The business says legal is slow. Legal says the business sends everything at the last minute and never uses the template. Both are describing the same defect from opposite ends, and neither will be fixed by hiring. The defect is that almost every in-house legal function reviews contracts uniformly, applying broadly the same process to a routine non-disclosure agreement and to a five-year master services agreement with an uncapped indemnity. Uniform review of non-uniform risk is a queueing decision, and it is the reason the queue exists.
This is an uncomfortable argument because it locates the cause inside the legal team rather than outside it. But the arithmetic is hard to escape. Take a legal function of four lawyers handling, say, nine hundred agreements a year. If the median agreement takes two hours of genuine legal attention plus the handling overhead of finding it, chasing context, following up and re-reviewing after comments, the team is fully consumed before anyone has done a single piece of strategic work. Now observe what those nine hundred agreements consist of. In most companies the large majority are low-value, standard-form, repeat-pattern documents where legal changed nothing at all. The team spent its year producing no change to the risk position on most of its throughput.
The argument in one line
A legal team that reviews everything reviews nothing well. Triage is not a compromise on rigour, it is the mechanism that funds rigour where rigour actually changes the outcome. The objective is not to review less, it is to move attention from the agreements where you are a formality to the ones where you are the control.
Triage by Risk, Not by Order of Arrival
Most legal queues are first in, first out with an override for whoever shouts loudest. That is the worst of both worlds: it is neither fair nor risk-weighted, and it makes turnaround unpredictable, which is the thing the business complains about more than the delay itself. Replace it with routing rules that assess each incoming agreement against a small number of dimensions.
- Contract value and duration. Both matter, and value alone is misleading. A modest annual subscription with a five-year committed term is a larger commitment than a one-off purchase at twice the annual figure. Route on total contract value and on the length of the lock-in, not on the invoice in front of you.
- Deviation from the playbook. The single most useful signal. An agreement on your own template with no changes is a different object from your template with twelve tracked changes, which is different again from the counterparty's paper. Route on which of those three it is before routing on anything else.
- Counterparty type and leverage. A large regulated financial institution or a global software vendor will not negotiate its standard terms, and pretending otherwise burns weeks. A small supplier that needs your business will accept your paper. Knowing which situation you are in should change the process, not just the expectation.
- Liability and indemnity exposure. Uncapped liability, indemnities that reach beyond the contract value, and carve-outs from the cap deserve legal eyes regardless of the deal size. This is the category where a small contract can carry a disproportionate tail.
- Personal data and regulated processing. Where the agreement involves personal data, cross-border transfer, or a sector-regulated activity, the compliance dimension can dominate the commercial one. Route these on the data question even when the money is small.
- Precedent risk. A concession granted to one customer that will be demanded by the next twenty is worth senior attention even at low value. Most-favoured-customer commitments and unusual pricing terms belong here.
The output of triage is a small number of lanes, and the discipline is to keep the number small. Three or four lanes is workable and defensible. Nine is a taxonomy that nobody remembers and everybody routes around.
| Lane | Typical population | Who reviews | Design intent |
|---|---|---|---|
| Self-serve | Company template, unamended, below a defined value threshold, standard counterparty type, no personal data beyond ordinary business contact details. | Nobody in legal. The business owner executes within delegated authority and the record enters the repository automatically. | Remove legal from transactions where it has never changed anything. This lane should carry the largest volume. |
| Guided | Company template with changes confined to a pre-approved fallback set, or counterparty paper that matches a known standard form. | Reviewed against the playbook by legal ops, a contract manager, or a junior lawyer. Escalates only where a change falls outside the fallback positions. | Convert judgement into a checkable rule so that reviewing does not require the most expensive person available. |
| Standard legal review | Counterparty paper, material deviations, moderate value, or a first-time relationship of ordinary size. | The assigned in-house lawyer, working to a published turnaround. | This is where the team's actual legal work should sit, and it should not be crowded out by the lane above. |
| Enhanced review | High value, long term, uncapped or unusual liability, significant data processing, regulatory sensitivity, precedent-setting terms, related party elements. | Senior in-house lawyer, with functional sign-offs from finance, information security or compliance as the risk requires, and external counsel where genuinely warranted. | Deliberate slowness, applied only where the exposure justifies it and defensible when the audit committee asks. |
The question is not how fast legal reviews contracts. It is how many contracts legal should be reviewing at all.
The Playbook That Makes Self-Serve Safe
Self-serve is the lane that frightens legal teams, and the fear is legitimate when it is implemented as permission to sign without a safety net. Done properly it is not the removal of legal control, it is the pre-positioning of legal control. The control moves from the review of each transaction to the design of the instrument the business uses.
Three artefacts carry that weight, and all three have to exist before the lane opens.
A template that is genuinely fit for its transaction
Not the master services agreement with clauses deleted, but a document drafted for the volume use case it will serve, in language a commercial manager can operate without interpretation. Templates that require legal to explain them are not templates, they are drafts. Version them, date them, and make the current one impossible to confuse with last year's.
A clause library with pre-approved fallbacks
For every clause the counterparty predictably pushes on, write the preferred position, one or two acceptable fallbacks, and the point at which the answer becomes no. This is the artefact that converts a lawyer's judgement into something a non-lawyer or a junior can apply consistently. It is also the hardest to write, because it forces the team to articulate positions it has been holding intuitively for years.
A hard stop list with no discretion attached
A short, memorable set of terms that route out of self-serve regardless of anything else: uncapped liability, indemnity beyond a stated multiple, foreign governing law or forum, exclusivity, assignment restrictions that bite on a change of control, personal data processing beyond a defined scope, anything touching a related party. Keep the list short enough to remember and absolute enough that nobody negotiates about whether it applies.
A sampling audit that runs regardless of whether anything looks wrong
Pull a random sample of self-served agreements each quarter and review them properly. This is what converts self-serve from an act of faith into a controlled process, and it is what lets you tell an audit committee that the delegation is monitored. It also surfaces template defects and fallback positions that are being stretched, usually before they become a pattern.
Delegated authority is a corporate governance question, not just a process one
Approval inside your workflow and authority to bind the company are different things, and conflating them causes real problems. A company's signing authority typically flows from board decisions, delegations recorded in resolutions, and instruments such as a power of attorney, and the arrangements differ between companies and can be affected by the sector, the shareholders and the transaction type. Your workflow thresholds should map onto that framework rather than sit alongside it, so that the person the system routes to for final approval is also the person actually authorised to commit the company. Related party transactions and matters requiring specific board or committee attention need particular care. Confirm the position in force for your company with the secretarial and finance functions rather than designing the matrix in legal alone.
Parallel Versus Serial, and Where Time Actually Goes
Once routing is right, the next largest source of delay is sequencing. Most approval chains are serial by inheritance rather than by design: legal, then finance, then information security, then the business head, then signature, each stage beginning only when the previous one ends. If each stage has a two-day turnaround, that is eight working days before anyone has been slow.
Serial sequencing is justified only when one reviewer's output changes what another reviewer is looking at. Finance should not price a deal whose scope legal is about to renegotiate. But information security assessing a vendor's controls and legal reviewing the liability clause are independent activities, and running them in sequence adds days for no informational gain.
Run independent reviews in parallel
Security, data protection, insurance and finance reviews rarely depend on each other's conclusions. Dispatch them simultaneously and let the slowest set the pace, rather than adding the durations together.
Keep dependent reviews serial, and say why
Where a downstream reviewer genuinely needs the upstream output, keep the sequence and make the dependency explicit in the workflow. An unexplained serial step gets treated as bureaucracy and quietly bypassed.
Put final signature last and alone
The authorised signatory should be the terminal step, not one approver among several. Signature arriving mid-chain is how companies end up executing agreements that a later reviewer had not cleared.
Cap the number of approvers per lane
Every additional approver adds delay and dilutes accountability, because each one assumes the others are looking properly. If more than four people are approving a routine agreement, the workflow is expressing organisational anxiety rather than managing risk.
SLAs and Escalation That Actually Fire
Publish a turnaround per lane, and measure it honestly
Publishing a turnaround commitment is the moment a legal team stops being a black box and starts being a service, and it is worth doing even though it feels exposed. The commitment should be per lane, expressed in working days, and measured from intake rather than from the moment a lawyer opens the file.
Escalation is the half that never gets built
The harder half is escalation. Almost every organisation has escalation rules. Almost none of them fire, because escalation depends on somebody noticing that a deadline passed and being willing to raise it about a colleague. Both of those conditions fail routinely, especially where the person sitting on the item is senior.
- Escalation must be automatic and time-based, triggered by the clock rather than by a person deciding to complain. Anything that requires a human to initiate it will not happen when the human is junior to the person being escalated.
- It must go to a named individual, not to a group inbox, and the first escalation should be to the approver's own manager rather than to the general counsel. Escalating straight to the top makes the mechanism feel punitive and drives people to route around it.
- There must be a defined end state. After the second escalation the item either moves forward under a documented exception, is formally rejected, or is reassigned to another reviewer. An escalation ladder with no terminal rung is a notification system, not a control.
- Reassignment on absence must be automatic. Planned leave, extended illness and departures are the commonest cause of stalled approvals, and a workflow that has no delegate configured will simply hold the item until somebody chases.
- Exceptions get recorded, not argued. When the business needs to proceed without a completed review, capture who authorised the exception and on what basis. Over a year the exception log becomes the most informative document the legal function produces, because it shows exactly where the process does not match the business.
Measure the clock from intake, or the numbers are fiction
The most common self-deception in legal operations is measuring turnaround from when a lawyer picks up a request. On that clock a team looks fast while the business waits three weeks for someone to pick anything up. The only honest cycle time starts when the business first asked and ends when the agreement is executed. It includes the days the request sat in a queue, the days spent chasing missing information, and the days lost to a bounced approval. Reporting the flattering internal number to a board is worse than not reporting at all, because it forecloses the conversation about the resourcing or routing that would actually fix the problem. Report the full clock, then break it down by stage so the delay can be attributed honestly, including the substantial share that will turn out to sit with the business rather than with legal.
Break cycle time into three components and report all three: waiting to be picked up, active review, and waiting on the counterparty or the business. In most functions the second is the smallest of the three, which is a useful and slightly deflating discovery for everyone involved in the argument about legal being slow.
The Political Problem Nobody Writes About
Every article on approval design assumes the obstacle is process. Frequently the obstacle is a person, usually a senior and well-regarded one, who has reviewed every contract personally for years and experiences triage as a statement about the value of that work. This is the most common reason a well-designed workflow is agreed in a meeting and never implemented, and pretending otherwise makes the whole exercise naive.
It is worth taking the concern seriously rather than treating it as territorial. The senior lawyer who reads everything is often the only person holding the pattern across the portfolio, catching the third occurrence of a clause that seemed harmless the first two times. When you remove her from the routine lane, that pattern recognition disappears unless you deliberately rebuild it somewhere else.
So design the transition to preserve what is actually valuable. Give the senior reviewer ownership of the playbook and the fallback positions, which is a higher-leverage version of the same judgement applied once rather than nine hundred times. Give her the quarterly sampling audit, which keeps eyes on the population without gating each transaction. Give her the enhanced lane, which is where the interesting problems are. And be honest in the framing: this is not about doing less careful work, it is about the fact that the current arrangement means the highest-value reviewer in the building spends most of her week on documents where she changes nothing.
Sequencing the Change Without Stalling the Business
Do not redesign the whole intake process at once. The change that matters can be sequenced over a couple of quarters, and each stage produces a result you can point at when you ask for the next.
Measure the current state for one quarter
Categorise every incoming agreement by value, paper, deviation and outcome, and record how long each stage took on the honest clock. You need this before you argue for anything, and the distribution will usually make the case for you: most functions discover that a large majority of volume produced no substantive legal change.
Open one self-serve lane, narrowly
Pick the highest-volume, lowest-variance agreement type you have, typically the mutual non-disclosure agreement or a small-value purchase form. Set a conservative threshold. Run it for a quarter with sampling. Nothing convinces a sceptical audit committee like a clean sample and a visible drop in queue length.
Write the playbook for the next two agreement types
Use the deviation data from your measurement quarter to know which clauses actually get pushed on, rather than guessing. Fallback positions written against real counterparty behaviour hold up. Ones written in the abstract get escalated on their first contact with a live negotiation.
Align thresholds with the delegation framework
Work with the company secretary and finance to map each lane onto the authority actually delegated by the board, so that workflow approval and authority to sign are the same decision rather than two parallel ones. Document the mapping, because this is the artefact an auditor will ask for.
Publish turnaround commitments and report against them
Once the volume has moved out of the queue, commit publicly to a turnaround per lane and report performance, including misses. This is what converts the relationship with the business from a complaint into a service conversation, and it is much easier to do after the routing change than before it.
Where My Agreements Fits
None of the above requires software to be true, and a team with strong discipline can run a triage model on a shared tracker for a while. What software changes is whether the model survives a busy quarter, because routing rules that depend on someone remembering to apply them decay in exactly the way every legal team has watched a spreadsheet decay.
CourtMesh's contract lifecycle application, My Agreements, provides the parts that carry this design. Contract intake gives you the single front door that makes cycle time measurable from the honest starting point rather than from when a lawyer opened the file. The pipeline view shows what is in flight and where it is sitting, which is what makes a bottleneck visible while it is still a bottleneck rather than after it becomes a complaint. Multi-stage approvals let you express the lanes as actual routes with recorded sign-off, so who approved what and when is a record rather than an email search. Obligation and renewal tracking picks up the commitments once execution happens, and counterparty management keeps the history of what you have agreed with a given party in one place, which is what makes the guided lane workable.
To be precise about the limits: no system decides which lane an agreement belongs in until you have made that judgement and configured it, no system writes your fallback positions, and CourtMesh does not give legal advice. The design work in this article is yours. What a platform contributes is that the design keeps being applied on the weeks when everyone is busy, which is the only period during which it matters.
Route by risk, and get your senior people back
The in-house legal backlog is usually a routing failure wearing the costume of a headcount problem. Sort the incoming population by value, paper, deviation, counterparty and exposure, let the business self-serve the lane where legal has never changed anything, put a real playbook behind the guided lane, and reserve genuine review for the agreements that carry the risk. Then measure the clock from intake, escalate automatically to a named person, and record every exception. My Agreements gives you the intake, the pipeline view and the multi-stage approvals to run that model, plus obligation and renewal tracking for what happens after signature. The thresholds and the playbook are yours to write. Keeping them applied on a bad week is what the system is for.
Explore CourtMesh
