The standard contingency fee has held at a third of the recovery for two generations. A third if it settles, forty percent at trial. It's so fixed that most owners treat it as a law of nature instead of what it actually is — a price. And every price that's gone unchallenged for forty years is a price that has never had to defend itself. As the owner, the question that should keep you up isn't "will my fee hold?" It's "what happens the day a competitor decides it shouldn't?"
Here's the answer. Within two years, the going rate in competitive markets slides from 33–40% toward 10–20% — not because the bar votes for it, but because a handful of AI-enabled owners discover they can win the same cases at a fraction of the labor cost, and the fastest way to take your clients is to undercut you on the one number the client actually understands. The compression has already started. The first firms are quietly quoting 25%. Nobody sent a press release.
When your competitor's cost falls, you don't feel it as their advantage. You feel it as your price war.
Why the fee was always mostly labor
A contingency fee covers two things: risk and labor. The risk is real — you might lose, you front the costs. But a huge slice of that one-third has always paid for labor: the intake calls, the records requests, the demand letters, the status updates, the endless follow-up nobody bills hourly but everybody funds. Strip the labor cost down by 70% with AI and the fee that made sense at a third makes embarrassing sense at half that. The owner who cuts their cost-to-serve first gets to choose: keep the fee and pocket the margin, or drop the fee and take the market. The owner who waits gets that choice made for them.
And the labor under the fee is exactly what's most automatable. Goldman Sachs estimates generative AI could expose the equivalent of 300 million full-time jobs and ranks legal among the most exposed categories. Thomson Reuters, surveying the profession directly, found lawyers expect AI to absorb a large and growing share of their hours within five years. The work that justified the fee is the work a machine does first.
$0.20
If a rival's cost-to-serve drops far enough, 20% of the recovery is a better margin for them than 33% is for you. At that point the client picks the cheaper firm in four seconds — and there's no loyalty to a fee.
The owner's move: compete on cost, not price
Most owners watch the wrong competitor. They eye the firm across town with the same overhead, the same playbook, the same one-third fee. That firm isn't the threat. The threat is the owner who rebuilt intake, records, and demand generation on AI, dropped their cost-to-serve through the floor, and is now happy to take your client at 20% because at their structure 20% prints money.
You don't beat that by defending your fee. You beat it by getting your own cost-to-serve down first — so that when the price moves, you move with it profitably instead of getting dragged there last and bleeding. The lowest-cost operator sets the market price. Decide now whether that's going to be you.
The lowest-cost operator sets the price for everyone. The only question is whether that operator is you or the person taking your clients.
The "it's not reliable yet" trap
The reflex is to wait for proof. And the failure modes are real — a Stanford study found leading legal AI tools still hallucinate on a meaningful share of queries, and courts have sanctioned lawyers for fake citations. But "it isn't ready" is exactly what every incumbent says right before the curve catches them. You don't need AI to replace the lawyer. You need it to replace the labor under the fee — and a records workflow that runs at 95% and gets reviewed is already cheaper than the human it replaces. That's a far more forgiving bar, and it's where your cost advantage actually comes from.
What to do before the rate moves
- Measure your real cost-to-serve. Most owners don't know what one case actually costs them to run. You can't cut what you can't see.
- Automate the process, not the lawyer. Intake, records, demand drafting, updates — the labor under the fee. That's where the margin is hiding.
- Decide your floor on purpose. If 20% is coming, be the owner who gets there deliberately and profitably, not the one forced there last.
- Watch the right competitor. Not the firm with your cost structure — the one quietly rebuilding theirs.
The one-third fee isn't a right; it's a price that hasn't been challenged. The challenge is here, it's early, and it rewards the owner who moves first. Cost-to-serve is the one edge in this business that compounds quietly and can't be matched overnight. Grab it before the owner across town does it to you.
Sources & further reading
- Goldman Sachs Research — "Generative AI could raise global GDP by 7%." goldmansachs.com The ~300M-jobs exposure figure and legal's high automation ranking.
- Thomson Reuters — "Future of Professionals Report." thomsonreuters.com How much of the billable hour professionals expect AI to absorb.
- Stanford HAI / RegLab — "AI on Trial: Legal Models Hallucinate." hai.stanford.edu The reliability counter-case — why "automate the labor, review the output" beats waiting.