Key takeaways
- Chat is only worth adding if someone, or a properly supervised system, answers quickly and books a next step.
- The SRA sees chatbots as a way to respond to potential clients quickly and out of hours, but firms stay accountable for what they say[1].
- Consumers want safeguards: 89% expect an AI legal tool to have a complaints process[2].
- Run the staffing sums first; covering a chat window for a full working week is a real cost.
- Pilot for three months with clear success measures before committing.
Three kinds of chat
When firms talk about 'adding live chat', they usually mean one of three quite different things. The choice between them matters more than the choice of software.
| Model | Who answers | Typical hours | Main risk |
|---|---|---|---|
| Staffed by the firm | Reception or the new business team | Office hours | Chats left waiting while staff are on the phone |
| Outsourced chat agents | A third-party team using the firm's script | Often extended or round the clock | Agents who do not know the firm, and quality control at arm's length |
| Automated chatbot | Software following a script or using AI | Round the clock | Wrong or misleading answers, and handling confidential information |
Each can work. Each can also make the firm look worse than having no chat at all: a window that pops up, promises help and then says nobody is available is a poor first impression.
The case for chat
- Some people will not phone. They are at work, in a shared space or anxious about talking, and a chat feels lower commitment than a call.
- Chat captures enquiries at the moment of interest, while the person is on your fees or services page, rather than relying on them to fill in a form later.
- Out-of-hours chat can collect contact details and book a call-back for the morning.
- Transcripts show you the questions prospective clients actually ask, which helps improve website copy and triage scripts.
The regulator recognises the potential. The SRA's Risk Outlook on artificial intelligence notes that legal chatbots can respond quickly to enquiries from potential clients and potentially offer a 24/7 service, gathering information and triaging cases for urgency, and that they can help firms serve clients when staff would not otherwise be available[1].
The case against, or at least for caution
- Chat creates another queue. If the same receptionist answers calls and chats, both suffer at peak times.
- Chat invites detail. People type their whole story, including information about the other side, before any conflict check.
- Short written answers can read as advice. 'You should be fine' in a chat transcript is a liability.
- Pop-ups annoy some visitors and can slow the page on mobile.
- Third-party widgets often set cookies and send data to the provider, which needs proper consent and contracts.
Accountability does not move to the software provider. The SRA is clear that firms remain responsible and accountable for outputs from AI they use, including errors in advice given by a third-party chatbot[1].
What consumers expect from automated chat
If you are considering an AI chatbot, consumer research is clear that people want safeguards. In the Legal Services Board's 2026 Public Panel survey, 89% of legal service users expected an AI legal tool to have a complaints process, and 66% expected that process to involve a real human[2].
In the same research, 81% agreed AI legal tools should never act automatically for a vulnerable person in high-stakes situations, and AI acting without the user's knowledge or consent was the single most urgent concern, named by 56%[2]. The earlier Legal Needs Survey found the main barriers to using AI to access legal services were lack of trust in the technology (51%) and lack of human oversight (49%)[3]. Those findings point to a simple design rule: tell people when they are talking to software, keep its job to gathering information and booking, and make it easy to reach a person.
Compliance checklist before you switch it on
- Disclose clearly whether a person or software is answering, at the start of every chat.
- Keep the chat's job to triage and booking. No views on merits, prospects or time limits.
- Ask for the other side's name before the story, and warn people not to share detailed confidential information until the firm confirms it can act.
- Check where transcripts are stored, who can see them, how long they are kept and whether the provider uses them for anything else. The Legal Ombudsman warns that not all free AI tools are secure or keep what users type confidential[4]; the same question applies to any tool your firm uses.
- Put a data processing agreement in place with the provider and update the privacy notice.
- Load the chat widget only after cookie consent if it sets non-essential cookies.
- Make sure every chat ends with a clear outcome: booked, call-back arranged, declined with a signpost, or contact details taken with a promised response time.
- Name a person who reviews a sample of transcripts every week.
Run the staffing sums
Staffed chat is not free just because the software is cheap. Median gross hourly pay for UK call and contact centre staff was £14.86 in 2026[5], before employer National Insurance, pension, holiday cover and management time.
Example (fictional): a firm wants chat covered 8am to 6pm, Monday to Friday, which is 50 hours a week. At the median contact centre rate that is about £743 a week in basic pay alone, and more once on-costs and cover are added. If the existing reception team absorbs chat instead, the cost shows up as slower phone answering. Either way, compare it with the instructions you expect chat to add, using your own average fee per matter.
Outsourced and automated options change the sums but not the question. Ask what each extra instruction from chat would be worth, and how many you need a month for the option to pay its way.
Questions to ask a chat provider
Whether you choose outsourced agents or software, put the same questions to every provider and get the answers in writing before you sign.
- Where are transcripts and contact details stored, in which country, and for how long? Can we set our own retention period and delete on request?
- Do you use our transcripts to train models or improve services for other customers? Can we switch that off?
- How does the chat hand over to a person, and what happens to a chat that arrives when nobody is available?
- Can we control the script and the questions completely, and see every change before it goes live?
- How do you stop the chat giving legal views, prices we have not approved or promises about outcomes?
- What reports do we get: chats started, contact details captured, bookings made, response times?
- How do you handle someone in distress or describing a risk to their safety?
- What happens to our data if we end the contract?
Vague answers to the first two questions are a reason to walk away. A firm's duty to keep the affairs of clients and prospective clients confidential does not change because the conversation happened in a widget.
When chat is worth it
Chat tends to earn its place when most of these are true:
- A large share of your enquiries already come through the website.
- Your practice areas attract people who prefer to write, such as employment, where an online service matters more to many enquirers.
- You can answer every chat within a minute or two in the hours it is shown, or you only show the window when someone is available.
- You have a clear handover from chat to a booked consultation.
- Someone senior is willing to own transcript review and compliance.
If several are false, a faster web form response and a visible phone number will usually do more for less.
A three-month pilot
Set the baseline
Record a month of website enquiries, form completion and response times before chat goes live.
Limit the scope
Show chat on two or three high-intent pages, such as fees pages, in hours you can staff.
Track every chat
Log each chat as an enquiry with the same stages: suitable, booked, instructed, lost with reason.
Review transcripts weekly
Check for anything that reads as advice, missed urgency or confidential detail taken too early.
Decide on evidence
After three months, compare total website enquiries and instructions with the baseline. Keep, change or remove chat based on that, not on how busy the chat window looked.
Frequently asked questions
Is live chat worth it for a small law firm?
Only if someone can answer chats quickly in the hours the window is shown and book a next step. Otherwise a fast response to web forms and a visible phone number usually do more.
Can a chatbot give legal information to enquirers?
It can explain what the firm does and how to book, but it should not give views on a person's case. The SRA is clear that the firm remains accountable for what its chatbot says.
Do we have to tell people they are talking to a bot?
Yes, as a matter of good practice and fairness, and consumer research shows people expect transparency and easy access to a human.
What data protection checks apply to chat?
Check where transcripts are stored and for how long, put a data processing agreement in place with the provider, update your privacy notice and handle chat cookies through your consent banner.
References
Solicitors Regulation Authority (2023). Risk Outlook report: The use of artificial intelligence in the legal market (opens in a new tab) ↑
Legal Services Board (2026). AI in Legal Services: Research Report (LSB Public Panel) (opens in a new tab) ↑
Legal Services Board (2024). Legal Needs Survey 2023: Summary report (opens in a new tab) ↑
Legal Ombudsman (2026). AI and Complaints: A Consumer Guide (opens in a new tab) ↑
Office for National Statistics (2026). Earnings and hours worked, occupation by four-digit SOC: ASHE Table 14 (2026 provisional) (opens in a new tab) ↑