I Gave My Real Phone Number to a Stranger's Bot

It Wasn't a Test. That's the Point.

I'm on a coaching call with a guy who just joined the mastermind. He's got a digital marketing agency - AI voice agents, local SEO, Google My Business, the works. He's been a construction surveyor for 40 years, knows the industry cold, and he's betting his next chapter on a suite of AI-powered services for builders, plumbers, electricians. Smart niche. Real experience. Good instincts.

We're about 30 minutes in. I'm walking him through the three-channel system - cold calling, live events, cold email - and he mentions he's already got a working voice agent demo live on his company website. I ask if I can see it. He gives me the URL. I pull it up.

A little chat bubble appears in the bottom left corner. I click it.

"Hi, you've reached MLDG Solutions UK. I am Jeanette. How may I help you today?"

I type: "Hey, I'm looking to buy. Can you tell me more about your pricing?"

The bot handles it cleanly. Says it can't provide specific pricing but offers to connect me with a specialist. Asks if I want to book a call. I say sure. It asks morning or afternoon. I say 10 a.m. It asks for my name, phone number, and email to confirm.

And I gave them.

My actual name. My actual cell number. My actual Gmail address.

Not a burner. Not a test account. Not "John Smith." Alex Berman. +1 972 922 9817. The real ones.

I didn't plan to. I wasn't trying to prove a point. I just... did it. Because the conversation felt natural enough that I forgot I was stress-testing someone's demo on a live coaching call.

That's the whole post right there. Everything else is just unpacking why that matters more than any conversion stat, any manufactured case study, or any testimonial he could ever put on a landing page.

The Involuntary Opt-In

There's a concept I want to name because I haven't seen anyone else talk about it directly: the involuntary opt-in.

It's the moment a skeptical, busy, distracted person gives something real - their time, their attention, their actual contact information - without consciously deciding to. Not because they were persuaded. Not because the copy was clever. Because the product just worked well enough that their guard dropped.

Most product demos never get there. Most AI chatbots, most voice agents, most "revolutionary" tools get stress-tested with fake names and joke inputs the second a technical person touches them. Because technical people - and salespeople especially - are trained to probe, to poke, to look for the seams. We're always half-outside the experience, evaluating it.

When that stops happening, when you stop evaluating and start just using - that's the involuntary opt-in. And it's the only honest signal that a product actually works.

I test a lot of software. I've built and sold companies in this space. I know what bad AI looks like. I've sat through a hundred demos of voice agents and chatbots that sounded great in a controlled environment and fell apart the second a real prospect pushed back even slightly. The hallucinations. The weird loops. The "I'm sorry, I didn't understand that" on the most basic follow-up question.

So when I got on this guy's website, I wasn't going in naive. I was going in as exactly the kind of person his prospects will be: experienced, slightly skeptical, moving fast, not particularly interested in being impressed.

And I gave the bot my real phone number.

That's a case study. Full stop.

Why Most AI Demos Fail This Test

Let me be specific about what usually goes wrong, because this guy's agent got a lot of things right that most people get wrong.

First problem with most voice and chat agents: they hallucinate on pricing. Someone asks "how much does this cost" and the bot either makes up a number, gives a number that's wrong, or - worst of all - confidently states something that directly contradicts what's on the website. Instant trust killer. This agent did the right thing: it acknowledged the question, said it couldn't give specifics, and immediately pivoted to a human handoff. That's the correct answer. It didn't dodge the question - it reframed it as a reason to talk to someone.

Second problem: most bots lose the thread when the conversation shifts. You ask about pricing, they answer about pricing, and then when you say "okay let's book a call" they start over from scratch like you never said anything. This one maintained context through the whole conversation. Booking flow felt continuous, not like a hard redirect.

Third problem - and this is the silent killer - most of these agents feel like forms dressed up as conversation. You can feel the decision tree. Every response has that slightly robotic cadence that says "I am processing your input and selecting from a finite list of responses." This one didn't have that. The conversation felt like a conversation. Not perfect, not indistinguishable from human, but natural enough that I stayed in it.

That third one is the hardest to engineer and the most important to get right. Because the moment a prospect feels the bot, they start playing with it instead of using it. And once they're playing with it, you've lost the sale.

The Niche Was Right. The Network Was the First Move.

Here's what I told him on the call, and I want to spell it out because I see people skip this step constantly: before you build any outbound system, before you set up domains, before you run a single cold email campaign - you go to the five people you already know who could buy this.

He told me he could probably name five people in construction who'd take a meeting. I told him that's where we start. Not because five customers makes a business, but because those five people will show you problems with your positioning, your landing page, and your offer that you literally cannot see from the inside. They'll ask about the thing that's missing from your homepage. They'll push back on the pricing structure in a way that tells you exactly what objection your cold emails need to pre-empt.

The warm network isn't a shortcut. It's market research you get paid for.

And in his case, the niche is genuinely smart. Forty years in construction - national housebuilders, development projects, the whole ecosystem. He knows the customer care problem from the inside: when you sell a house, you're on the hook for repairs and fittings for two years minimum. That means inbound call volume that spikes at the worst times, after-hours emergencies, and a customer service burden that's expensive to staff with humans and almost impossible to staff well at odd hours.

A voice agent that handles after-hours customer care calls for a national housebuilder? That's not a "nice to have." That's a headcount reduction with a better customer experience attached to it. The ROI writes itself.

And then you extend it: plumbers, electricians, any trade business with emergency call-outs. Same problem. Same solution. Same buyer psychology.

That's a real niche. Not "local businesses" as a category - which I'll be honest, I warned him off of - but a specific vertical where he has 40 years of relationship capital and genuine domain credibility. The voice agent is the wedge. Everything else - the SEO, the Google My Business work, the Facebook stuff - that's the upsell stack once you're in the door.

The Three-Channel System for Getting the First Meetings

For anyone in a similar position - new business, solid offer, zero pipeline - here's the exact sequence I laid out on this call.

Channel one: your existing network. Name five people who already know you and would take a meeting. Contact all five this week. Not a mass email, not a LinkedIn blast - an actual message that says here's what I'm building, I'd love 20 minutes to show you. This gives you your first feedback loop and potentially your first sale before you've written a single cold email.

Channel two: live events. For his niche specifically, I pulled up a list of construction and electrician trade shows. Here's the thing most people don't realize about trade shows: you don't need a booth. You don't need a ticket half the time. You find out which hotel the conference is at, you show up in the lobby, and you meet people. Shake hands. Have the conversation. He's the only guy in that room who has four decades of surveying experience and an AI voice agent running on his website. That combination alone is a conversation-starter that nobody else has.

Channel three: cold email. But not before the infrastructure is right. I had him in the mastermind with access to Cold Email University for a reason - the technical setup has gotten more complex than it used to be. One domain sending cold email is a liability now. You need to rotate through ten, twenty domains minimum so you're not torching your deliverability on one mistake. He'd already bought five domains, which was the right instinct. The course walks through the Outlook setup, the Google Workspace setup, getting Instantly or Smartlead configured - all of it step by step. If you're building this out yourself, Instantly and Smartlead are both solid for managing multi-inbox sending at scale.

For the actual scripts - especially when you're targeting a niche where you have real credibility - grab the Top 5 Cold Email Scripts. The ones that work in B2B services with a specific vertical lead with the outcome, not the feature. "I help national housebuilders reduce after-hours customer care costs" is a first line. "We offer AI voice agent solutions for the construction industry" is not.

The Cold Calling Problem (And the Only Real Answer)

He admitted on the call that cold calling is where he struggles. Rejection is hard. He knows the "every no gets you closer to a yes" speech and it doesn't make the dialing easier. I respect that honesty more than the people who pretend cold calling is fun.

My honest take: cold calling doesn't get easier. I've done it myself. I've hired people to do it. The discomfort doesn't disappear with volume - it just gets more familiar. The only actual exit from cold calling is either hiring someone else to do it or building enough inbound that you don't need it.

But here's why I still push it early, especially for a new business: you need meetings before you need anything else. Not a perfect website. Not a polished deck. Meetings. The fastest way to get meetings when you have no pipeline is to call people. Everything else - cold email, LinkedIn, content, trade shows - has lag time. Cold calling has same-day feedback.

My recommendation to him: set a timer for one hour. Dial until it goes off. Don't set a meeting target - targets create anxiety and excuses when you fall short. Time targets just require you to show up. Whatever comes from the hour comes from the hour. Then stop, take notes on the objections you heard, and go refine the pitch.

Once you have enough meetings that you're overwhelmed - and that's the actual goal, getting to overwhelmed - then you hire someone to handle the meetings. Then you hire someone to handle the sales calls. Then your fulfillment is already handled (his is). At that point you're managing a business, not doing all the jobs in it.

Step by step. In order. Don't skip ahead.

On Finding the Right Leads for a Niche Like This

If you're building out a list of construction companies, housebuilders, or trade businesses in a specific region, the approach I'd use is layered. Start with Google Maps data for local firms - a Google Maps scraper will pull business names, categories, and basic contact info fast. For the national and larger regional players, pull from a B2B email database filtered by industry and company size. Then verify and enrich the contacts before you mail anything - bad data in, wasted sends out.

For UK-based outreach specifically, scrub your lists carefully - the regulatory environment is different from the US and Canada, and you want to make sure you're mailing the right people through compliant channels. That's not a reason not to do cold email; it's a reason to do it right.

Back to the Bot

After I booked the appointment - after the bot confirmed my 10 a.m. slot and said it would send a confirmation to my email - I sat back and told him: I like this. I'm in on this.

And I meant it. Not as a courtesy. Not as coach-speak to keep him motivated. I meant it because I'd just done the thing his prospects are going to do, and it felt fine. It felt like booking an appointment. Which is exactly what it was.

He mentioned the margins: five to nine times on the voice agent product. His cost is seven cents a minute. He's selling packages that start at a dollar a minute - and the human alternative, at minimum, runs you four dollars a minute for someone who doesn't even know the business. The economics are obvious once you lay them out side by side. The sale isn't "AI is cool." The sale is "this is cheaper than your current solution and it works at 3 a.m."

That's the pitch. That's all the pitch needs to be.

The demo closes it. But only if the demo actually works - if the bot handles the booking flow cleanly, maintains context, doesn't hallucinate pricing, and feels like a conversation instead of a form. His does. I know because I used it. Without planning to. Without realizing I was handing over real contact information until I'd already done it.

What the Involuntary Opt-In Actually Proves

I want to be precise about what this story does and doesn't prove, because I'm not in the business of overstating things.

It doesn't prove the product will close every deal. It doesn't prove the pricing is right or the market is big enough or that the cold email campaigns will work. Those are separate questions.

What it proves - specifically and only - is that the core product experience is good enough to use. Not just good enough to demo. Good enough to use. That's a different bar and most products don't clear it.

If you're building anything right now - a SaaS, an agency service, an AI tool - run this test before you run any other test. Don't ask friends if it's good. Don't run a survey. Don't look at click-through rates or time-on-page. Find someone skeptical and experienced, put them in front of the product without warning, and watch what happens.

If they start genuinely using it - if their guard drops and they just do the thing the product is designed to make them do - you have something. If they stay in observer mode the whole time, poking at it and narrating what they're doing, you have more work to do.

The involuntary opt-in is the test. Everything else is noise.

He passed it. I just happened to be the one who proved it.

If you're building outbound systems around an offer like this - whether it's AI services, agency work, or anything B2B - the Best Lead Strategy Guide is worth reading before you launch your first campaign. And if you want the kind of real-time feedback that turns a good demo into a signed contract, that's what Galadon Gold is built for - not courses you watch alone, but live sessions where you actually send the emails, make the calls, and get the meeting on the calendar before you log off.

Because momentum beats perfection. Every time.