Home / The matching model
Six layers
Analysed properly, then decided by a person.
Most referral services are a form and a spreadsheet. Ours is a structured intake feeding a scored capability matrix, with hard constraints applied first and a human making the final call — then a feedback loop that adjusts the weights based on how the people we introduced were actually treated.
We will also tell you what it is not. It is not a black box, it does not predict what your claim is worth, and it does not decide anything on its own. Any service claiming an algorithm picks your lawyer is describing a lead-routing rule and dressing it up.
Structured intake
Around forty discrete variables, captured as structured data rather than a free-text box that someone skims. Mechanism, jurisdiction, elapsed time, injury markers, treatment pattern, functional and economic impact, counterparty class, evidence posture, insurer posture, existing representation, working-style preference, and access requirements.
The questions were built backwards from the forks that actually reassign a file.
Feature extraction
Raw answers become derived signals. Elapsed months become limitation proximity. A counterparty class becomes a possible short-notice obligation. A cluster of head, memory and consciousness answers becomes a brain-injury indicator that outranks the collision category it arrived in. A denial date becomes a review-window flag. You see every one of these before you submit — the read-back on the final screen is this layer, exposed.
Nothing is inferred that we would not show you.
Hard constraints
Before any quality judgment is made, the field is cut by things that are not negotiable: province and applicable regime, language of service, conflict screening, whether a firm is licensed and currently practising in the right jurisdiction, and live capacity — whether they can genuinely take a new file of this size this month. Capacity alone removes more firms in a given week than anything else.
A brilliant firm with no bandwidth is the wrong firm today.
Capability matrix
Every remaining firm is scored against a matrix held per claim type and per province — not "personal injury" but birth injury, catastrophic designation, commercial trucking, disability terminations at the change-of-definition date, cap-exit medical characterisation, benefits appeals. Weighting shifts with the file: litigation posture dominates where liability is contested, disbursement capital dominates where the claim needs six figures of expert work, and speed of first contact dominates where a deadline is close.
The weights change per file. That is the whole point.
Human review
The model narrows. A person decides. Every match is read by someone before a name is given, because the reasons a specific lawyer suits a specific person do not reduce cleanly to features — bereavement, capacity, family dynamics, a client who has already been through one bad experience. Where the model and the reviewer disagree, the reviewer wins and the disagreement is recorded.
Software has never made the final call here and will not.
Outcome feedback
We follow up after the introduction and again later in the file. Did they call back when they said they would? Do you know who is handling your file? Were you told the weaknesses honestly? Those answers adjust routing weights directly. A firm can be technically excellent and still be routed less because the people we sent stopped feeling looked after.
This is the only mechanism that can move a firm’s standing.
Worked example
One file, four reassignments
Watch how a single case moves through the model as each detail arrives. This is what "matching on minute details" actually means in practice.
"I was rear-ended in Mississauga."
Category: motor vehicle, Ontario. At this point the pool is large — several dozen eligible firms. Nothing has been decided.
"It was a delivery truck."
Commercial carrier. Electronic logging, telematics and maintenance records are in play, defence counsel is likely already engaged, and preservation demands need to go out this week. The pool contracts sharply to firms with demonstrated carrier work.
"I was dazed for a few seconds — I didn’t black out."
Brain-injury indicator. This now outranks the collision category entirely. The pool contracts again to firms with genuine acquired-brain-injury depth, because the medical evidence strategy is a different discipline from soft-tissue work.
"I’ve been off work four months and my partner has stopped working to help me."
Household under financial pressure and an unpaid attendant-care claim that is being lost every week it goes unrecorded. Weighting shifts toward firms that move fast on interim benefits and advances — not just the ones who litigate well.
"Just tell me what to do. I can’t research this."
Directive working style. Between two firms of comparable capability, the one whose partner runs files hands-on and calls clients directly is chosen over the one that communicates through a well-organised clerk team. One name, with the reasoning, in your inbox.
Where the research comes from
Grounded in sources you can check yourself
Our jurisdictional layer is built from provincial limitations statutes, insurance legislation and regulations, occupiers' liability and municipal notice provisions, published tribunal and court decisions, and law society registers and conduct rules. Where a rule is indexed annually or subject to legislated change, we record a verification date and show it on the page rather than letting a stale figure look current.
We publish structure and mechanism, not dollar figures. Caps, deductibles and benefit limits are re-indexed every January in most provinces, and a confidently wrong number does more damage to an injured person than an honest "this is indexed — confirm the current figure."
Every jurisdictional page on this site ends the same way: this is structure, not advice, and the application to your facts is a question for a lawyer. We are confident about the map. We do not pretend to know your terrain.