Home ›
Blog ›
AI Cold Calling Scripts That Convert
Cold Calling
AI Cold Calling Scripts That Convert in 2026
By Roman Stanek
Last updated: April 15, 2026
~10 min read
I've run over 10,000 AI cold calls for Australian tradies. Most scripts I started with were too long, too polished, and too robotic. The ones that actually book meetings are direct, specific, and short. Here's the exact anatomy that works — and the 4 real script variants I tested, with the results.
10,000+
AI calls made for AU tradies
Amy production data
7.2%
Best script booking rate
Script D — direct variant
<30s
Optimal opener length
First word to the ask
Why Most AI Cold Calling Scripts Fail
Before the scripts, let's cover the mistakes that kill conversion. They're consistent across every industry.
Too long. The average AI cold call script I see in the wild takes 90 seconds just to get to the ask. By then, 60% of prospects have mentally checked out. You have 15-20 seconds to earn the right to stay on the line. That's it.
Too robotic. Patterns like "Hello, my name is Amy and I'm calling on behalf of..." trigger the "it's a spam call" reflex immediately. Natural openers sound like a person who already knows something about you.
No pattern interrupt. If the first sentence sounds like every other cold call, the brain categorises it as a cold call and switches to rejection mode. You need something that stops the automatic hang-up instinct.
No inline objection handling. Human callers adapt on the fly. AI can't — it follows a decision tree. If your script doesn't anticipate "I'm not interested" or "send me an email," you lose the call. The objection handling must be baked into the script.
Features before relevance. "We use advanced AI-powered technology to..." is worthless until the prospect understands why this call is relevant to them. Lead with relevance, not features.
The Anatomy of a Script That Works
5-Part Structure (Total: Under 90 Seconds)
1. Pattern Interrupt
Say something unexpected. Reference something specific about them — their suburb, their trade, something local. Not a compliment. A specific observation. This buys you 5 more seconds.
2. Leverage
One sentence that tells them what's at stake for them. Not your product's features — what they're losing or what they could gain. Keep it concrete: "$42K a year in missed calls" beats "improve your business communications."
3. Credibility
One specific proof point. A number. A result. A name they'd recognise. This is where social proof lives — make it real, not vague.
4. The Ask
One specific, low-friction ask. Not "would you be interested in learning more?" That's too vague. "Can I show you this in 10 minutes?" or "Can I send you a short demo?" — specific, time-bounded, easy to say yes to.
5. Objection Handling
Pre-built responses to the top 3 objections for your market. "Not interested" → redirect to the offer. "Send me an email" → book a call first. "Already have someone" → ask what they'd improve. Each response leads back to the ask.
Amy's 4 Script Variants (The Real Ones)
I built and tested 4 script variants for Amy — the AI caller I run for Australian tradies. Each was tested on 500+ calls. Here are the actual scripts.
Amy (opener):
"Hey, is this [Business Name]? Great — quick one. My name's Amy, I'm calling from [your business]. We build free demo websites for trade businesses in [Suburb]. We actually put one together for your business already based on what's working for plumbers in your area — it's ready to go. Would you have 10 minutes this week to take a look?"
Note: "Free demo website" is a tangible, specific offer. Much better than "can I tell you about our services." The word "already" creates curiosity.
Objection — "I'm not interested":
"No worries at all. Just so I know — do you have a website at the moment, or is that something you're looking to sort out?"
→ If yes: "Perfect — I'd love to show you what we did differently for [Competitor suburb] plumbers. Still just 10 minutes."
→ If no: "That's actually why I called — we built something for your business, no cost to look at it. Can I send you a link?"
Objection — "Send me an email":
"Happy to. What's the best email to send it to?" [collect email] "Perfect — I'll also include a link to book 10 minutes if you want to go through it together. What days work best for you this week?"
Note: Always pivot email requests back to a booking. Emails alone convert at under 1%.
Amy (opener):
"Hi there, hope you're having a good morning. This is Amy from [your business] — we help trade businesses in Australia get more customers using AI. I was wondering if you'd have a few minutes to chat about how that might work for your business?"
Why it underperforms: "Hope you're having a good morning" is a cold call cliché — everyone knows what follows. "I was wondering if you'd have a few minutes" is an easy no. The offer is vague ("AI," "more customers"). Tradies need specific, not conceptual.
Objection — "Not interested":
"That's completely understandable. Could I just ask — is there anything in particular that's holding back your business growth at the moment?"
Why this fails: Asking a probing question after a "not interested" feels intrusive. Tradies respond to this by hanging up.
Amy (opener):
"Hey, this is Amy. Quick question — how many calls does your business miss per day when you're on the job? I ask because we're working with electricians in [Suburb] and the average is 3-4 missed calls per day — about $42,000 a year in jobs that go to someone else. We built a system that answers those calls automatically. Can I show you how it works — takes 10 minutes?"
Why it works: Leads with a question that creates self-reflection. Uses a specific stat ($42K) that hits hard. The offer follows the pain — it feels like the solution, not a pitch.
Objection — "We don't miss calls":
"That's great — honestly, you're ahead of most. When you're on-site or under something, does the call go to your mobile directly or to voicemail?"
→ If voicemail: "How many of those actually leave a message? Our data shows 62% of callers hang up rather than leave a voicemail. I can show you how we capture those — still 10 minutes?"
→ If mobile: "And what's your answer rate when you're mid-job? Even 80% means 20% of calls aren't getting through. Worth 10 minutes to see what that costs?"
Amy (opener):
"Hey, this is Amy — I'll keep this quick. I run AI systems for trade businesses in [Suburb]. Built one for a plumber in [Nearby Suburb] — went from missing 4 calls a day to answering every single one automatically. Booked 11 extra jobs in the first month. Can I show you the same thing for your business? Takes 10 minutes, I'll show you exactly what it does."
Why it wins: Ultra-short opener. Specific result (11 jobs, first month). Direct ask with a time boundary. No fluff. AU tradies respect directness — they don't have time for corporate speak mid-morning.
Objection — "Not interested":
"Fair enough — one thing though. If there were 4 calls going to voicemail every day right now, would that be worth fixing?"
→ If yes: "That's exactly what the system does. 10 minutes — I can show you the actual call recordings so you can hear it working. What day suits you?"
→ If no: "No worries. If that changes, we're at automatewithroman.com. Cheers."
Note: Script D ends hard objections cleanly. No desperate over-handling. Tradies respect a caller who takes no for an answer gracefully — it builds trust for the automatewithroman.com mention at the end.
Objection — "I already have someone handling this":
"Good to know — and I don't want to waste your time. Quick question: what's the one thing you'd change about how they're doing it?"
→ This opens the conversation. Most "I already have someone" responses are deflections, not genuine satisfaction. The question surfaces the real objection.
A/B Test Results: What the Data Showed
After 2,240 calls across 4 scripts, here's the clean comparison:
Script D — Direct Close
580 calls
7.2%
Script C — Pain First
540 calls
5.6%
Script A — Website Offer
600 calls
4.8%
Script B — Soft Approach
520 calls
3.8%
Key finding: Directness beats softness for AU tradies by nearly 2:1. Tradies are busy people. They respect callers who get to the point and don't waste time. Script B's "hope you're having a good morning" opener was actively hurting conversion — it signals the call is a pitch, not a relevant conversation.
What to Never Say in an AI Cold Calling Script
Say This
- Specific numbers ("11 jobs in the first month")
- Time boundaries ("takes 10 minutes")
- Local references (suburb names, trade-specific pain)
- Direct asks ("Can I show you this week?")
- Clean exits ("No worries — cheers")
Never Say This
- "Is this a good time?" (easy no)
- "I know you're busy" (signals insecurity)
- "I just wanted to reach out" (meaningless)
- "We're the leading provider of..." (credibility-kill)
- "Would you be interested in learning more?" (too vague)
- Any sentence over 20 words
Adapting Scripts for Different Trades
The structure stays the same. The specific details change. Here's what to customise for each trade:
- Plumbers: Emphasis on emergency calls and burst pipe urgency. "A customer with a burst pipe calls you. Goes to voicemail. They call the next plumber." Urgency angle works well.
- Electricians: Emphasise switchboard upgrades and the compliance angle. Property managers respond to "we've put together a compliance quote for your suburb's most common switchboard issues."
- Painters: Property managers and real estate agents are the best target, not homeowners. "We work with PMs in [Suburb] who need reliable painters for end-of-lease repaints." Reliability angle over price.
- Builders: Focus on architects and project managers rather than end consumers. B2B referral angle: "We work with architects who need reliable builders on retainer for new residential projects in [Area]."
- Landscapers: Strata companies and body corporates. "We maintain common areas for 12 strata properties in [Suburb] — do you handle anything similar in your area?"
Call Timing for Australian Tradies
Script quality and call timing are equally important. The best script called at the wrong time still fails.
- 7-9 AM Sydney/Melbourne: Primary window. Tradies are awake, haven't hit the job site yet. Decision-making capacity is high. This is when Amy runs.
- 5-7 PM: Secondary window. After work. Tradespeople are home, winding down. Good for mobile numbers; less effective for business lines.
- 10 AM - 4 PM: Avoid. Most tradies are fully on-site. Answer rates drop 40-60%. Even if answered, the call is often cut short.
Read the full AI lead generation setup guide for call scheduling, lead list building, and the full infrastructure stack.
After the Script: What Happens When They Say Yes
The script ends with a booked meeting. The AI confirms the time, sends a calendar invite, and triggers an SMS with the Calendly link as backup. When the prospect doesn't show up — which happens 20-30% of the time — a follow-up SMS goes out 30 minutes after the missed meeting: "Hey, we had a 10-minute call booked — I can send you the demo recording instead if it's easier. Just reply YES."
This rescue sequence recovers 15-20% of no-shows. It's built into the system — no manual effort.
For the inbound side — handling calls that come in to your business — see the AI receptionist guide for Australian tradies.
Want the Scripts Configured for Your Trade?
I'll set up the right script variant for your trade and target suburbs, run the first 100 test calls, and refine based on real data. Book 30 minutes to see the live demo.
Apply
When AI Cold Calling Scripts Don't Apply
- High-trust, relationship-based sales — if your industry relies on long-term relationships and personal referrals (think commercial construction tenders, $500K+ contracts), cold AI calling is the wrong opener. Use warm intro approaches instead.
- Calling residential numbers without consent — AU's Do Not Call Register applies. These scripts are built for B2B outreach. Calling homeowners at random requires DNC compliance and explicit consent frameworks.
- Markets with under 500 prospects — if your total serviceable market is a list of 200 businesses, bulk AI calling exhausts the list quickly. Use targeted manual outreach for small markets.
- Industries where AI disclosure is legally required — some regulated industries require callers to identify as AI upfront (financial services, healthcare in certain jurisdictions). Check your specific compliance requirements before deploying.
- Complex multi-stakeholder sales — if closing a deal requires buy-in from 5 people and a 6-week procurement cycle, a 10-minute demo close is unrealistic. Use the AI to book a discovery call, not a close.
Frequently Asked Questions
What makes an AI cold calling script different from a human one?
AI scripts must be shorter, more direct, and pre-handle the top objections inline — because AI can't improvise the way an experienced human caller can. Every response path needs to be mapped in advance.
How long should an AI cold calling script be?
Opener to ask: under 30 seconds. Full script including one handled objection: under 90 seconds. If you're still talking at 2 minutes and haven't gotten a yes or hard no, the prospect has already checked out.
Which script variant works best for Australian tradies?
Script D (direct close) — 7.2% booking rate vs 3.8% for the softly worded Script B. Tradies respond to directness. They're not in the business of polite conversation during the workday.
What's the best time to AI cold call Australian tradies?
7-9 AM Sydney time is the primary window — before they're on the tools. Secondary: 5-7 PM after work. Mid-morning to mid-afternoon call rates drop 40-60% as people are fully focused on the job.
What should you never say in an AI cold calling script?
"Is this a good time?" (easy no), "I know you're busy" (insecurity signal), "I just wanted to reach out" (meaningless), "We're the leading provider of..." (credibility-kill), and anything vague about "learning more." Plain, direct, specific language only.
Want This Built for Your Tradie Business?
I set up Amy — the AI cold caller for AU tradies — end-to-end. Voice, scripts, scraper, CRM. Apply for a 30-min fit call.
Apply to Work 1-on-1 with Roman
Or join my free community — AI Mastery Genesis on Skool — where I drop the templates I use to build these agents.