A general counsel puts a number in a board pack. Contracts reviewed this quarter: four hundred and twelve. It is a fine number, defensible and easy to produce, and it gets a nod from the chair. Two quarters later the team is quietly optimised around it. Reviews are marked complete a little faster. The lawyer who spent three weeks redesigning the vendor onboarding process so that a category of contract stopped reaching legal at all has, on the reported measure, made the function look worse. Nobody decided this. The metric decided it.
This article is about choosing that number with intent. The argument is simple and it has consequences: a legal team will drift towards whatever its leader reports upward, because that is the only signal most teams get about what the organisation values in their work. So the question is not which metrics are easiest to gather, or which ones other companies use. It is which behaviours you are willing to cause.
You are not choosing a way to describe the team's work. You are choosing what the team will do more of.
Goodhart's Law in a Legal Department
The economist Charles Goodhart is usually paraphrased as observing that when a measure becomes a target, it ceases to be a good measure. In-house legal is unusually vulnerable to this, for three reasons that compound each other.
The first is that legal output is hard to observe directly. Nobody can see the quality of a negotiation from outside it, so proxies get used, and proxies are exactly what Goodhart's law attacks. The second is that legal's most valuable work is preventive, and prevention is invisible by construction. A dispute that never happened generates no data point, while a dispute that is fought generates matter counts, spend figures and outcome records. Measure what you can see and you will systematically reward the department that is busy with problems over the one that stopped them. The third is that a legal team is small enough for an individual to notice their own number, which is where the gaming starts, usually without anyone feeling dishonest about it.
The core of it
Every legal metric has a cheat, and the cheat is usually not dishonest, it is just the rational response to what is being counted. Count contracts reviewed and you get shallower reviews. Count cycle time alone and you get contracts marked complete before they are. Count cases won and you get settlement decisions distorted to protect a percentage. The discipline is to name the cheat when you choose the metric, and pair the metric with something that makes the cheat visible.
The Four Families, and Why You Need All Four
A workable measurement set for an in-house function draws from four families. Any one family on its own distorts, and the distortion is predictable enough to plan around.
Throughput: how much work moved
Matters opened and closed, agreements executed, advice requests handled. Cheap to gather and genuinely useful for capacity planning and resourcing arguments. Reported alone it rewards volume over judgement, which is why it belongs in an internal operating review rather than at the top of a board slide.
Cycle time: how long the business waited
The elapsed time from a request being made to the business having what it needed. This is the metric the rest of the organisation cares about most, and the one most often measured dishonestly. It is also the family where improvement is most attributable to something legal actually changed.
Risk and outcome: what happened as a result
Dispute volume and trajectory, exposure movement, recurring issue types, regulatory findings, matter outcomes against the position taken at outset. The hardest family to instrument and the closest to the reason the function exists. It moves slowly, so it is reported quarterly and read as a trend rather than a score.
Cost: what it took
Internal cost per matter or per agreement, external counsel spend by matter type and firm, spend against budget, and the internal versus external mix. Finance will engage with this family more readily than any other, which makes it the useful entry point for a conversation about resourcing.
The pairing rule matters more than the individual choices. Throughput without cycle time hides a queue. Cycle time without a quality or risk measure invites speed at the expense of care. Cost without outcome makes the cheapest possible legal function look like the best one, which is a conclusion no audit committee actually wants to reach and several have reached by accident.
Vanity Metrics, and What to Use Instead
Most legal dashboards are assembled from whatever the existing systems happen to emit. That is how a function ends up reporting emails answered and documents produced, numbers which are real, easy and almost entirely uninformative. The table below sets out the common ones, a better substitute, and how to instrument the substitute without inventing a data collection burden that nobody sustains.
| Vanity metric | Why it misleads | Better substitute | How to instrument it |
|---|---|---|---|
| Contracts reviewed | Counts touches, not value. Penalises the lawyer who removes a whole category of contract from legal review, which is the highest-value thing anyone on the team can do. | Proportion of agreements executed without legal review, alongside the deviation rate on those agreements. | Requires an intake point and a status field on each agreement recording which lane it went through. No timesheet needed. Pair with a quarterly sample audit of self-served agreements so the quality side is visible. |
| Advice requests answered | Rewards responsiveness to questions rather than the removal of the reason the questions keep arriving. A team fielding the same query forty times looks productive. | Repeat query rate: how often the same question arrives from different parts of the business. | Tag incoming requests against a short, fixed topic list at intake. Ten to fifteen tags, chosen once. A rising tag is a signal to write guidance or change a template, and the drop afterwards is the result you report. |
| Cases won | Selection effects dominate. A team can raise its win rate by settling anything difficult, which is often the wrong commercial decision. It also treats a favourable settlement as a non-event. | Outcome against the assessment recorded at the outset of the matter, plus recovery or exposure reduction relative to the amount claimed. | Record an expected range and a reserve position when the matter is opened, then compare at disposal. The discipline of writing the assessment down at the start is most of the value, independent of the metric. |
| Total external legal spend | Falls when work is deferred and rises when the business does something ambitious. On its own it makes underinvestment look like success. | External spend per matter by matter type, and the internal versus external mix, both read against matter volume and complexity. | Requires matter codes on invoices and a consistent matter type taxonomy. Agree the taxonomy with finance once so that legal and finance are not reporting different totals to the same committee. |
| Turnaround time from when legal opened the file | Excludes the queue, which is usually where most of the delay lives. Produces a flattering number that forecloses the resourcing conversation. | End-to-end cycle time from business request to execution, broken into waiting, active review, and waiting on the counterparty. | Timestamp the request at intake, at first legal action, at each approval, and at execution. Four timestamps, all captured by the workflow rather than typed by anyone. |
| Policies and templates published | Measures production, not use. A library of forty templates nobody can find is worse than six that everyone uses. | Template adoption rate: proportion of executed agreements of a given type that used the current standard form. | Compare the executed agreement against the template version recorded at intake. Falling adoption is an early warning that the template no longer fits the business. |
| Compliance training completion | Measures attendance. Near universal completion is the norm and tells you almost nothing about behaviour. | Incidence and recurrence of the issue the training addresses, tracked by business unit. | Requires an issue log with a consistent categorisation and a unit attribution. Slow to move and worth reporting annually rather than quarterly. |
Notice what the substitutes have in common. Each is instrumented by capturing something at the point work enters or leaves the function, not by asking lawyers to record how they spent their day. That distinction is the difference between a measurement system a team tolerates and one it resents.
Measuring Prevention, Which Is the Hard Part
The most valuable work an in-house team does is the work that stops something. A clause negotiated in year one that caps an exposure in year four. A vendor rejected at onboarding because a litigation and insolvency screen showed a pattern the finance review would never have surfaced. A product launch redesigned so that a regulatory question never arises. None of these produce an event, and events are what data is made of.
You cannot measure prevention directly. You can measure four things that stand in for it honestly, provided you describe them as proxies rather than as proof.
- Dispute origination rate by source. Track new disputes against the contract, product line, business unit or vendor category they arose from. A falling rate from a source where legal made a specific intervention is the closest thing to evidence of prevention that this domain offers. It requires patience, because the lag between the intervention and the trend is often measured in years.
- Recurrence of known issue types. When the same category of problem stops recurring after guidance, a template change or a process fix, that is prevention with a date attached. This is why the fixed topic taxonomy at intake earns its keep.
- Early involvement rate. The proportion of significant transactions where legal was engaged before the commercial terms were agreed rather than after. Legal brought in late is legal reduced to documenting a decision, and the correlation between late involvement and subsequent disputes is something most experienced GCs can already see in their own portfolio.
- Issues caught before commitment. Count the occasions where a review, a diligence step or a counterparty screen changed a decision: a contract not signed, a term renegotiated, a supplier not onboarded. Keep a short log with one line per instance. It is the least statistically rigorous item on this list and frequently the most persuasive in a board room, because each entry is concrete.
How to talk about prevention without overclaiming
Boards are rightly sceptical of avoided-cost figures, and legal functions damage their credibility by producing them. Saying that a clause saved the company a specific sum invites a challenge you cannot win, because the counterfactual is unobservable. Report the intervention and the trend, not a rupee figure: this category of dispute has fallen since we changed the template, here are the numbers, here is what else changed in the same period. A modest claim you can defend outperforms an impressive one you cannot.
Litigation Portfolio Metrics That Are Worth the Effort
Companies with a live dispute portfolio, which in India means most companies of any size, need a distinct set. Indian litigation runs on timescales that make quarterly outcome reporting close to meaningless, so the metrics have to be built around ageing and movement rather than around resolution.
- Portfolio ageing. Matters bucketed by time since institution. The shape of this distribution tells you more about the portfolio than any single number, and a thickening tail of long-pending matters is a resourcing and provisioning signal well before it is a legal one.
- Exposure, stated as a range with a stated basis. Aggregate amounts claimed, the internally assessed exposure, and the movement between them. Be careful to keep the internal assessment distinct from the figures that feed the contingent liability disclosure in the financial statements, which are prepared under the applicable accounting requirements and should be agreed with finance and the auditors rather than lifted from a legal tracker.
- Disposal rate against institution rate. Whether the portfolio is growing or shrinking in matter count. A portfolio growing faster than it disposes needs either more capacity or a different upstream strategy, and this metric is how you make that argument without anecdote.
- Movement, or the absence of it. Matters with no substantive progress over a defined period, whether from adjournments, awaited orders or internal inaction. This distinguishes matters that are stuck for reasons outside your control from matters that are stuck because nobody has looked at them, and only the second category is something you can fix.
- External spend per matter, by matter type and by firm. Read against complexity and stage, not in isolation. The most useful version of this metric is the outlier list, not the average.
- Recovery and enforcement conversion. For matters where you are the claimant, what proportion of favourable outcomes actually converted into recovery. Execution is where value quietly disappears, and a function that reports wins without reporting recovery is reporting half the story.
The Board Pack Versus What You Keep Internal
Not every metric belongs in front of the audit committee, and the failure to separate the two audiences is how legal functions end up either drowning a board in operational detail or telling it nothing useful.
A board or audit committee is asking a small number of questions: is there an exposure that could materially hurt us, is it getting better or worse, are the controls we approved actually operating, and is anything happening that we should know about before somebody else tells us. Give them six to eight items on a single page, consistent quarter to quarter so the trend is readable, with the movement explained in a sentence each. Include the uncomfortable items deliberately, because the credibility of the whole pack rests on whether the committee believes you would tell them.
What goes upward
- Litigation portfolio: matter count, ageing shape, movement in assessed exposure, and any single matter that is individually material
- Regulatory and compliance events: notices received, inspections, findings, and their status
- Contract cycle time against the published commitment, with an explanation of any deterioration
- External legal spend against budget, with the drivers of variance
- Status of the controls the board itself approved, including delegated signing authority and any exceptions granted
- One forward-looking item: the legal or regulatory development most likely to affect the business in the next few quarters
What stays internal
Individual workload distribution, per-lawyer throughput, queue depth by reviewer, template adoption by business unit, and the raw request tags. These are operating instruments for running the team and improving the process. Sent upward they invite the board into personnel management, which serves nobody, and they push the team towards managing their individual numbers rather than the work.
Benchmarking Traps, Particularly in India
Somebody on the board will eventually ask how the legal team's size compares to peers. The honest answer is that ratios of lawyers to revenue or lawyers to headcount travel badly, and they travel especially badly into India.
Three factors distort them. First, the scope of the in-house mandate varies enormously between companies: some Indian legal functions carry company secretarial work, compliance, regulatory affairs, insurance and sometimes contract administration, while others carry none of it. Comparing headcount without comparing mandate compares nothing. Second, the dispute profile is structural rather than chosen. A company in a sector with a high volume of routine recovery, tax or consumer matters will carry a matter count that would look alarming in a different industry and is entirely normal in its own. Third, the balance between in-house and external work is a deliberate strategic choice with a wide defensible range, and a lean team with a large panel budget and a large team with a small one can be equally well run.
If the comparison has to be made, compare the function against its own history and against its own stated objectives, and be explicit about the mandate when presenting any external comparison at all. The most useful benchmarking exercise most Indian GCs can run is against last year, not against a global survey whose respondents do work their team does not do.
Instrumenting Without Building a Surveillance Culture
Here is where measurement programmes get resented, and resentment is fatal because the data collection depends on the goodwill of the people being measured. The instinct, borrowed from private practice, is to introduce timesheets. Resist it in most in-house settings. Timesheets in a department that does not bill produce a large, continuous administrative burden, data of doubtful accuracy because it is reconstructed at week's end, and a culture in which people feel watched. The measurement you need almost never requires them.
Instrument the workflow, not the person
Capture timestamps as work moves between states: request received, assigned, first action, approved, executed, matter opened, matter disposed. These are properties of the work item, generated automatically as the process runs, and they answer most operating questions without anyone recording anything about themselves.
Tag at intake, once, from a short fixed list
Matter type, business unit, and a topic tag chosen from ten to fifteen options. One dropdown at the moment a request arrives. This single field enables the repeat query analysis, the origination analysis and most of the prevention proxies. Keep the list short and stable, because a taxonomy that changes every quarter destroys the comparability that made it worth having.
Measure the team, publish the team, discuss individuals privately
Function-level metrics belong on a shared dashboard where everyone can see them. Individual-level data should exist for capacity and development conversations and should not be circulated. The moment per-person numbers travel upward, people start managing the number, and your data quality degrades in the same week your culture does.
Tell the team what is being measured and why
Explain which behaviours you are trying to encourage and what each metric's cheat is. Teams that understand the intent tend to flag distortions themselves, and a lawyer telling you a metric is pushing them somewhere unhelpful is the most valuable feedback the system can produce.
Review the metric set annually, not quarterly
Metrics need to run long enough to show a trend and to reveal their own distortions. Changing them every quarter guarantees you learn nothing, and it signals that the numbers are decoration. Set them, live with them for a year, then change deliberately with a stated reason.
The failure that looks like success
The most dangerous outcome of a measurement programme is a dashboard everybody trusts and nobody interrogates. Numbers on a slide acquire authority quickly, and by the third quarter people are managing to them rather than through them. Any metric that improves without anyone being able to explain what changed should be treated as a data problem until proven otherwise. Ask the question in every review: what would this number look like if we were quietly getting worse, and would we be able to tell the difference from here?
Where to Start If You Are Five People
Most of this is written for a function with legal operations support. A five-person team in a growing Indian company has no such thing, and the correct response to a comprehensive framework is to ignore most of it. Adopt three metrics, and only three, for the first year.
End-to-end contract cycle time, measured from the business request
This is the metric that changes the relationship with the rest of the company, and it is the one that will reveal where your capacity is actually going. Two timestamps and an intake point are enough to start. Report the median and the ninetieth percentile, because the tail is what the business remembers.
Litigation portfolio ageing and movement
A list of every live matter with its age, its stage, its assessed exposure and the date something last happened. This one artefact answers most of what a board will ask and surfaces the matters that have quietly stopped moving. It is also the foundation for every other litigation metric you might add later.
Request volume by topic tag
One dropdown at intake. Within two quarters it will tell you exactly which template to rewrite, which guidance note to publish, and which business unit needs a conversation. It is the cheapest metric on this list and the one most likely to change what the team does.
Add the cost family in year two, once finance has a matter type taxonomy it agrees with. Add prevention proxies in year three, when you have enough history for a trend to mean something. A small team that measures three things consistently for three years will understand its own function better than a large one that measures twenty things for one quarter and then rebuilds the dashboard.
Where CourtMesh Fits
Almost every metric in this article depends on one thing: work entering the function through a defined point rather than through individual inboxes. Without that, the timestamps do not exist and the tagging never happens, and the measurement programme becomes a manual reconstruction exercise that dies in its second quarter.
My Agreements, CourtMesh's contract lifecycle application, provides that entry point on the transactional side. Contract intake and a pipeline view give you the request timestamp and the stage-by-stage visibility that make honest cycle time possible, multi-stage approvals record where an agreement waited and with whom, and obligation and renewal tracking covers the post-signature commitments that would otherwise never appear in any measure at all. On the disputes and risk side, unified search across the Supreme Court, the 25 High Courts, the District Courts and tribunals, over a corpus of roughly 310 million cases drawn from official government portals, together with counterparty litigation and insolvency screening and a private organisation-scoped watchlist, is what lets a counterparty check become a repeatable step you can point to rather than an ad hoc favour, which is precisely what the issues-caught-before-commitment log needs in order to be credible.
The honest boundary: no platform tells you which metrics your board should see, and none of this substitutes for the judgement about what your function is for. CourtMesh does not give legal advice. What it contributes is that the underlying events, when a request arrived, where it waited, what was committed, what is pending against a counterparty, are captured as a by-product of doing the work rather than as an additional reporting task that competes with it.
Choose the number before it chooses your team
Whatever a general counsel puts in the board pack becomes the department's real objective within two quarters, so pick it deliberately. Drop the counts of contracts reviewed and requests answered, measure cycle time from when the business asked rather than from when legal opened the file, track the litigation portfolio by age and movement instead of by wins, tag incoming work so you can see what keeps coming back, and describe prevention as a trend rather than as a rupee figure you cannot defend. Instrument the workflow instead of the person, keep individual numbers inside the team, and leave the set alone long enough to read a trend. CourtMesh gives you the intake, the pipeline, the approvals and the counterparty and litigation visibility that generate those numbers as a by-product of the work. Deciding which ones deserve a board's attention is the part that stays yours.
Explore CourtMesh
