9 Mechanism Script Structures

Never rely on a winning ad — I don't care if it's booking 15 estimates a week. The moment you stop testing, you're one algorithm shift away from a dead week and a crew sitting idle. Most contractors have one ad working and 12 that aren't. That's not a testing system. That's a gamble with a good streak.

The ad dies — fatigue, platform change, season shifts. Nothing behind it. Estimator calendar goes from full to three appointments. At $8K average job, that's two or three crew days lost. Not because the market dried up — because the pipeline was a single lane. "We found what works" is a warning, not a finish line.

When something works you leave it alone — that's triage, not laziness. The problem is "leave it alone" only works if a system is loading the next angle in the background. Most contractors also think testing means guessing. What I'm showing flips that: the homeowner market tells you which angles work — you just give it enough to vote on.

Estimator calendar stops being a source of anxiety — multiple angles loaded, when one softens another picks up. CPL stabilizes instead of drifting up. You learn your market (which fear closes fastest in your geography). And you stop being the bottleneck — system runs the test, market scores it, you review a one-pager and make one decision: cut or scale.

Two rules before anything else: never use your business name as your Facebook page name (your page name is audience targeting in disguise). And never use Lead Gen as your campaign objective — not for foundation repair, not for waterproofing, not for any ticket north of $5K. Those two rules connect directly to why your pixel might be trained on the wrong person right now.

Meta's algorithm optimizes for exactly what you told it. You said "get me form fills" — it found people who fill out every free form they see. That's your pixel's model of your customer. Every dollar you spend going forward finds more people who look like them. The longer you run it that way, the deeper the hole. Meta isn't broken — it's doing exactly what you told it to do.

"Facebook leads are just low quality now" — that's not a Meta problem, that's a signal problem. Meta can only optimize for what you tell it. Six months of Lead Gen objective = pixel trained on the wrong person. The fix isn't a new platform or more budget. It's changing what you send back to the algorithm.

CPL goes up — fewer leads but they're real. Setter converts 8/10 instead of 3/10. Cost per booked estimate drops as Meta builds a model from your actual closed jobs. "These leads are garbage" objection disappears. And the 14M+ closed-won jobs anchor accelerates the calibration — you're not starting from scratch, you're starting with a model that already knows what a closed foundation job looks like.

If you're sending paid traffic to a form, you're burning money. Lead comes in at 7:14 PM. Office picks it up Thursday morning. Homeowner already booked with your competitor. The form didn't lose the job — the gap did. Here's how to close that gap with AI, without sounding like a robot, without your office touching a lead until it's pre-qualified.

Lead hits a form, goes to an inbox, someone has to decide: is this real? So either they respond slow (12 other things going on) or they put a tire-kicker on the calendar because they don't want to miss a job — and your estimator drives 45 minutes to quote a $600 crack fill for a price shopper. That's your office bottleneck. Not a people problem. A process problem. Costing you booked estimates, estimator hours, and close rate simultaneously.

"My office already handles this" — the question isn't whether they're doing it, it's how fast and how consistently. If response time varies by person, time of day, or how busy the week is, you have a variable. Variables leak revenue. "AI will give bad advice" — the AI isn't diagnosing anything. It collects four pieces of information. The moment someone asks "is this serious?" the AI says "that's exactly what we want our estimator to find out — let me grab a couple details first."

9:47 PM Wednesday. Homeowner DMs "FOUNDATION." AI responds in 20 seconds. By 10:02, intake packet complete — project, zip confirmed, homeowner, active urgency. "We'll have someone call you tomorrow morning." Setter gets Slack notification. Thursday 8:15 AM, setter calls. Lead picks up, is expecting the call. Estimate booked for Friday. Estimator walks in knowing the full context. Job closes. Compare that to the same lead hitting a form at 9:47 PM.

Never leave a voicemail. Nobody picks up a number they don't recognize. When your office calls a fresh lead, hits voicemail, and moves on — that lead is already gone. They've texted three other guys. Here's how to respond to a quote request in under 60 seconds, qualify the job before anyone opens a calendar, and book the estimate — without your office watching anything.

The 2-minute window is real. Inside 2 minutes, contact rate is roughly 8× higher than waiting 30 minutes. The homeowner didn't submit one form — they probably submitted three. Yours, the guy ranked below you, whoever had a targeted ad in their feed. First one to talk to them wins the estimate. If your process is: lead sits in inbox, office checks it in the morning, calls during business hours, leaves voicemail — you've already lost.

"Our office follows up when they can" — at a growing company, "when they can" is 3–5 hours after the lead came in. "Homeowners will wait for a good contractor" — true for referrals and repeat customers who know you. Not true for someone who found you through an ad. You haven't earned the wait yet. Speed is how you prove yourself before you've ever met them.

Show rate goes from 55–70% to 80–85%. At 20 estimates a month, that's 3–4 more jobs you're actually in the room for. Calendar quality improves — qualification gate stops trucks going to tire-kickers. Office load shifts: team stops chasing 12 overnight leads, starts booking confirmed estimates with complete packets. Speed-to-lead stops depending on who's free.

If you're leaving a quote appointment without presenting a proposal in person, you're already behind. The contractor who presents in person with options, on the spot, gets called back first. Never leave without good/better/best. Never present one price. You're not losing on price — you're losing on positioning.

Homeowner gets your PDF, two other PDFs. All three have a number, a scope line, a signature box. They're not making a spreadsheet — they're calling back the guy who felt right. If your quote looks like a line item, you're in a price war you didn't sign up for. The quote-to-job gap isn't about price — it's about trust. Trust starts dying the second you drive off without presenting options.

"The appointment went great — I sold it in the home." You didn't sell it if you left without a signature. You had a good conversation. "A quote is a quote." No — a quote is a guess about what the homeowner cares about. One price tells them the only thing that matters is that number. Three prices let them buy up and feel like they made a decision instead of just agreeing to a number.

Close rate up 10–15 points — not because you got cheaper, but because you stopped being forgettable. Quote-to-job gap compresses — homeowners decide faster when they trust the person who showed them the evidence. You stop competing on price as often. Contractors stop saying "we lost it on price" and start saying "we lost it because follow-up was too slow" — which is a fixable problem.

Two rules, day one: never start a job without a before video and a referral ask. Never end a job without an after video, a review, a testimonial, and another referral ask. Most guys follow zero of them. Every completed job you close without capturing proof paid you once. You left two, maybe three more on the table.

You spent $8K–$40K getting that homeowner to say yes. Job is done. You pack up and drive to the next one. No before. No after. No review. No referral ask. That job is dead — paid you once. Meanwhile that customer is talking to their neighbor who has the same foundation crack. Maybe they mentioned your name. Maybe not. You don't know because you never asked. You're running the same Facebook ads, paying the same lead cost, grinding for the same new customer — when the guy who would've referred three jobs was standing right there.

"We do good work — word of mouth handles it." Word of mouth is invisible. Can't track it, can't scale it, can't turn it into an ad, can't count on it to fill next quarter's calendar. "We'll remember to ask for reviews." No you won't — not consistently. Job finishes, everyone's tired, another job to get to. "We'll remember" means maybe 20% of the time, halfhearted, converts at a fraction of a proper ask. The reason isn't laziness — nobody built the play.

$3M waterproofing company, 15 jobs/month. Before: 3–4 Google reviews/month, ad creative unchanged for 6 months. After the play: 12–15 new reviews/month, rating goes 4.1 → 4.6, two referral jobs a week, ad account rotating fresh before/after every two weeks, close rate on new estimates goes up because estimator is pulling up local job proof on a tablet in the homeowner's kitchen. Cost per booked job dropped — not from more ad spend, but from ads working better and referrals filling the calendar.

One rule: never log into more than one place to answer a business question. If you're the one connecting your ad account to your CRM to your call tracker to your calendar to QuickBooks — you're not running a business. You ARE the system. The moment you take a vacation, go on a job, or have a bad Monday, the business goes blind.

The "6-login morning": ad account, CRM, call system, calendar, QuickBooks, spreadsheet — 45 minutes and you still don't know if your show rate dropped this week or which leads from Monday are still alive. You're the integration layer. You hold this together in your head. That's not a system — that's a liability.

"More dashboards means more control" — you want visibility, so you go to the source. "My business is too complex for one tool" — it's not. Leads in, they get called, some book, some show, some close, you do the job, collect, ask for a review. The problem isn't complexity — nobody set it up to be seen in one place.

Monday morning. One view on your phone: 14 leads last week, 11 contacted in 5 min, 8 booked, 6 showed, 4 closed, $8,400 average job, $610 cost per closed job, 4.8× ROAS. One sales guy ran 3 estimates, closed 1 — that's a coaching conversation. Another ran 3, closed 3. Show rate dipped 75% vs 84% last week — check alert log, confirmation texts didn't go out because CSM was out Thursday. Fix: automate the confirmation. Didn't log into 6 places. Didn't call 3 people. Looked at one screen, made 2 decisions, went to work.

One rule I never break: I never find out something broke from a customer complaint. The alert fires first. Always. Most contractors running any kind of automated system don't actually know when it stops working. They find out two weeks later when close rate is down and they're reverse-engineering what happened. By then the leads are gone, the jobs are gone, and you're just doing math on a loss.

Automated follow-up breaks Thursday. Leads come in Thursday, Friday, Saturday. Nobody contacts them. You don't notice because booked calls lag. By Monday's report, 12–18 leads got zero contact in their first 5 minutes. Contact rate drops by half when you miss that window. At $8K average ticket and normal close rate — $30K–$50K in potential revenue gone and you didn't feel it happen. That's delayed discovery cost. A dashboard tells you what happened. A fire alarm tells you what's happening.

"We review the monthly report" — that's a post-mortem. You're reading the autopsy. "We have dashboards — we know what's going on." Dashboards require someone to look at them, at the right time, with enough context to know something's wrong. Nobody pulls dashboards on a random Tuesday at 2 PM when the issue started Sunday. The real reason: you built the system, it worked, you assumed it would keep working. Systems drift. Integrations break. Workflows fall out of sync. You need something that watches it for you and screams when something's off.

Wednesday afternoon, workflow silently stops — integration hiccup between CRM and SMS. Old world: you find out Friday morning when booked calls are low. You've lost two days. New world: 40 minutes after the workflow stops, Slack alert fires. Ops person checks, restarts integration, manually pushes the 6 leads who didn't get touched. Total downtime: 40 minutes. Another: show rate drops 18 points in 4 days. Alert fires on day 2. Pull the booked calls from that window — all from a new Facebook campaign targeting outside your service area. You shut it down Wednesday, not Friday. Saved a week of ad spend. Catch it while the job is still saveable.

One rule I run every operation by: never make the same mistake twice. Not "try harder next time." Not "I'll talk to the team." The system catches it, the system fixes it, the system checks that it's fixed. If your business is making the same mistake in week 6 that it made in week 1 — that's not a people problem. That's a system that has no memory. Three moves to build that memory.

Lead falls through a crack Monday. You catch it Wednesday, say something in the group chat, team says "won't happen again." Next Monday — same crack, different lead. Or estimate goes out wrong. You fix it. Two weeks later, different estimator, same error. You're the error-catching system right now. Every mistake lives in your head, not the workflow. When you scale that workflow, you scale the mistakes. More volume means more of the same cracks, running faster.

"They just need to pay more attention." "Better training will fix it." If paying attention were the fix, you'd have fixed it already. Your team isn't forgetting on purpose — the process never told them what to do when something goes sideways. There's no place to log it, no one to route it to, no check the following week to see if it's still happening. The workflow has no nervous system. That's what we're building.

Tuesday. CSM notices three appointments this week had wrong crew lead assigned. Logs it — customer impact, caught internally. Run the 3 questions: process exists but the reassignment column is buried and only checked if someone remembers. Workflow gap. Fix: reassignment column becomes a required field before appointment can be marked confirmed. 20 minutes to set up. One week later, ops checks the log — zero crew lead assignment errors. Fix held. Logged resolved. Nobody got yelled at. Nobody had a demoralizing conversation. The process got smarter. That's what a business with a nervous system feels like.