How tortlconnect ranks every match.

Why we publish it

A ranking model is a moral document.

Whatever a directory ranks attorneys by is what the directory believes a good outcome looks like. If the ranking is hidden, the consumer has no way to know whether the model thinks good outcome means the same thing they do.

tortlconnect optimizes for fit. The features below are the only things that move the score. Money does not move the score. Recency of payment does not move the score. There is no tier that lifts an attorney above their fit-determined position.

If a directory will not show you the math, the math is not on your side.

The features

What each feature measures.

Each feature is a pure function: it takes the case, the attorney, and any pre-resolved context, and returns a number between zero and one. The descriptions below are paraphrased from the implementation. The implementation itself lives in a single feature library and is the authoritative source.

Specialty depth

How closely the attorney's declared focus aligns with the case type. Depth in the practice area beats a generalist who lists it.

Reputation

A normalized rating with a neutral floor for new attorneys — track records matter, but a clean newcomer is not penalized for the absence of one.

Recovery probability

A composite of liability strength, case type baseline, tortl verification, and remaining statute-of-limitations runway. The most honest single signal in the model.

Geographic proximity

Distance in miles, banded. Twenty-five miles or less is full credit; beyond two hundred the score falls below twenty percent.

Fee alignment

Whether the attorney's typical contingency percentage fits inside the consumer's declared maximum. Light tolerance below; hard penalty when the gap exceeds seven points.

Response-time fit

Average response hours, weighed against case urgency. An urgent case demands a fast attorney; a routine case forgives a 72-hour window.

Liability appetite

Whether the attorney takes strong-only cases, anything, or specifically welcomes complex liability — matched against the case's liability grade.

Case complexity fit

Derived from comparative negligence percentage, multiple insurers, pre-existing conditions, denial history, and liability grade. Matched against the attorney's complexity appetite.

Activity recency

How recently the attorney has been active in the platform. New attorneys aren't dormant — they're new — and are scored on a neutral floor until a pattern forms.

Language match

Whether the consumer's preferred language is covered by the attorney's declared languages.

Case value fit

Whether the case's estimated damages fit inside the attorney's declared value range. A small tail of credit for cases just outside the band.

Expected fee value

The case's estimated damages × recovery probability × the attorney's typical contingency, normalized against the attorney's declared value range.

Intake completeness

How much of the case file the consumer has actually filled in. Sparse intake is a strong predictor of a ghost client.

Critical-failure penalties

When a single feature should outvote the rest.

A weighted average can hide a catastrophic failure. An attorney can score well across nine features and have a 72-hour response time on a case where the statute expires in two weeks — and the consumer still sees a top ranking. tortlconnect addresses this with multiplicative penalties: when a feature falls below a defined threshold, the entire score is dampened by a fixed factor.

Penalties stack multiplicatively. A score that earned a 0.85 in the weighted sum and triggers both the recovery and the specialty soft-fail becomes 0.85 × 0.60 × 0.85 = 0.43 before any further adjustment.

Hard filters

When a match is excluded entirely.

Below penalties sit hard filters. These do not dampen a score; they remove the match from the result set altogether. A consumer never sees an attorney who cannot legally take their case, and an attorney never sees a case that is already past the deadline to file.

Not licensed in the case's state

tortl will not surface an attorney to a consumer in a state where they cannot practice. Bar-licensure is verified at onboarding and re-checked.

Does not practice the case type

Listed practice areas are the floor for eligibility — an attorney who has not declared they handle the case type is excluded.

At monthly capacity

Attorneys set a monthly intake cap. Once they hit it, they are excluded from new matches until the next month.

Statute of limitations has expired

If the case's statute window has passed, the match is filtered. The score is irrelevant; the underlying claim is gone.

The quality boost

One asymmetric lift, applied honestly.

When a tortl-verified case with strong liability and at least $50,000 in estimated damages is matched to an attorney with a 4.5-star rating or higher, the final score is multiplied by 1.10, capped at 1.00.

This is the only asymmetric lift in the system. It exists because a clean case with a top-rated attorney is the matching engine's success case — and our incentive is to make sure those matches surface first. The boost is documented here so it isn't mistaken for a hidden tier.

What tortlconnect does not do

The list is short, intentionally.

We do not sell match ranking.

The algorithm above is the only thing that orders a consumer's match results. Bidding does not enter the score. The browse view — a consumer scrolling the open directory of in-state attorneys — is a separate surface where impression order can be bid on, and is labeled as such in product.

We do not sell consumer information.

Consumer case data is shared with an attorney only when the consumer initiates contact through tortl. The matching algorithm reads the case file; it does not transmit it.

We do not boost recency-of-payment.

An attorney who pays on the first of the month and an attorney who pays on the twenty-eighth score identically. Money flow is not a feature.

We do not surface attorneys outside their license.

The hard filter on jurisdiction is non-negotiable. We will not show a consumer an attorney who cannot represent them, regardless of how good a fit they would otherwise be.

We do not blur the line between ranking and impression.

The directory's browse view (where bidding affects impression order) and the consumer's match results (where it does not) are visually distinct in product and labeled differently. A consumer always knows whether they are looking at "your matches" or "all attorneys in your state."