The contract management ownership trap: why procurement versus legal is the wrong question

What happened when one manufacturer changed its ownership model?

A global manufacturer operating across many regions and a large supplier base came to us after two difficult quarters. They had funded contract management through procurement. Intake was clean, onboarding was faster and sourcing discipline had improved. Negotiations, however, were a mess.

First drafts ran well over target. Data processing agreements were triggered only when someone remembered. Legal was buried in statements of work carrying many different indemnity variants, because clause fallbacks had never been encoded. Security reviews were skipped under time pressure.

The diagnosis was organizational, not technical. Decision rights were vague, risk triage happened by searching old emails, and approvals depended on memory rather than rules. When we shadowed live deals from end to end and audited the clause library, the root cause was plain: nobody owned the handoffs.

We did not change tools. We changed the ownership model. Legal took policy, procurement took flow and IT took the platform. Within a quarter, first drafts were faster, high risk deviations on the pilot templates fell, data processing agreements routed automatically and internal audit described the control points as clear. The business stopped calling contract management a bottleneck.

Why does single owner contract management fail?

When procurement leads alone

Procurement optimizes orchestration and throughput, from intake through sourcing, evaluation, award, onboarding and purchase order. The wins are real: standard intake, competitive bidding, faster onboarding and cleaner savings attribution. Where it breaks:

• Negotiation guardrails wobble because deviations are not scored and fallbacks are not encoded
• Privacy, security and data protection reviews depend on someone remembering
• Renewals, service levels and audits drift after signature, and savings are claimed but not evidenced

When legal leads alone

Legal optimizes governance and control: tighter reviews, a smaller risk surface and playbooks that reflect regulation. Signed contracts have fewer defects and better audit trails. Where it breaks:

• Request quality varies because category and risk data are not structured at intake
• Cycle times grow and the business routes around the system
• Adoption stalls because sourcing and onboarding have not changed, so better policy does not reach outcomes

Either way, the result is a local maximum. Speed versus risk becomes a turf war instead of an operating model.

What does shared ownership look like?

The answer is not compromise. It is explicit decision rights that respect each mandate.

All three share responsibility for adoption and outcomes. Disputes go to an executive group of the general counsel, the chief procurement officer and the chief information officer.

Handoffs to put in writing

• Risk triage at intake: category, data sensitivity, spend and geography trigger privacy and security reviews by rule
• First draft: who drafts each contract type and risk tier, and the target time
• Deviation handling: AI deviation scores map to approved fallbacks and route to the right approver by threshold
• Vendor readiness: who owns banking, know your customer and insurance checks, and what ready to transact means
• Obligations register: an owner, a due date and a system of work for renewals, service levels, audits and price protections

Why does AI make this tension sharper?

Much of what is sold as AI for contracts is still better search. Useful, but not enough. AI that reasons over contracts should cross reference master agreements, statements of work and addenda to find compound risk, score deviations against your library and suggest approved alternatives, trigger reviews automatically when risk is detected, and keep the lineage from intake to obligation so savings and risk reduction can be proven.

When AI helps both sides, adoption follows. But AI cannot reason over chaos. If your library holds dozens of indemnity variants with no ranked fallbacks, AI will amplify the ambiguity. In Aavenir, Avy works against the playbook your team defines, and a person approves every result.

What should you decide before turning on AI?

1. Clause library hygiene: consolidate variants, rank fallbacks and tag attributes such as jurisdiction, data sensitivity and liability caps
2. Templates and playbooks: tie templates to use cases such as buy side, sell side, services and software, with stage by stage rules the AI can act on
3. Intake: standardize request forms so category, risk flags, personal data and spend arrive structured
4. Approval matrix: set thresholds within which work can be pre-routed or approved without escalation
5. Obligations: decide what to monitor and where each alert goes

Which metrics actually matter?

Measure flow and risk together, and give every metric an owner and a decision it informs.

• Cycle time (procurement): request to first draft, signature and first delivery, tracked by median and by the slowest cases
• Clause deviation rate (legal): share of agreements with non standard language, weighted by risk
• Competitive sourcing adoption (procurement): share of eligible spend that went through competitive bidding
• Onboarding lead time (procurement): request to a supplier ready to transact
• AI suggestion uptake (legal and procurement): where suggestions are accepted and where they need tuning
• Obligation compliance (business owner and legal): renewals, service levels and audits met on time

Start with three to five, publish them monthly and slice them by contract type, region and deal size. A practical first quarter runs in three phases: baseline and guardrails, a pilot on one high volume template, then a first dashboard and a second template.

What are the common pitfalls?

• Tool first, policy later: fuzzy fallbacks make AI amplify ambiguity
• Hero implementations: if success depends on one power user, it will not scale
• Vanity AI: insights that do not trigger work do not change outcomes
• Siloed wins: savings without guardrails, or control without throughput, both stall

At Aavenir, this is how we connect sourcing, onboarding and contracting on ServiceNow: as one workflow that respects both orchestration and governance. See Contractflow.

The bottom line

Procurement and legal are both right. Contract management works when each owns what it does best, IT owns the platform, and the handoffs between them are written down. Do that, clean up your policy, and AI can finally help you move faster and stay safer at the same time.