Why Contract Review Automation Is Lagging — And What We Can Do About It

Despite years of investment and progress, contract automation is still underachieving.    

We've come a long way in generating contracts quickly from templates. Tools now allow business teams to self-serve NDAs and basic agreements by filling in a simple form—no legal bottlenecks, no delays.    

But when it comes to reviewing, revising, and negotiating contracts—especially at scale—automation remains the exception, not the norm. The real opportunity lies here, yet most enterprises still rely on slow, manual processes.    

The costs? Higher legal spend, sluggish deal cycles, and inconsistent standards.    

The cost of poor and inefficient contract management are real.  World Commerce and Contracting estimated a few years ago.    

• Average cost of poor contracting is 9.2% of a company’s annual income
• Average cost of poor contracting is 15% for large capital projects
• 65% of major projects fail or under-perform, averaging an overrun cost of 80%

The direct and indirect (effect of delays and friction) costs of contract negotiation and management is difficult to estimate but thought to be between 4-6% of the value of each contract.  Its an unacceptably high tax on business.    

Even modest automation could unlock significant savings and promote better visibility of contract terms by data collection during the process.    

So why aren’t we further ahead?    

What’s Holding Us Back?    

An uncomfortable truth in simple terms: the barrier isn’t technology. It’s mindset. More nuanced versions of the truth cite a number of factors.    

1. Over-caution from legal, compliance and risk teams    

Automation of contract review is often treated as taboo. Many lawyers accept AI as a tool for augmentation—but not automation. There’s a deep (and understandable) reluctance to let AI act with autonomy, even when contracts are low-risk and templated.    

2. Lack of outcome-based data    

We don’t yet have enough clear, shared evidence that AI-led contract reviews are as effective—or more efficient—than human lawyers on routine contracts. And without real-world data, many teams simply won’t trust the shift.  Often there is reluctance to establish structures that allow this data to be collected.  How often do decisions get made and perceptions formed on 'gut instinct'.    

3. Bias toward “visible” tech    

Legal functions tend to prefer dashboards, clause libraries, and editing tools embedded in Microsoft Word. But the true efficiency gains come from invisible automation—AI that reviews, triages, and approves contracts before legal ever sees them.    

It’s easy to champion tech you can hold. Harder to rally behind the tech that quietly does the work.    

Where Automation Can Work—Right Now    

Not all contract reiew should be automated. But many low-value, low-risk agreements can be handled faster, better, and cheaper with structured workflows and AI.    

Think:    

• NDAs
• Sub-$75k vendor and customer agreements
• Standard SaaS subscriptions
• Simple consultancy arrangements

A Realistic Workflow for Automated Review    

Here’s what an automation pipeline could look like:    

1. Ingest the contract (via upload/email/API)
2. Enrich with CRM or VRM data (e.g. value, risk, geography, party info)
3. Triage based on value and risk rules.  For example, Low-risk: less than $75k per year, defined rules for low risk delivery (type of goods and services and geography) and rules around document structure (own template, purchase orders or statements of work under existing framework contracts and certain standalone contracts).
4. For low-risk contracts: AI reviews the contract against a simple list of big rocks (potential bombs) - such as unlimited liability, IP transfers, restrictions on business.  The simpler the better. Remember we are talking low value contracts.
5. Ideally the AI fixes the contracts either with a list of comments or a redline.  Or fixes are done by a junior contract specialist.  Or some combination
6. Return to the business team, with a clean copy or comments / mark-up
7. Optional: Human-in-the-loop QA pass.  This is important in building confidence in the AI and can be relinquished at some point.
8. Essential: Data capture. This gives visibility over the terms of contracts automated and enables testing on accuracy to take place.
9. For med-high risk contracts, the automation focuses on directing the contract to the right reviewer (optimising resources) and having the AI review the contract first.

All actions logged to your CLM or legal matter system.    

The end goal? Faster cycle times. Lower cost. Structured data.    

Guardrails, Not Gatekeeping    

Automation doesn’t mean turning off your brain—or your legal function.    

It means defining guardrails in advance:    

• What constitutes big rocks (potential bombs) based on objective data?
• Which liabilities should be capped or limited and which should be potentially unlimited
• Which clauses must be escalated?
• What fallback positions are acceptable?
• How should governing law be handled in different jurisdictions?

These are rules legal teams are already using informally. Automation just makes them codified, repeatable, and measurable.    

Role of AI, technology and testing    

With advances in the capabilities of large language models, there are many of good options available to contract teams to implement a contract review automation workflow.    

But legal and contract teams will only be convinced by a combination of reliable data and a good user experience.    

This means that it is vital that the quality and accuracy of the AI is objectively measured.  Easily done with simple contracts and a simple set of big rocks.  A simple scoring sheet showing error rates could look like this.    

Agentic AI is showing great potential    

Agentic AI is AI that works autonomously after its initial set-up, all within defined parameters and rules.  A key advantage is that it can problem solve around obstacles.    

It makes automation much easier to set-up, run, monitor and adjust.    

At Lexical Labs we have embedded Agentic AI into our software Tiro.  It has the potential to transform the automation use case in the following ways:    

• Identifying gaps in information required for the triage and either retrieving the information or alerting a user.
• Retrieving additional documents necessary to run the workflow. For example, retrieving the Master Services Agreement relevant to a new Statement of Work
• Triggering automated processes as part of the overall workflow.  For example, one might automate a quality assurance process by having one LLM check the work of another.

Benefits    

Once set-up, contract review automation can be instantly transformative.    

Business teams will receive a response in minutes if fully automated or an hour or two if there is still a human in the loop.  As well as accelerating the contract process, it will transform the user experience for the business teams.    

Time and cost saving has the potential to be greater than traditional augmentation AI.    

As an example, lets say 3 out every 10 contracts are low risk contracts and therefore completely removed from the legal in-tray.  Lets also assume that the same AI reviews also the higher risk contracts and saves 40% of the time (as compared to manual review). The overall efficiency gain is 58% compared to manual review of all 10 contracts and a further 18% as compared to AI augmented review.    

The same testing process on accuracy can also be applied to calculating the efficiency gain.    

Data capture and visibility.  Contract data is captured as a digital and actionable asset for future use.    

Wellbeing and morale: there are numerous studies from other areas of enterprises that demonstrate that automation of low value tasks can benefit wellbeing, morale and staff retention.    

What Needs to Change    

To unlock contract automation, we need a mindset shift across legal, procurement, and business teams.    

✅ Move from perfect to acceptable

✅ Treat review as a process, not a craft

✅ Focus on outcomes, not visibility

✅ Embrace systems that keep low-risk work off lawyers’ and specialists desks

✅ Start measuring results    

What to Do Next: Six Steps to Take    

If you’re serious about unlocking value through contract automation:    

1. Start small. Pilot one or two contract types.
2. Segment risk. Define what “low risk” means for your business.
3. Set guardrails. Build rule sets and fallback positions.
4. Track results. Capture turnaround time, error rates, escalations.
5. Enable humans. Build in QA and override steps where needed.
6. Integrate smartly. Use APIs to plug into your CRM and CLM systems.

Final Thought    

We have the tools. We have the need. What’s missing is the will.    

The next leap in legal and contracting efficiency won’t come from more dashboards or fancy redlining tools. It will come from letting go—letting automation take over where risk is low, standards are clear, and the business needs speed.    

This isn’t just a chance to optimise. It’s a chance to rethink what legal work really needs a lawyer.    

‍