Most writing about legal AI is aimed at people who are already deep into it. That is not where most legal teams are. Most are at the very start, fairly sure they want a tool, and trying to work out how to pay for it. Some are also carrying scars from the last time. They bought a CLM, or a contract tool, or a workflow platform that was sold on a similar promise and never quite delivered, and now they are bracing to ask for budget again, knowing the CFO remembers how the previous bet turned out. If that is you, on either count, this is about the case you are about to build.

I should be upfront that I build and sell legal technology, so I have an interest in you buying it. What follows is still the most useful thing I can tell you, which is how to make the case on a number you can actually stand behind rather than the one that is easiest to reach for.

Headcount reduction is the obvious lever, and usually the wrong one

When you take an AI tool to finance, the cleanest answer to “what is the return” is headcount. If the tool does the work, you need fewer people, and the saving is the salaries you stop paying. It is the answer CFOs are most used to hearing, and the one that unlocks money fastest. So it is tempting to build the whole case on it.

For most legal teams, that case will not hold, and it is worth understanding why before you stake your credibility on it.

When a city widens a road to relieve congestion, traffic eases for a while. Then the easier road draws more drivers, the new capacity fills, and congestion returns at a higher volume. Transport economists call this induced demand. Legal review behaves the same way. When review is slow, the business is naturally cautious about what it sends, and a fair amount of work gets handled informally or deferred. As review gets faster, that caution relaxes. Work that was being kept away from legal starts flowing back, and matters that were never worth the wait become worth doing.

The result is that your team becomes more productive per person, and the organization tends to spend that productivity on handling more work rather than on doing the same amount with fewer people. The saving you projected as removed salaries reappears as expanded capacity. The headcount you planned to cut stays roughly where it is.

This does not make the tool a poor investment. It makes headcount reduction the wrong number to build the case on. The better news is that there are several stronger numbers, and which one fits depends on your organization.

The savings that actually fit your organization

There is no single business case for legal AI, and that is the bit most people get wrong. The right case depends on the shape of your team and your business, and the strongest version is usually built from one primary number, with the others adding weight behind it. Here are the angles worth considering, roughly in order of how concrete they tend to be, along with how to size each one before you walk in.

Avoided hiring, if your organization is growing. If demand on your team is rising and your current model means meeting it by hiring, the honest version of the headcount argument is avoidance rather than reduction. The tool lets you absorb a materially larger volume of work without the lawyers you would otherwise have had to add. To a CFO, avoided cost is real cost, and this saving is strong precisely because it asks nobody to leave.

The way to make it credible is to anchor it to a growth driver the business already forecasts. Your legal workload does not rise on its own. It tracks something, usually deal volume, company headcount, revenue, or contract count. Find the metric your workload follows, show how it has grown, and show how your team has grown alongside it. That ratio is your hiring curve. If the business plans to keep growing at the same rate, you can project the lawyers that growth would otherwise require, and the saving is the slice of that curve the tool lets you avoid. A CFO can argue with a vague claim that you will need fewer people. It is much harder to argue with a line like “we have added one lawyer for every ten million in revenue for three years, the plan adds forty million, and this lets us serve it with the team we already have.”

Expect the obvious objection, which is how the CFO knows you would have hired anyway. Pre-empt it by showing the strain that is already there: the backlog, the response times, the work already going to external firms because there is no internal capacity. The hiring was coming. The tool changes whether it has to.

Reduced outside counsel spend, in almost any organization. When your team cannot handle the volume, work goes to external firms, and that spend is already a visible and often painful line on the budget. A tool that lifts internal capacity lets you keep more of that work in house, and the saving lands against a number finance already tracks and already wants to bring down. It does not depend on your organisation growing, which is why it is frequently the strongest case of all.

To size it honestly, pull your external invoices for the last year and sort them by why the work went out. Some of it went out for genuine specialist expertise or contentious matters, and that work is not coming back. An AI tool does not replace your litigation counsel or your specialist tax advice, and claiming it will only damages your credibility. But a surprising share of external spend is high-volume, lower-complexity work that left simply because you did not have the hands: routine reviews, standard agreements, overflow during busy periods. That is the addressable portion, and it is the number to put in front of finance. Being upfront that not all of the bill is recoverable makes the figure you do claim far more believable than a sweeping promise to slash external costs.

Redeployed senior time, if your team is flat. If your headcount has been stable for years, or you sit in a large institution that holds it flat by policy, the avoidance argument will not help you, and you should not force it. Your case is about what your existing people spend their time on.

Right now your senior, expensive lawyers are doing a great deal of work that does not need their seniority: first-pass review, standard NDAs, the same handful of clauses a hundred times, because there is no one else to absorb it. There are two ways to put a number on what that costs. The direct one is to estimate the share of senior time spent on routine work and multiply it by what that time costs. A General Counsel spending a day a week on first-pass NDA review is an expensive way to review NDAs, and that figure tends to land once it is written down. The larger cost is the opportunity cost, which is harder to quantify: the advisory work not done, the strategic input not given, the risk left unmanaged, all the things your senior people would attend to if they were not clearing a queue.

I would be honest that this is a softer argument than cash saved, because it does not remove a line from the budget. It changes what the business gets for money it is already spending. So pair it with a concrete number wherever you can, and use the reallocation as the story that explains why the team becomes more valuable rather than cheaper.

Accelerated revenue, if legal sits on the critical path. In businesses where deals wait on contracts and contracts wait on legal, there is a revenue argument that often beats every cost argument in the room, because CFOs care about revenue in a way they rarely care about efficiency.

A bottleneck in legal pushes closings, and delayed closings push revenue into later periods. To size it, you need three things: the average value of a deal, the number of deals in a period, and the time legal currently adds to the cycle. Even a modest reduction in turnaround, multiplied across a year of deals, produces a number that gets attention, whether you frame it as revenue brought forward or as deals that no longer slip a quarter.

Reduced risk, as the weight behind the case. There is one more angle, and I would use it to make a case heavier rather than to carry one on its own. The work people currently route around legal is not only deferred volume. It is unmanaged exposure sitting on the business today.

When review is a bottleneck, people find ways around it, and every one of those ways is a small uncontrolled risk: contracts signed without review, outdated templates reused, counterparty terms accepted to save time, obligations nobody checked, liability caps nobody negotiated, auto-renewals nobody tracked. Individually, each is minor. In aggregate, across a whole business, they are the terms that surface in a dispute years later when it is far too late to change them.

The mistake is to try to attach a precise number to this, because you cannot, and a CFO will see straight through a fabricated one. What you can do is make it tangible. Point to the actual near-misses: the contract that went out with the wrong indemnity, the renewal that lapsed, the audit that turned up agreements nobody could locate. Use the volume of contracts being signed without review as a measure of how much exposure is currently going unmanaged. Then make the insurance argument, which a CFO already accepts everywhere else in the business. You are not buying certainty that nothing goes wrong. You are reducing the chance and the size of the things that do. Framed that way, risk does not need a false number. It needs a true picture, sitting behind a case built on the concrete savings above.

Building a case you can stand behind

The work of this conversation is not finding the cleverest argument. It is choosing the one that is actually true for your organisation and leading with it. A growing company gets the avoidance case. A flat institution gets the senior-time case. A team with a heavy external bill gets the outside counsel case. A revenue-critical business gets the acceleration case. Most teams have more than one of these available, and the risk angle adds weight to whichever they choose.

This matters more if you are carrying a scar from the last tool. The CLM that disappointed, the platform that never landed, the spend the CFO still remembers, none of that means the category is broken or that you were wrong to try. The previous tool was very likely sold to you on a justification that looked clean on a slide and was never connected to how the work behaves. When the case is disconnected from reality, the results cannot match the pitch, and the tool takes the blame for a promise it was never going to keep. The way to avoid repeating that is to make a case grounded in a saving the tool can actually produce in your specific organization.

And it is worth remembering, underneath all the CFO-facing language, what you are actually buying. Not a smaller team, in most cases, but a better one. Senior judgment recovered from routine work. Risk that was drifting out of your sight brought back into view. A team spending more of its time on the matters that need a lawyer and less on the queue. That is the real return, and the business case is simply the version of it your CFO can put a number against.

You are at the start of this. You do not have to get the tool wrong to get the case wrong, and the case is the part most within your control right now. Build it on the saving that is true for your organization, and you give the tool, and yourself, the conditions to actually deliver.

‍

Back to all posts