Conditional-fee and contingency work: estimating risk with AI
By the ChamberOne team 5 minute read
A firm that takes cases on a conditional or contingency basis is, in a sense, an investor. It funds the work, carries the risk and is paid only if the case succeeds, in whole or in part. That makes the first decision, whether to take a case, one of the most important a partner makes. AI can help organise the information behind that decision. It cannot do the predicting.
Key points
- Check regulator and court rules on conditional or contingency fees before designing any arrangement.
- Use AI to organise evidence and build a chronology, then verify against the source.
- Forecast costs by stage from your own records, with a base, long and early-settlement case.
- Treat estimates as internal views, never as promises to the client.
First, check what you are allowed to do
Rules about conditional, contingency and success-based fees differ between places and between types of case. Some jurisdictions allow them freely in certain claims, some restrict them, and some require specific written terms or limit the percentage or the kind of dispute. Before building any model for a fee arrangement, check the guidance from your regulator or bar, and any court rules on costs.
This article is about the business side of the decision: how to assess and manage the risk. It does not say what is permitted where you practise.
What a case-selection decision needs
When a firm funds a case, the partner is weighing several things at once. Each can be written down, and the writing itself improves the decision.
- Liability: how strong is the case on the facts and the law, and what is the best available evidence?
- Quantum: what is a realistic range for the amount recovered, and what is the downside if the court awards less?
- Cost: how much time and how many disbursements will the case take to get to settlement or trial?
- Time: how long until the firm is paid, and what does that do to cash flow?
- The opponent: can they pay, are they insured, and how do they tend to behave in litigation?
- The client: are they credible, consistent and willing to see the case through?
Where AI helps with evidence
At intake, a firm may be handed a bundle of documents, a client account and a hopeful smile. Reading and organising that material takes time that is itself an investment. An AI tool can build a chronology from the documents, extract the main dates and parties, list what evidence supports each element of the claim and flag gaps, such as a missing medical report or an undated email.
That gives the partner a structured picture sooner and at lower cost. Every item still needs to be verified against the source, since a misread date or an invented fact in a case assessment can lead to a poor decision.
Cost forecasting
Forecasting cost is easier when you have done similar cases before. If your records show how many hours past cases of this type took at each stage, and the disbursements incurred, you can build a stage-by-stage budget for the new one. A tool that reads your own time and billing records can help you assemble those figures.
Keep the forecast honest. Show a base case, a case where the matter runs long and a case where it settles early. Note the assumptions. Revisit the forecast at fixed points, such as after the defence is received or after disclosure, and decide at each point whether to continue.
A portfolio view
One case is a gamble. A portfolio of cases is a business. A partner can look across all funded cases and ask how much money is tied up, how much is expected to come in and when, how many cases are in each stage and how concentrated the risk is in one kind of claim or one opponent.
AI can help gather the data and draft summaries, but the numbers come from your own records, and the probabilities come from partners' judgment. Be wary of a model that offers a precise percentage chance of winning. Whatever it says, the figure has no solid basis unless it was built from relevant, verified outcome data, and for most firms it was not.
Estimates are not promises
An internal estimate helps the firm decide. It should never be passed to the client as a prediction or a guarantee. Clients must understand that outcomes are uncertain, that estimates may change and what happens if the case fails. Check your regulator's guidance on what must be explained to clients on costs, risks and funding. Put the explanation in writing, in plain words.
Also be careful about how you describe AI-assisted assessments. 'Our software rates your case highly' is a risky thing to say. An assessment is a view held by the lawyer, with reasons, and the lawyer owns it.
- Use the estimate to help the firm decide
- Explain that outcomes are uncertain and estimates may change
- Describe an assessment as the lawyer's view, with reasons
- Pass the estimate to the client as a prediction or guarantee
- Say 'our software rates your case highly'
A review routine that keeps you honest
- 1
Record your original assessment of every funded case: strengths, weaknesses, estimated cost and range of recovery.
- 2
At each stage, update the assessment and say what changed and why.
- 3
When a case ends, compare the result and the cost with the original estimate.
- 4
Once a year, look at the pattern. Which kinds of case did you overestimate? Which opponents behaved differently from expectation?
- 5
Use the lessons in the next case-selection meeting.
- 1Record the original assessment
- 2Update it at each stage
- 3Compare the result with the estimate
- 4Look for yearly patterns
- 5Use the lessons in case selection
- Then back to step 1.
What AI cannot do here
It cannot tell you how a particular judge will see the evidence, how a witness will perform, or whether the defendant will settle. It cannot tell you whether a client will stay the course. It cannot take the financial risk. Those remain with the partners. The most valuable thing a tool can offer is a clearer, faster view of the facts, so that the partners' judgment rests on better ground.
Be disciplined about the cases you decline, as well. Keep a short note of why each was refused, and tell the client promptly, in writing, so they can look elsewhere before any time limit runs out. Declined cases also teach you something: patterns in what you turn away can show where your criteria are too tight or too loose, and the notes are useful evidence of a considered process.
This article is general information about the law, legal practice and technology, current when it was published. It is not legal advice. Laws and guidance change: check the sources linked above and take advice on your own circumstances from a lawyer qualified where you are.
Read next
AI and the billable hour: what actually changes and what does not
How faster drafting puts pressure on hourly billing, and which fee models suit which work.
5 minute readFixed-fee pricing with AI-assisted scoping
How to use past matter data and AI summaries to scope a fixed fee, define exclusions and set a review point.
5 minute readRecovering unrecorded time: how AI time capture works and its limits
How suggestions from calendar, email and documents help fill timesheet gaps, and why the fee earner must still confirm each entry.
5 minute read