A useful quantified decision is not one that talks about the same subject, it is one that awards an amount on the same head of loss, with comparable parameters. Court decisions are now open to the public, but they are distributed as text, not as amounts you can use head by head. All the work therefore consists of moving from a body of decisions to a series of amounts comparable to the file.
In personal injury, the rule of law is rarely the point in dispute. The principle of full compensation is settled: the victim must be compensated for all their loss, without gain or shortfall. The Dintilhac classification (nomenclature Dintilhac, the court-endorsed classification of the heads of personal-injury loss), which has no binding force but has established itself in practice, sets the breakdown into heads. What is discussed, head by head, is the amount.
Looking for case law in personal injury is therefore not looking for a statement of principle. It is looking for a value: what a court actually awarded for souffrances endurées (the pain and suffering endured up to consolidation, rated 1 to 7) at a given degree, or for a permanent functional deficit (déficit fonctionnel permanent, "DFP") at a given rating and age. This difference in objective completely changes the search method.
Decisions handed down by the judicial courts are made available to the public free of charge and in electronic form (article L111-13 of the Code of Judicial Organisation). The surnames and first names of natural persons, whether parties or third parties, are redacted before this release.
The opening happened gradually, on a timetable set by decree no. 2020-797 of 29 June 2020 and by the order of 28 April 2021, amended several times since, court level by court level and matter by matter. On the judicial side, distribution runs through Judilibre, the data processing operated by the Cour de cassation, France's supreme court for private-law matters (decree no. 2021-1276 of 30 September 2021).
Three sources therefore coexist in practice: open judicial data, the paid document bases of legal publishers, and the collections of decisions put together by practitioners and courts. All three provide the full text. None provides, natively, a series of amounts by head.
Opening up the decisions removed an access barrier, not the underlying difficulty. Four obstacles remain intact.
The amounts are buried in the text. A personal-injury appeal ruling can contain thirty amounts: the claims, the insurer's offers, the sums awarded, the interim payments already made, the recoverable costs. Nothing mechanically distinguishes the sum awarded for the DFP from the sum claimed and refused.
The file parameters are not structured. The DFP rating, the age at consolidation, the degree of souffrances endurées appear in the body of the decision, in words, in a style specific to each court. A full-text search for "DFP 15%" misses every decision that writes "déficit fonctionnel permanent of fifteen per cent".
The scope of the amount varies. The recourse of third-party payers is exercised head by head, on the sole indemnities repairing the losses they have covered (article 31 of law no. 85-677 of 5 July 1985). Depending on the decision, the amount you read is the head before the claim is set off, or the balance going to the victim. Comparing one with the other silently distorts the comparison.
Time deforms the series. A sum awarded ten years ago does not compare as is with a recent decision. On a head where amounts have shifted, mixing vintages produces an average that describes no real practice.
A decision is only relevant if it is close to the file on the parameters that drive the amount of the head examined:
| Head of loss | Parameters to match |
|---|---|
| Déficit fonctionnel permanent (DFP) | Rating retained, age at consolidation, nature of the sequelae |
| Souffrances endurées | Degree on the 1 to 7 scale, duration and intensity of care |
| Préjudice esthétique (cosmetic damage) | Rating retained, age, location |
| Loss of professional earnings | Professional situation before the accident, capacity to return to work |
One criterion to set aside: the court or the panel that handed down the decision. Reusing the identity data of judges and court registry staff for the purpose or effect of evaluating, analysing, comparing or predicting their professional practices is prohibited by article L111-13 of the Code of Judicial Organisation, and this breach is punishable by the penalties set out in articles 226-18, 226-24 and 226-31 of the Criminal Code. Selecting decisions by jurisdiction for practical reasons is one thing; building a profile of a judge is another, and that one is unlawful.
This is exactly the work Plato has industrialised, on a base of more than 84,000 quantified decisions: extracting from each decision the heads, the amounts and the medical parameters, then allowing a file to be compared with real decisions at the same rating, the same degree and a close age. Each amount stays traceable back to the decision it comes from, which is the condition for producing it before a judge or against an insurer.
To place a specific file, the quantified case-law base can be searched at jp.plato.legal. For the valuation method upstream, see our personal-injury valuation guide, the detail of the déficit fonctionnel permanent and of souffrances endurées.
Where can you find quantified personal-injury case law?
In open judicial data, distributed notably via Judilibre, in the bases of legal publishers and in practitioners' collections. These sources provide the full text: extracting the amounts by head remains to be done, by hand or with a specialised tool.
Are court decisions public and free?
Yes. Article L111-13 of the Code of Judicial Organisation provides for their release to the public free of charge in electronic form, after redaction of the surnames and first names of natural persons.
Why isn't a full-text search enough?
Because the amounts and medical parameters are written in natural language, differently across courts, and because one decision contains both the sums claimed and the sums awarded. Full text finds decisions, not series of comparable amounts.
What makes two decisions comparable?
Matching on the parameters that drive the head: rating of déficit fonctionnel permanent and age at consolidation, degree of souffrances endurées, rating of préjudice esthétique. The triggering event and the date of the decision matter too.
Can you compare the practices of a judge or a court?
No. Reusing the identity data of judges and court registry staff to evaluate, analyse, compare or predict their professional practices is prohibited (article L111-13 of the Code of Judicial Organisation), under the penalties set out in articles 226-18, 226-24 and 226-31 of the Criminal Code.
Should you keep the head amount or the balance going to the victim?
You need to know which, and never mix the two. Third-party payers' recourse is exercised head by head (article 31 of the law of 5 July 1985), so the same decision can show a very different head amount and balance.
How many decisions do you need to place a file?
Enough to see the dispersion, not just a central value. Two or three isolated decisions say nothing about practice; a series of strict matches lets you argue a supported amount and anticipate the other side.
Can an AI do this search?
A generalist AI cannot: it has no base of quantified decisions and may invent a reference. A vertical tool built on an extracted, verified case-law base can, provided each amount stays traceable back to its source.