Proving the Value of AI Starts With Agreement
By Tara Waters, Practice Director, Transformation Office, Harvey
Most law firms are not simply trying to measure the value of AI, they’re trying to prove it. Those are different endeavours: a firm can estimate what AI has changed about the cost of delivering its work, but proving its value to the client requires agreement on what that value is. Agreement takes two parties, which is the part almost every ROI framework misses.
Proof needs a counterfactual and firms typically do not run matters twice. The best a firm can do is compare the closest matters it has and produce a defensible estimate of its own cost to deliver. But that is not the figure in dispute. When a client pushes back on price, the argument is about what the work was worth to them, and no amount of matter comparison will tell a firm that.
So the firm has to ask the client, and the timing is crucial. A general question about what a client values produces platitudes. The same question after the invoice is raised produces a negotiating position. The question asked while the matter is still being scoped produces something a firm can actually price against.
Agreement starts with a conversation but does not stay one. It requires an information trade, which may be uncomfortable in both directions. The firm has to say where AI was used, what it replaced and what the work now costs to deliver. The client has to say what they have paid for this work elsewhere, what they still spend on it internally and where else they could send it. Each of those is the thing the other side has no way of knowing, and each is the thing its holder has every short-term reason to keep.
Which is why agreement rarely happens, not because nobody thought of it. Whoever goes first gives up a bargaining advantage. A firm that opens its cost base invites a discount request. A client that names a figure sets a ceiling. Both are behaving sensibly, and neither is wrong to wait.
What breaks the standoff is sequence rather than goodwill. Firms already agree a great deal before a matter starts: scope, staffing, budget and often the fee itself. None of it covers what a better outcome would be worth, which is the number everybody argues about afterwards.
In practice that means adding the question to the scoping conversation that already happens: how to measure in the client’s terms, against the baseline the client holds rather than one the firm has estimated, and writing down what a better outcome would look like on the matter. Faster by how much, on which risk, to what standard, measured how? Both sides review it at the close, including the parts that did not land.
That costs the firm something real: the ability to decide afterwards what the work was worth. What it buys is a measure the client has already accepted.






Comments