Written by Marijn Overvest | Reviewed by Sjoerd Goedhart | Fact Checked by Ruud Emonds | Our editorial policy

How to Use Claude for Contract Analysis in Procurement

The contract review problem procurement keeps deferring

In conversations with procurement leaders across Europe, the Middle East, and North America over the past twelve months, a pattern keeps repeating. The contract gets signed. It goes into a folder on SharePoint, or a contract management system, or, more often than anyone admits, a shared drive nobody has catalogued in three years. Six months later, a category manager asks: "what does the contract actually say about that?" and nobody can answer without an hour of reading.

This is not a legal problem. It is a commercial problem. The penalties that were negotiated carefully at signature are sitting in the contract uninvoked because nobody has read it against actual supplier performance. The service-level credits that should have triggered three quarters ago have not been claimed. The price escalation cap is indexed against a commodity that has moved, and nobody has recalculated what the current ceiling actually is.

Procurement Tactics' 2026 AI Readiness in Procurement survey found 40% of procurement teams spend 60% or more of their week on manual data work and reactive firefighting. The contract review that should happen every quarter is the work that gets displaced by that firefighting. It is not that procurement teams do not know it matters. It is that they do not have the hours.

Claude changes the time cost of a rigorous contract review from a day of deep work to under half an hour. That shift turns the quarterly review from aspirational to realistic, and it is usually the first procurement workflow where an AI tool earns its place, because the commercial findings tend to pay for the whole AI investment several times over in a single review.

What a good contract review actually looks like

A review is not a summary. A summary tells the procurement team what the contract says. A review tells them what the contract says relative to what is actually happening, and where that gap creates leverage.

The structure that consistently produces usable findings has three sections.

Section 1, Contract summary

This is the low-value section. It lists the commercial essentials: supplier name, contract value, contract term, renewal mechanism, key pricing clauses, key service levels. The procurement team already knows most of this. Its purpose is reference, not analysis.

Section 2, KPI scorecard

This is where the review starts to produce value. For every contractual KPI, on-time delivery, quality, service-level response times, fill rates, defect rates, the scorecard shows the contractual target, the actual performance over the measurement period, the variance, and the status. Amber if within tolerance; red if consistently below target for multiple periods.

The status column is what drives the next section. A KPI that has been red for three consecutive quarters is not a management issue; it is a commercial one. The contract almost certainly contains a remedy, a rebate, a credit, an escalation path, that has not been invoked because nobody has connected the performance data to the contract language.

Section 3, Improvement log

Every clause creating risk or leverage gets a row: what the clause says, what the commercial or operational exposure is, the recommended action, and a priority level. A force majeure clause that is too broad. A liability cap with no consequential loss provision. A price escalation indexed to a commodity that has moved twenty percent in the supplier's direction. An audit right that expires before the renewal. Each one gets written up in terms a procurement professional can take to a supplier or a renewal conversation.

The review is commercial, not legal. A lawyer reviewing the same contract would focus on different exposures. That is a feature, the procurement team does not need a legal review every quarter. They need a commercial one, and they need it in a format they can actually use.

A worked example: the €207,000 sitting inside one contract

To make this concrete, consider an anonymised example drawn from the Claude Cowork course material, a procurement team managing a packaging supplier relationship. Annual contract value: €3.8 million. Contract term: 24 months, with renewal in six months. Primary commercial terms: 95% on-time delivery target, 0.5% rebate per percentage point below target, 90-day payment terms, volume commitment floor.

Actual performance over the preceding twelve months: on-time delivery averaging 84%. Quality performance within tolerance. Fill rate acceptable. Service levels acceptable. One commercial KPI, the OTD, has been running roughly eleven points below target for three consecutive quarters.

Claude's review flags this immediately. The contract contains a rebate clause for OTD underperformance. Eleven points below target, across the twelve-month period, against annual spend: a valid claim for just over €207,000 in rebates. The claim has never been invoked.

That finding does not usually emerge from a casual read of the contract. It emerges from a disciplined comparison of actual performance against contractual terms, which is what the framework forces Claude to do. The procurement team now walks into the renewal conversation knowing exactly what the supplier owes. Whether they claim the rebate, use it as negotiation leverage, or agree a forward remediation plan is a commercial decision. But the conversation is on different ground than it would have been.

In my experience running this exercise with procurement teams across industries, a six-figure finding in the first contract review is common. It is not always there, some suppliers genuinely deliver against their KPIs, but the frequency is high enough that the first review typically pays for every other piece of AI work the procurement team will do that year.

The three inputs Claude needs

The quality of the review is determined by the quality of the inputs. Three files are enough.

The contract itself. PDF is fine; a scanned image-only PDF works less well than a text-searchable PDF. If the contract includes addenda, amendments, or side letters, include them. One missed amendment can invert the interpretation of a clause.

The supplier performance scorecard. Twelve months of contractual KPI performance in a table. Monthly rather than quarterly if the data is available, because the review catches trends that quarterly aggregation hides. Without this data, Claude can produce a contract summary and a clause-risk review, but the KPI scorecard becomes speculative, and that is where most of the commercial findings come from.

A short company briefing. A paragraph on the procurement organisation, the category, the spend context, and the commercial priorities. The briefing is what tells Claude whether the review should emphasise cost recovery, risk reduction, operational reliability, or negotiation leverage. The same contract produces different review emphases depending on what the procurement team actually needs from the review.

A fourth input is optional but useful: the active supplier list showing qualified alternatives. If an alternative supplier has been qualified through a recent RFI, the review can include a leverage section noting where the supplier's performance gaps create opportunities for dual-sourcing or migration.

The clarifying questions Claude asks, and why they matter

A well-designed review prompt instructs Claude to ask clarifying questions before drafting. In practice, four typically come back, and the answers change the output.

What is the performance measurement period? Claude needs to know whether the KPI scorecard is for the most recent contract year, the preceding twelve months rolling, or a specific quarter. The answer determines what "current performance" means in the review.

How does the team define material breach? The threshold for escalating a KPI failure to a contractual remedy is organisation-specific. Some teams treat one quarter below target as material; some require three. The answer shapes which clauses the improvement log flags as action items.

Are there side agreements or amendments not included in the master contract? Suppliers frequently negotiate commercial side agreements, volume discounts, service-level waivers, relationship-specific pricing, that are not in the main document. Missing these can invert the review's conclusions.

What is the primary objective: cost recovery, risk reduction, or renewal leverage? The same contract produces different reviews depending on the objective. A cost-recovery review focuses on unclaimed rebates and credits. A risk-reduction review focuses on liability, indemnity, and termination. A renewal-leverage review focuses on the commercial gaps that can be renegotiated. All three are valid; they are not the same output.

Procurement teams skipping the clarifying exchange, pasting a generic prompt and accepting whatever Claude produces, end up with a review that is technically accurate but commercially unfocused. The five minutes spent answering the clarifying questions is usually the highest-leverage five minutes of the entire exercise.

What the contract review is not

Two important limits.

The review is not a legal analysis. It is a commercial one. A contract review produced by Claude is a useful input to a legal conversation, it frames the commercial exposures and the leverage opportunities, but it does not replace a lawyer's review of the same contract. For high-value or high-risk contracts, the sequence should be: Claude review → procurement action plan → legal review for the items that require a legal perspective. Not the other way around.

The review is not a substitute for the negotiation. The procurement professional still runs the renewal conversation, the remedy claim, or the relationship escalation. Claude produces the analysis and the evidence. The decisions about what to do with them, when to push, when to concede, how to frame the supplier conversation, are commercial judgements that sit with the procurement team. The teams getting the most value from this workflow use the review to prepare for decisions, not to automate them.

Teams that want to build this as a repeatable capability across a procurement function, not just a one-off exercise for a single contract, tend to embed it inside a broader programme of AI skills development. The AI Fundamentals for Procurement Teams program covers the prompt engineering, template design, and workflow structuring that turns a single contract review into a quarterly rhythm across the category portfolio.

From one contract review to a quarterly rhythm

The highest-value version of this workflow is not a single contract review. It is a repeatable one, every major supplier contract reviewed at a set cadence, with the findings tracked across reviews.

A quarterly cadence is usually right for top-tier suppliers (the top twenty by spend or the top ten by strategic importance). A semi-annual cadence works for tier-two relationships. Below that, a review at renewal plus one review halfway through the contract term is usually enough.

The practical sequence: create a Project in Claude for each major supplier. The Project contains the contract, the scorecard template, and the review template. Each quarter the scorecard is updated with fresh performance data, and the review is re-run. The Project accumulates context, previous findings, actions taken, outcomes, which means every subsequent review is sharper than the last.

Over four quarters, the Project becomes a living record of the commercial relationship. It is the version of supplier relationship management that most procurement teams know they should have and rarely achieve, because the manual version takes too long. The AI-assisted version takes a few hours a quarter. That is the shift.

Frequently asked questions

How long does a contract review with Claude take?

Fifteen to twenty-five minutes end-to-end for a standard supplier contract, assuming the three inputs (contract, scorecard, briefing) are prepared. Longer contracts with many amendments take proportionally longer, but rarely more than forty-five minutes.

Can Claude review contracts in languages other than English?

Yes. Claude handles the major European and Asian business languages well. For contracts in less common languages, the quality drops somewhat but remains useful as a first pass; a local reviewer should validate key findings.

Is it safe to upload a supplier contract to Claude?

For most commercial contracts, yes, on a paid Claude plan with the standard enterprise data-handling terms. Contracts containing regulated data, third-party confidential information without supplier consent, or information subject to specific data-residency requirements should be reviewed against your AI policy before upload.

What if the contract is in a scanned PDF that isn't searchable?

Run the PDF through an OCR tool first. Claude can read image-only PDFs with lower accuracy, but OCR takes a few minutes and significantly improves the review quality.

Should legal review happen before or after the Claude review?

After. The Claude review frames the commercial exposures and leverage opportunities; legal review then focuses on the specific items that require a legal perspective. Running legal first wastes legal time on aspects that procurement can assess directly.

How often should a contract be reviewed?

Top-tier strategic suppliers: quarterly. Tier-two: semi-annually. Tier-three: at renewal and mid-term. The AI-assisted review makes quarterly realistic in a way the manual version never was.