As of 13 August 2026, AI can ask qualifying questions to website visitors.
Most people should hand this to a purpose-built tool.
Can you do it?
5 minutesto a draft.
30 minutesto something you’d act on.
Cost, all in£0/month
Skill neededpower-user
Who has to check ita colleague
What the alternative costsA purpose-built website live-chat product such as Tidio Lyro is the direct alternative; its price is not specified in the supplied data.
If this goes wrong: the chat filters out a good prospect, collects information you did not need or gives a visitor a misleading route, and your team has to recover the missed opportunity.
What to actually do
Use a tool built for this
The route this page recommends
Hand it to a person
Second choiceA person who owns the outcome does this end to end, worth it when the failure is dear.
Do it yourself
The distant thirdA chat interface, power-user skill, and roughly 30 minutes until you can act on the result.
How to actually do it
- Open your current sales qualification guide, ideal customer profile and product or service page, then gather the customer types, buying triggers, disqualifiers, service areas and hand-off options.
- Write down the maximum number of chat questions, the contact details you actually need, the sales team or booking route for each lead type, and any information the chat must never request.
- Paste that information into the prompt and replace every bracketed slot with your business's own criteria, routes and wording.
- Run the prompt in a chatbot and copy the script, routing table and test conversations into a working document.
- Compare every product claim, eligibility rule, price reference and hand-off instruction against your current website and sales materials, then correct the script before deployment.
- Open Tidio Lyro or another website chat platform, add the approved script and routes, and configure contact collection only after the consent message.
- Test the live chat as each of the five example visitors, including refusing contact details and asking for a person, then send the transcript and routing result to a colleague for approval.
- Publish the chat, review early transcripts against the intended lead categories and change any question that is causing good prospects to be rejected or routed incorrectly.
Prompt
Design a website lead-qualification chat for [business name], which sells [product or service] to [target customer]. The chat should ask no more than [number] questions, one at a time, in a plain and helpful UK English tone. Qualify visitors using only these criteria: [list the criteria]. For each criterion, state the acceptable answers and what to do when the answer is unclear. Ask for contact details only after explaining why they are needed and obtaining the visitor's agreement. Do not ask for sensitive personal data, invent product facts, make promises, or pressure the visitor. At the end, classify the visitor as [lead categories], explain the classification in one sentence, and choose one route: [sales hand-off, booking link, email follow-up, self-serve information, or other route]. Include a clear human hand-off message for visitors who ask for a person, object to a question, or do not fit the criteria. Return: 1) the exact chat script, 2) the routing rules in a table, 3) the information the business must supply before publishing, and 4) five test conversations covering a strong lead, weak lead, unclear answer, refusal to provide contact details and a request for a human.
Open it prefilled in ChatGPT or Claude, or copy it into Gemini, which takes no prefill link.
What it gets wrong
- AI cannot decide which qualification criteria reflect your actual sales strategy without your customer and pipeline context.
- AI cannot reliably recognise every indirect buying signal, evasive answer or unusual but valuable prospect.
- AI cannot take responsibility for excluding a visitor, making a claim about your service or deciding whether a hand-off is commercially sensible.
- AI cannot determine your lawful basis for collecting personal data or replace your privacy notice and internal data-handling decisions.
- AI cannot build trust with a hesitant visitor as consistently as an experienced salesperson who understands the relationship.
Even on a YES, the friction has a name: judgement under ambiguity, consent and privacy and context depth.
How we scored this
Five axes, each scored nought to two by hand: ten means AI carries the task cleanly, and the thresholds that turn a total into YES, PARTLY or NO are published in the methodology. Each axis name links to its definition.
| Axis | Score (0–2) |
|---|---|
| Output | 2 |
| Inputs | 2 |
| Verification | 1 |
| Liability | 1 |
| Effort delta | 2 |
| Total | 8 / 10 |
The methodology and its thresholds are published in full.
FAQ
- Can ChatGPT qualify leads on my website?
- Yes, it can design the questions and rules, but a chatbot alone does not place the conversation on your website. Use a website chat product such as Tidio Lyro to run the conversation, and test its routing against real sales scenarios before publishing.
- What questions should an AI chatbot ask potential customers?
- Ask only what changes the next action, such as the visitor's need, timing, fit, location or preferred route to contact. Set the criteria from your sales process, ask one question at a time and avoid collecting personal data that you do not need.
- Can an AI chatbot tell if a lead is serious?
- It can classify answers against rules such as need, timing and fit, but it cannot reliably judge intent from vague or polite replies. Keep an option for a human hand-off and have a salesperson check the rules against actual conversations.
- Is it safe to collect lead details with an AI chatbot?
- It can collect details if you explain the purpose, obtain agreement and use a properly configured business system. You remain responsible for deciding what data to collect, how long to retain it and what your privacy information must say.
Nearby answers
Assessed by gpt-5.6-luna (gpt-5.6-luna) on 2026-08-13, second-checked by an independent model. Wrong somewhere? Email [email protected] and it gets re-checked.
The newsletter
AI news, new answers and product picks, straight to your inbox.