The Enquiry That Never Became a Matter: Fixing Legal Intake

The enquiry that never became a matter

Every firm has a number nobody reports: how many people tried to become clients and did not. They called during a hearing and reached voicemail. They filled in a form on Saturday and heard back on Tuesday. They asked a question the receptionist could not answer and did not call again.

None of this appears in any system, which is why it is rarely anybody’s problem. It is also, for most small and mid-sized practices, the single largest source of lost revenue — larger than write-offs, larger than unbilled time.

This is what “virtual assistants” and AI intake are actually for. Not replacing anyone. Making sure the enquiry survives contact with a busy week.

Why intake fails, specifically

Nobody is free when the call comes. Legal work is appointment-shaped. The people who could answer are in hearings, in meetings, or with another client. The caller does not know that and does not wait.

The first response is slow. Response time is the strongest predictor of conversion in almost every service business, and the legal profession is not exempt. An enquiry answered in an hour and one answered in two days are not the same enquiry.

The information gathered is inconsistent. Whoever picks up asks what they think to ask. Two enquiries about the same matter type arrive with different facts, and someone has to go back for the rest.

Nothing is recorded when the answer is no. Declined and lost enquiries vanish. A firm that cannot say how many people it turned away, or why, cannot tell whether it is turning away the right ones.

What AI intake should and should not do

Should: answer immediately, in a structured way. Acknowledge the enquiry, capture who they are, the matter type, the key dates, and how to reach them. Speed matters more than sophistication here — the point is that nobody waits until Tuesday.

Should: put it into the system as a record. The enquiry becomes a row with a status, not a note in someone’s inbox. That is what makes the number reportable, and reportable numbers get managed.

Should: flag urgency by rule. Limitation dates, custody matters, anything with a hearing this week. A simple rule that escalates on keywords catches most of it, and does not require a model to be clever.

Should not: give legal advice. Obvious, and worth writing into the configuration rather than assuming.

Should not: decide whether to take the matter. Conflicts, capacity, viability, and whether this is a client the firm wants are judgement calls with professional consequences.

Should not: be invisible to the caller. A person who believes they are speaking to a lawyer, and is not, is a complaint waiting to be made. Say what it is.

The conflicts problem, which is not a technology problem

An intake system that gathers names is gathering the raw material for a conflicts check, and that check has to happen before anyone gives advice or takes information they should not have.

Automation helps at the edges — the names are captured consistently, they can be matched against existing parties, and the check can be prompted rather than remembered. But the decision remains a lawyer’s, and a system that appears to clear conflicts automatically is worse than one that does not try.

What to measure

If you deploy anything in this area, these are the numbers that tell you whether it worked:

• Median first-response time, before and after.
• Enquiries with no recorded outcome — the ones that fell through. This number should approach zero, and it is the whole point.
• Enquiry-to-matter conversion, by source and matter type.
• Time from enquiry to matter opened.
• Out-of-hours enquiries answered — usually the clearest single gain, because it is where the old process had nothing at all.

Notice that none of these is about how many messages the AI handled. That number describes the software, not the practice.

The part people underestimate

Most of the benefit here is not intelligence. It is that something responds at all, consistently, and writes it down in the same place every time. A dumb form that always captures the same eight fields and always creates a record beats a clever assistant that sometimes does.

Which means the sequencing matters: get one place where enquiries live and one consistent set of questions, then automate the response. A firm without the first will find that automation produces faster chaos.

Where Lawnova fits

Lawnova treats intake as the front of the matter lifecycle rather than a separate product: voice intake captures an enquiry as a structured record, case management is where it becomes a matter, and smart notifications mean an urgent enquiry reaches a person rather than a queue.

The design assumption is the one above — that the failure is usually silence rather than sophistication. Answer, record it, tell someone. Most of the recoverable revenue is in those three steps.