TL;DR

  • When someone asks "where did you get my number," name the true source in one plain sentence. Never dodge.
  • Say the call is from an AI before they ask, and give an easy way out.
  • Below are 6 example responses.

Six example responses by scenario

Each reply names a source, owns the AI status and offers a way out. Swap the bracketed parts for facts from the contact record. Buyers ask how the agent handles this question, so build every reply to settle trust first.

1. App or site sign-up

Say: "Good question. You signed up on our app on [date], and we used the number from that sign-up. I'm an AI assistant calling for [Brand]. Would you like me to stop calling?"

Why it works: People forget they shared a number. A short recap of the sign-up settles the doubt.

Avoid: "We have your details on file."

2. Existing customer reminder

Say: "We have this number from your [loan] account with [Brand]. I'm an AI assistant, and I'm calling about your payment due on [date]. If you'd rather speak to a person, I can arrange that."

Why it works: It names the account and the reason for calling in one breath. The customer sees the call is about their own money.

Avoid: "This is a routine call." It gives no reason.

3. Web form or inquiry lead

Say: "You filled in our form on the [page name] page on [date]. That's where we got your number. I'm an AI assistant following up on your question. Should I take you off the list instead?"

Why it works: Only the real lead would know the page and the date. This fits a lead follow-up campaign type.

Avoid: "You showed interest in us." It is vague and sounds like a script.

4. Cold list outreach

Say: "Fair question. We found your number on a [business directory / purchased list]. We don't know each other yet. I'm an AI assistant, and I can remove you right now if you like."

Why it works: It names the list type and keeps the tone honest. Honesty keeps the call from turning hostile. Cold outreach is a campaign type of its own, so record the list source.

Avoid: "You're one of our valued customers."

5. Referral source

Say: "[Name] gave us your number. They thought you might want to hear about [offer]. I'm an AI assistant calling for [Brand]. Want me to remove your number?"

Why it works: A named person is easy to check. The contact can call that person and confirm.

Avoid: "A friend shared it." Without a name, this sounds like an excuse.

6. Fraud fear or unknown source

Say: "Are you a bot? Yes, I'm an AI assistant. I can't confirm where this number came from on this call, and I'm sorry. I'll add you to our do-not-contact list now. I can also have a person call you back from [Brand's published number]."

Why it works: It admits the gap and stops guessing. A callback on a published number lets the person check the caller on their own terms.

Avoid: Making up a source to end the questions. The record becomes wrong and the trust is gone.

How to use these replies

Follow four rules in every reply.

  1. Name the true source in one plain sentence.
  2. Configure your agent so it answers yes when asked if it is an AI.
  3. Offer opt-out right away.
  4. Keep it to two or three short spoken sentences. Buyers fear AI that sounds scripted, and long answers sound scripted.

Load the facts before the first call. The source must come from campaign data, never from improvisation. Put the lead source, the sign-up date and the consent status in every contact record. If those fields are empty, the agent has nothing true to say. It will dodge or guess, and both lose the call.

Then ask the agent your six questions. Talk to it in the browser first. Then call your own phone and ask each one, including "Are you a bot?" Check that each answer names the source in your records, says it is an AI and offers a way out. Where the reply is "I can't confirm," design the agent to offer a human callback. In Insight7, calls where only the AI spoke are not counted as reached.

Use one product to do the work. Insight7 campaigns are in beta. A campaign has a type (customer survey, lead follow-up, cold outreach or win-back), consent marking, calling hours, retry rules, a do-not-contact list and a test call to your own phone. The agent can be tested in the browser with no phone number needed. Insight7 can also score every call against a weighted scorecard your team owns, not a sample.

Check the rules where you call. This is not legal advice, so confirm with counsel.

How to adapt them to your calls

Rewrite each reply per campaign type. A survey, a lead follow-up, a cold outreach and a win-back call each have a different true source, so each needs its own wording. Write the reply the way people in that market speak. Listeners tune out accents that sound foreign, so have a native speaker read each line aloud before launch.

Score every call for this moment. Add three criteria to your scorecard: "named the true source," "confirmed AI status honestly" and "offered opt-out." Treat a wrong claim of source as a critical error that fails the whole call. QA leaders say one critical error should fail the call even when the rest follows procedure.

Check the scores against a human. Have your QA manager listen to a handful of scored calls each week. Insight7 call scoring has built-in calibration that lets managers review, override and fine-tune AI ratings.

Read the short calls. Pull the transcripts where people hung up soon after the question. Count them yourself, because no outside figure will fit your list. These calls show which lines sound like a script. Rewrite one line, test it in the browser, and score the next batch.

Related reading: AI Voice Agent Latency Explained and the Response Times Customers Will Tolerate, How to Clone a Team Member's Voice for AI Calls Safely and With Consent and 9 Design Principles for Natural-Sounding Conversational AI in the Contact Center.

Questions people also ask

How do you answer "where did you get my number" on a sales call?

Name the source in the first sentence, such as the sign-up, the form or the referral. Then say who is calling and why. Add the date so the person can check.

What is the best reply to "how did you get my number" in cold calling?

Give the list type and make removal easy. Say "a business directory" or "a purchased list" and offer to delete the number now. Do not hint at a relationship that does not exist.

Does an AI caller have to say it is an AI?

It depends on where you call. The EU AI Act and Colorado's Chatbot Safety Act require disclosure. Set your agent to answer truthfully whenever asked. Disclosing first is the safest habit.

Is it legal for AI to make outbound calls without consent?

Often not. In the US, the FCC ruled in 2024 that AI voices fall under the TCPA's rules for artificial or prerecorded voices. These generally require prior express consent. Rules vary elsewhere, so ask counsel.

What should the AI do if it does not know where the number came from?

Say so plainly and stop guessing. Design the agent to offer a human callback when it lacks context. Then log the call so someone can find the real source afterward.

How do you stop a customer from thinking an AI call is a scam?

Give the brand, the reason and the source in the opening lines. Then offer a callback on a published number. People want a way to reach a human. Never ask for sensitive details.

What should happen after someone asks to be removed from the list?

Confirm the removal aloud, end the call politely and add the number to the do-not-contact list at once. A copy of a campaign leaves out anyone added to that list since the original.

Insight7 campaigns carry consent marking and a do-not-contact list, and every call can be scored against your scorecard. This week, write the real lead source, sign-up date and consent status into every contact record of your top campaign.

Book a demo of Insight7