Check this page with an assistantOpens a chat asking it to summarise this article and name the evidence behind each claim.
Claude opens with the prompt on your clipboard: Anthropic does not support prefilled prompts on the web, and we would rather copy it than ship a button that drops it.
The Case You Lose Is the One That Rang at Seven
Legal intake fails on availability far more often than it fails on persuasion. Someone who has just been arrested, served, injured, or fired is not researching firms over a week. They contact several, and they retain one of the ones that responded while the problem still felt urgent.
This is why intake is the highest-leverage unglamorous system in a small firm. The marketing spend that produced the call has already been paid. The website that ranked has already been built. The only thing standing between that work and a signed engagement is whether anybody picked up, and for most firms under ten lawyers the honest answer outside business hours is nobody did.
The tempting conclusion is to bolt a chatbot onto the site and let it handle everything. That instinct is right about the problem and dangerous about the solution, because legal is the one vertical where the intake tool itself is regulated. The rest of this page is about where those limits sit, because they determine the build.
Who Starts the Conversation Decides Which Rule Applies
A visitor who chooses to open a chat window or fill in a form on your own site has initiated contact, and what they encounter is advertising. Model Rule 7.3 restricts something different: soliciting professional employment through live person-to-person contact with someone known to need legal services in a particular matter.
That single distinction carries most of the compliance weight, and it maps cleanly onto system design. Inbound automation, where the prospective client arrives under their own steam and engages a tool you have published, sits in the advertising rules: it must be truthful and not create unjustified expectations about results. Outbound automation that identifies people who appear to need legal help and contacts them is the pattern the solicitation rule exists to restrain.
It is worth saying plainly because vendors sell across verticals without adjusting for this. An automated sequence that texts everyone who filled in an unrelated form, or that works a purchased list of people with a known legal problem, is a normal product feature in most industries and an ethics problem in this one. Buy the tool, keep the feature switched off.
The System Collects Facts. It Never Answers “Do I Have a Case?”
Evaluating whether a set of facts supports a claim is legal advice. A non-lawyer system that offers it raises unauthorized practice concerns under Model Rule 5.5 and simultaneously creates the impression of a relationship the firm has not agreed to. The division is clean: the automation gathers and schedules, a lawyer assesses.
In practice this is less limiting than it sounds, because the questions that actually qualify a matter are factual rather than evaluative. When did this happen. Has anything been filed. Have you spoken to another firm. What county. Is there a deadline you have been given. A system can capture all of that, compare it against criteria the firm sets, and route accordingly without ever offering an opinion.
Where firms get into trouble is the friendly-sounding middle ground: a bot that says a matter “sounds like it may be worth pursuing” or that a deadline “has probably not passed.” Both read as reassurance to a worried person and both are assessments. The disclaimer at the bottom of the widget does not undo the sentence at the top of it.
The Question That Disqualifies You From the Case
Information learned from a prospective client can carry duties even when the firm never takes the matter, and under Model Rule 1.18 that information can disqualify the firm from acting for the opposing party. An intake tool that collects detailed facts from every visitor before any conflict screening manufactures that exposure at volume.
This is the failure mode with no analogue in any other industry, and it is the one almost nobody selling intake automation into legal will raise with you. A gym that captures too much information from a lead has a data hygiene issue. A firm that captures too much from the spouse of an existing divorce client, at two in the morning, through a widget nobody was watching, has a conflict it now has to deal with.
The fix is sequencing rather than restraint. Identity and adverse parties come first and get screened against the conflicts system before the tool invites anything narrative. Only after that clears does the automation ask what happened.
| Intake stage | Safe to collect | Hold until cleared |
|---|---|---|
| Identify | Name, contact details, county, matter type | Any narrative account of events |
| Screen | Names of other parties involved | Documents, evidence, uploads |
| Qualify | Dates, filings, existing representation | Strategy or settlement expectations |
| Schedule | Availability, preferred contact method | Anything resembling advice |
| Consult | Everything, with a lawyer present | Nothing at this point |
Firms resist this because the sequenced version feels colder than a sympathetic assistant inviting someone to explain their situation. The sympathetic version is exactly the one that collects disqualifying information from a stranger at scale. Warmth belongs in the consultation, where a lawyer is on the other end of it.
Being Found Is the Half Nobody Automates
Intake automation only compounds work that already happened upstream. It converts the contacts your visibility produced; it does not produce them. A firm with excellent intake and no discoverability has built a very fast system for handling a small number of people.
The two halves have different failure signatures, which makes them easy to tell apart. If consultations are booked but few, the constraint is upstream and the work is discoverability. If contacts arrive and go cold, the constraint is intake. Firms routinely buy the second fix for the first problem, because intake is the thing being sold to them this year.
Diagnosing which one you have takes an afternoon rather than a retainer, and the sequence for it is in traffic diagnosis. Where the constraint turns out to be visibility, legal is a genuinely hard vertical for it, because the subject matter sits squarely in the territory covered by YMYL and E-E-A-T and thin firm pages do not survive there.
Questions Firms Ask
- Is an intake chatbot considered solicitation?
The distinction that matters is who started the conversation. A form or assistant that a visitor chooses to use on your own website is advertising, and advertising is governed by the truthfulness rules rather than the solicitation rule. Automated outbound contact to a person you know needs legal help with a specific matter is the thing Model Rule 7.3 restricts, and automating it does not make it something else.
- Can an intake bot tell someone whether they have a case?
No, and this is the line that decides whether the tool is safe to run. Assessing whether a set of facts supports a claim is legal advice, and a non-lawyer system giving it raises unauthorized practice problems under Model Rule 5.5 while creating the impression of a relationship that does not exist. The bot gathers facts, checks basic fit against criteria you set, and books time. A lawyer evaluates.
- Why does the order of intake questions matter ethically?
Because information received from a prospective client can carry confidentiality duties even if you never take the matter, and under Model Rule 1.18 that information can disqualify the firm from representing someone on the other side. An intake tool that collects detailed matter facts from everyone who lands on the site, before any conflict screening, is manufacturing that risk at scale.
- Do bar advertising rules differ by state?
Substantially, and the ABA Model Rules are a template that states adapt rather than a law that binds anyone directly. Some states impose filing requirements, retention periods for advertising records, or specific disclaimer language. Any intake automation has to be reviewed against the rules of every jurisdiction the firm advertises into, not just the one where the office sits.
- Does faster intake actually win more cases?
It wins more of the cases you were already going to be considered for, which is a narrower claim than the one usually made. Someone with an urgent matter contacts several firms and retains one that responds. Responding first does not overcome being the wrong firm for the matter, and it does not create demand that was not there.
Primary Sources
- ABA Model Rules of Professional Conduct: table of contents, including Rules 1.18, 5.5, 7.1 and 7.3
- ABA Model Rule 1.7: Conflict of Interest, Current Clients
- ABA jurisdictional rules comparison charts, for state-by-state variation
- FCC: Telemarketing rules under the Telephone Consumer Protection Act
- Google Search Central: Creating helpful, reliable, people-first content
