A local referral business can help someone reach a suitable service provider. Calling the customer a “lead” does not change what they expect: an honest explanation of who they contacted, where their information goes, and who may call them. AI should help organize that request, not conceal a resale operation.
Start with a limited, non-emergency service you understand and a few providers you can verify. Avoid beginning with medical, legal, financial, or emergency inquiries. Those can involve sensitive information, additional professional rules, and consequences that a basic intake system is not equipped to manage.
Explain the business before collecting details
State clearly that you operate a referral service, identify your business, and explain how you are paid. Do not present yourself as the contractor, use another company’s credentials, or create fake local offices. Let visitors see the provider choices and the limits of any checks you perform.
Verify each provider’s identity, service area, availability, and relevant licenses or insurance through appropriate sources. Record when you checked and what the evidence actually shows. A license check is not a guarantee of workmanship. Recheck information when it expires or when complaints raise concerns.
For advertising, check the rules of the exact platform and product you intend to use. Google’s Local Services Ads requirements prohibit advertisers from selling or passing leads to other businesses and prohibit routing requests to businesses not listed in the ads. That product is not a channel for the resale model described in many “lead arbitrage” pitches.
Make the handoff a deliberate choice
Design intake so the person knows the named provider, the information to be shared, and the expected contact method before agreeing. A useful starting design is one requested referral to one selected provider. That is a transparency practice, not a claim that one checkbox satisfies every legal requirement.
Keep the service request separate from optional marketing subscriptions. Do not treat an inquiry as permission for unrelated companies to contact someone indefinitely. If the selected provider cannot help, return to the customer with another choice rather than silently distributing the details elsewhere.
Record the wording shown, the person’s choice, the recipient, and the time. Preserve only the evidence reasonably needed under your approved retention policy. Make correction, cancellation, and privacy requests easy to submit, and ensure providers receive relevant changes promptly.
Give AI an administrative role
A practical first workflow uses AI to suggest a service category, flag missing details, and prepare a short summary for a person to review. For example, “interior painting, two rooms, seeking an estimate next month” is a useful description. The tool should not invent a budget, homeownership status, urgency, or willingness to buy.
- Identify the referral business and disclose the automated assistant if one is used.
- Ask for the minimum details needed to understand the requested service and general area.
- Show the available provider and sharing explanation.
- Confirm the customer’s contact choice and check the summary.
- Send the approved request, then track whether it reached the intended provider.
Leave sensitive documents, payment-card details, and detailed personal histories out of the intake. Do not let an AI system bid on access to people’s information or distribute it automatically to unapproved recipients. A web form and a reviewed handoff may be sufficient without a voice bot.
Check contact, recording, and privacy rules separately
The FTC’s Telemarketing Sales Rule guidance explains obligations that can apply to telemarketing, including do-not-call and prerecorded-call provisions. Applicable rules depend on the activity and jurisdiction. Do not assume that an incoming inquiry authorizes later automated sales calls or texts.
Call recording and transcription raise separate consent and privacy questions. Have qualified counsel review the actual scripts, forms, routing, recordings, and follow-up arrangements before launch. A supplier’s compliance badge is not a legal assessment of your operation. The workflow here does not involve cold AI calls or pretending a synthetic voice is a real employee.
Agree with providers how they may use the information, prohibit onward resale, restrict staff access, and establish deletion and incident procedures. Check what the AI supplier stores or uses for training. Minimize the data available to each participant rather than sending complete conversation histories by default.
Test whether referrals help both sides
Define a billable referral in writing: the right service and area, a genuine contact request, agreed sharing permission, and a usable contact method. State how duplicates, wrong numbers, unavailable providers, disputes, and refunds are handled. Do not promise a sale or exclusive access unless you can deliver it.
Calculate advertising, intake, software, human review, refunds, and support costs against collected revenue. Track complaints and unwanted contact alongside provider response times. Pause a provider who repeatedly ignores the agreed boundaries; more referrals are not a success if customers regret asking for help.
For an alternative arrangement, see managing local advertising for a named business. That service has a different customer relationship from operating an independent referral network. This article is general information, not legal advice or a promise of profitability.
AI disclosure: This article was written and revised with AI assistance. Sources checked September 28, 2026. See our disclaimer.
Illustrative photograph by Giorgio Trovato on Unsplash.
