifically promised, in terms you could test. And what the governing law adds or removes whatever the document says.
Why do contract reviews fail?
Rarely because of the reading. The 2021 EY Law Survey with the Harvard Law School Center on the Legal Profession found that 69 percent of organizations do not require contracting staff to use a template or pre-approved model, that 71 percent say they do not have the technology to monitor contracts for deviations from standard terms, and that 75 percent have no pre-approved fallback terms. A reviewer with no standard to read against and nothing agreed to fall back to is working from memory and escalating every ask, which is what the business experiences as legal being slow.
Can AI do a contract review?
It can do the reading, which was never the bottleneck. It is weaker at spotting what is missing, because that needs a reference for what should have been in the document. And accuracy claims in this category should be treated carefully: no vendor-independent benchmark has put contract review tools and general AI assistants on the same contracts, the 2 published head-to-head tests were each run by a vendor of one of the products, and a 2025 study in the Journal of Empirical Legal Studies found purpose-built legal AI research tools hallucinating between 17 and 33 percent of the time on products as they stood in spring 2024.
Does a contract review depend on which law governs the contract?
Yes, and more than most guidance admits. The governing law decides what can be excluded from liability, how long you have to bring a claim, whether a pre-contract assurance survives an entire agreement clause, and which terms will not be enforced however clearly they are drafted. In the United States much of that is state law rather than federal, and for goods the Uniform Commercial Code applies where the common law otherwise would. The process is the same everywhere. The answers are not.
How we checked this page
The survey figures are taken from the 2021 EY Law Survey conducted with the Harvard Law School Center on the Legal Profession, read in the full report that Harvard hosts free. The value erosion estimate is quoted verbatim from World Commerce & Contracting with Deloitte, June 2023, with its own characterization as an estimate preserved, and described by the method the report states rather than as a survey of 1,236 organizations, which is not what it was.
A correction made during review, on our own sourcing. An earlier draft of this page reported that 71 percent of contracts are not monitored for deviations from standard terms, and made a point of the unit being contracts rather than organizations. The report says it both ways: the narrative reports respondents saying they lack the technology to monitor, and an infographic in the same document compresses that to the stronger claim about contracts. We now use the narrative version and say why in the body, because congratulating ourselves on precision while quoting the looser figure would have been the worse failure. The same draft said the blockquote above was verbatim; it is, except that the report prints a typo in that sentence which we have silently corrected and now disclose.
What this page deliberately does not do. It does not repeat the stage timings that appear in vendor process guides, because no study measures them; it gives the end-to-end figure that has been measured instead, and says what that figure does and does not cover. It does not give a single national answer on US contract law, because most of it is state law and a confident national answer on limitation periods or liability exclusions would be wrong somewhere. Where the page offers judgment rather than evidence, on how many comments to raise and how long a single read takes, it says so in the sentence rather than in a footnote.
On jurisdiction. This page separates US-governed and English-law agreements rather than blending them. The English law points are cited to the legislation, with one exception flagged in the text: the treatment of non-reliance clauses under section 3 of the Misrepresentation Act 1967 rests on First Tower Trustees in the Court of Appeal rather than on the words of the section, and the page says so because the drafting argument it closed was live until 2018. On the US side, the two exceptions that a page like this usually omits are stated: Louisiana never adopted UCC Article 2, and the CISG displaces Article 2 on qualifying international sales even where the contract picks a US state’s law.
Sources. EY Law and the Harvard Law School Center on the Legal Profession, The General Counsel Imperative: How does contracting complexity hide clear profitability?, 2021 EY Law Survey. World Commerce & Contracting with Deloitte, The ROI of Contracting Excellence, June 2023. World Commerce & Contracting, Most Negotiated Terms, 21 October 2024, available to members. APQC Open Standards Benchmarking, cycle time to establish a supplier contract. Vals Legal AI Report, February 2025. Unfair Contract Terms Act 1977, section 2. Misrepresentation Act 1967, section 3. Limitation Act 1980, sections 5 and 8. Companies Act 2006, section 51. Law of Property (Miscellaneous Provisions) Act 1989, section 1. Late Payment of Commercial Debts (Interest) Act 1998. Magesh and others, Hallucination-Free?, Journal of Empirical Legal Studies, 2025. Last reviewed 22 September 2026.
