You run three Google Ads campaigns, a Yelp profile, a yard-sign program, and a truck-wrap fleet. Calls come in. Jobs get booked. Invoices get paid. But when you sit down at the end of the quarter and try to figure out which channel actually made you money, you’re staring at a spreadsheet that tells you how many clicks you bought, not how many water heaters you installed.

Most trades business owners I talk to can tell me their total marketing spend within a few hundred dollars. Almost none can tell me which channel delivered the highest margin per acquired customer. The gap between those two numbers costs the typical $3M plumbing or HVAC shop somewhere between $50,000 and $200,000 a year in wasted ad spend, under-invested winners, and guesswork.

The problem isn’t effort. It’s that the manual work required to connect a phone call on Tuesday to a booked job on Thursday to a paid invoice three weeks later involves touching four systems, remembering to tag the lead source in your dispatch software, and hoping your CSR wrote it down correctly when the customer called at 4:47 p.m. on a Friday.

AI call tracking changes that. Not the old call-tracking number swap that tells you which campaign rang the phone. I mean an agent that listens to the call, extracts the lead source if the customer mentions it, tags the job automatically, watches whether it converts to a booked appointment, tracks the invoice, and calculates your true cost-per-acquired-job across every channel without anyone on your team lifting a finger.

That’s what this article walks through. How the tracking works, what it replaces, and what it looks like when an AI agent handles the whole loop from ring to revenue attribution.

The manual tracking tax

Here’s the workflow most trades shops run today.

A homeowner calls your main line. Your CSR or the owner answers, takes the details, creates a job ticket in the dispatch system. If they remember and if there’s a field for it and if the customer mentioned how they found you, the CSR types “Google” or “referral” into a dropdown. That dropdown has eight options. Three of them mean the same thing. One is labeled “other.”

The call converts. Your tech goes out, diagnoses a failed compressor, writes an estimate for $4,200. The estimate sits in the system. No one follows up on day two. No one follows up on day five. The customer calls a competitor. You never know whether that lead came from the $1,200 you spent on Facebook that month or the $150 Nextdoor ad.

Now multiply that by 60 inbound calls a week. Half of those calls happen when your CSR is on another line or your office is closed. A quarter of them don’t leave a voicemail. You’re missing the lead source on 40% of the calls that do convert, because the customer didn’t volunteer it and your team didn’t ask.

At the end of the month, you open your Google Ads dashboard. It says you got 43 clicks and spent $890. You open your dispatch software. It says you booked 11 jobs tagged “Google” and invoiced $31,000. Sounds great, except you also booked 19 jobs tagged “unknown” that month, and three of them were $8,000+ commercial HVAC replacements. Were any of those Google leads? No idea.

You’re flying blind. You keep spending on every channel because you can’t prove which ones don’t work. The channels that do work don’t get more budget because you can’t prove they’re the winners. That’s the manual tracking tax.

What AI call tracking actually does

An AI call-tracking agent sits on your phone line and listens to every inbound call in real time. It transcribes the conversation, identifies whether the caller is a new lead or an existing customer, extracts the lead source if mentioned, qualifies the job type and urgency, and writes a structured record into your dispatch system before your tech even rolls the truck.

If the caller says “I saw your truck on Maple Street” or “I found you on Google,” the agent tags it. If they don’t mention it, the agent asks: “Just so we can send the right crew, how did you hear about us?” Most customers answer. The agent logs it.

The same agent tracks what happens next. Did the call convert to a booked appointment? Did the tech write an estimate? Did the estimate convert to an invoice? Did the invoice get paid? Every step gets recorded with the original lead-source tag attached.

At the end of the month, you open a dashboard that shows you cost-per-lead, lead-to-appointment conversion rate, appointment-to-invoice conversion rate, and total revenue by channel. You see that your Google Ads delivered 14 leads at $63 each, nine of them booked, six invoiced, and $22,000 in collected revenue. Your Yelp profile delivered three leads, two booked, one invoiced for $1,400. Your truck wraps delivered eight leads, six booked, five invoiced for $19,000, and cost you nothing but the original wrap install.

Now you know where to spend more and where to cut. That’s what AI call tracking does.

The three parts of the tracking loop

The system breaks into three parts: call capture, conversion tracking, and revenue attribution.

Call capture is the 24/7 Dispatch Voice Agent. It answers every inbound call, even when your office is closed or your CSR is on another line. It qualifies the job, asks for the lead source, books the appointment directly into your dispatch calendar, and sends the customer a confirmation text. The agent doesn’t just log the call. It handles the entire intake, so you never lose a lead to voicemail or a busy signal.

One HVAC company in our network installed this agent and saw after-hours bookings jump from two a month to 11 in the first 30 days. Every one of those calls included a tagged lead source, because the agent asked for it as part of the intake script.

Conversion tracking is the Estimate Follow-Up Agent. It watches every estimate your techs write. If the estimate hasn’t converted to an invoice within 48 hours, the agent sends a follow-up text or email. Day five, it follows up again with a slightly different message. Day 14, it makes a final offer or asks if the customer wants to reschedule.

The agent tracks which follow-ups convert and ties the conversion back to the original lead source. If a Google lead converts on day six after two follow-ups, you see that in the attribution report. If a referral lead converts immediately without follow-up, you see that too. Now you know not just which channels bring in leads, but which channels bring in leads that close fast and which ones need more nurturing.

Revenue attribution is the reporting layer. The system pulls invoice data from your accounting or dispatch software, matches it to the original call record, and calculates total revenue and margin by channel. You see cost-per-acquired-job, not just cost-per-click. You see lifetime value by source if you track repeat customers. You see which campaigns deliver emergency calls that convert at 80% versus scheduled maintenance calls that convert at 40%.

All three parts run automatically. No one on your team has to tag a lead, remember to follow up, or build a pivot table at month-end. The agent does it.

If you want a step-by-step checklist for capturing after-hours calls and tagging them correctly, grab the After-Hours Call Recovery Plan for Trades. It’s a one-page worksheet that maps the intake questions, the tagging logic, and the follow-up sequence most trades shops use to recover 15-25% more revenue from evening and weekend leads.

What this replaces in your business

Right now, someone on your team is doing five things manually.

First, they’re answering the phone or listening to voicemails and trying to figure out where the lead came from. If the customer doesn’t say, they guess or leave it blank. That’s 10-15 minutes per day of partial data entry.

Second, they’re manually tagging lead sources in your dispatch system. If your system even has a lead-source field. If they remember to fill it in. If the dropdown options match your actual marketing channels. Most don’t. Most shops have a “Google” option that could mean Google Ads, Google Local Services, or organic search, and no one knows which.

Third, they’re trying to reconcile your ad spend with your booked jobs at the end of the month. That’s an hour of spreadsheet work, pulling reports from three platforms, matching phone numbers, and guessing at the leads that didn’t get tagged. The result is a rough estimate that’s wrong by 20-30%.

Fourth, they’re following up on estimates manually, if they follow up at all. Most shops send the estimate and hope. The ones that do follow up are doing it with a reminder in Outlook or a sticky note on the desk. It’s inconsistent, it’s late, and it doesn’t track which follow-ups work.

Fifth, they’re making budget decisions based on gut feel, because the data isn’t clean enough to trust. You keep running the Yelp ads because you’ve always run them, not because you know they deliver a positive return.

AI call tracking replaces all five. The agent answers the call, tags the source, tracks the conversion, follows up on the estimate, and builds the attribution report. Your team stops doing data entry and starts doing the work that actually generates revenue.

Real-world example: HVAC shop in Phoenix

A $4M HVAC company in Phoenix was spending $3,200 a month across Google Ads, Yelp, and a local radio spot. The owner knew his total marketing budget. He didn’t know which channel was working.

His CSR was tagging lead sources manually, but only about half the time. The dispatch system had a lead-source field, but it was a free-text box, so entries included “Google,” “google,” “Google Ad,” “internet,” and “online.” Impossible to aggregate.

At the end of each month, the owner would pull a report from the dispatch system, count the jobs tagged with each source, and divide his total marketing spend by the number of jobs. That gave him a rough cost-per-lead, but it didn’t account for the 40% of jobs with no source tagged, and it didn’t track which leads actually converted to paid invoices.

We built him a 24/7 Dispatch Voice Agent that answered after-hours calls and asked every caller how they heard about the company. We added an Estimate Follow-Up Agent that tracked every estimate and followed up on day two and day five. We connected both agents to his dispatch software and built a revenue attribution dashboard that pulled invoice data and matched it to the original call.

Three months in, he had clean data for the first time. Google Ads delivered 22 leads at $87 each, 16 booked, 11 invoiced, $47,000 in revenue. Yelp delivered nine leads at $122 each, four booked, two invoiced, $6,800 in revenue. The radio spot delivered one lead that didn’t convert.

He killed the radio spot, doubled his Google Ads budget, and kept Yelp running at the same level because the leads were lower volume but higher ticket. His total marketing spend went up by $800 a month, but his revenue from tracked leads went up by $34,000 over the next quarter. Cost-per-acquired-job dropped from an estimated $290 to a measured $176.

That’s what happens when you can see the whole loop.

The Omni build for tracking marketing to revenue

When we build this system for a trades business, we start with the Omni Audit for trades businesses. It’s a 60-minute working session. No deck. Three outputs: a process map of your current lead intake and follow-up workflow, a list of the manual tracking steps that are costing you the most time or accuracy, and a one-page agent spec that defines exactly what the AI will do.

Most trades shops need two agents to close the tracking loop.

The first is the 24/7 Dispatch Voice Agent. It’s an Omni voice build. It answers your main line, qualifies the job, asks for the lead source, books the appointment in your dispatch calendar, and sends a confirmation text. It handles after-hours calls, overflow calls, and any call your CSR can’t pick up. Every call gets tagged with a source, a job type, and a timestamp.

The second is the Estimate Follow-Up Agent. It’s an Omni ops build. It watches your dispatch or quoting system, tracks every estimate that goes out, and follows up automatically on day two, day five, and day 14. It logs every follow-up attempt and every conversion, so you can see which lead sources convert fast and which ones need more touches.

Both agents write structured data into your existing dispatch system. No new software to learn. No separate CRM to maintain. The agents integrate with ServiceTitan, Housecall Pro, Jobber, FieldEdge, or whatever you’re already using.

The third piece is the attribution dashboard. We pull invoice data from your accounting system, match it to the call and estimate records the agents created, and build a report that shows cost-per-lead, conversion rate, and total revenue by channel. You get a link to a live dashboard that updates daily. No month-end spreadsheet work.

The whole build takes four to six weeks from audit to launch. The agents start tracking on day one. You see your first clean attribution report at the end of week two.

If you want to see what that looks like for your business, book a 60-min Omni Audit. We’ll map your current workflow, identify the tracking gaps that are costing you the most, and spec the agents that close the loop. No cost for the audit. You walk out with a process map, a gap analysis, and a one-page build spec you can use whether you build with us or not.

Why this matters more than you think

Most trades business owners underestimate how much revenue they’re leaving on the table by not tracking lead source to invoice. The visible cost is the ad spend on channels that don’t work. That’s real, and it’s usually $500 to $2,000 a month for a typical $2M-5M shop.

The hidden cost is bigger. It’s the high-performing channels you’re under-investing in because you can’t prove they work. It’s the leads that convert on day six after two follow-ups, but you never followed up so they went to a competitor. It’s the emergency calls that come in at 7 p.m. on a Saturday and go to voicemail because your CSR is off, and you never knew they called.

We see shops recover $4,000 to $15,000 a month in previously lost revenue just by capturing after-hours calls and tagging them correctly. The attribution piece adds another layer. Once you know which channels deliver the highest margin per acquired customer, you can shift budget from the losers to the winners and grow revenue without spending more.

One electrical contractor in our network was spending $1,800 a month on Facebook ads and $900 on Google Local Services. He thought Facebook was working because he got a lot of likes and comments. The attribution data showed Facebook delivered 11 leads, three booked, one invoiced for $850. Google Local Services delivered six leads, five booked, four invoiced for $9,200. He flipped the budget. Three months later, his revenue from tracked leads was up 40% and his cost-per-acquired-job was down 28%.

That’s the difference between guessing and knowing.

What happens after the agents launch

The first week after launch, the agents are learning your business. The voice agent is handling calls and booking appointments. The follow-up agent is tracking estimates and sending messages. You’re watching the logs and making small adjustments to the scripts and the timing.

Week two, you see your first attribution report. It’s rough because you only have two weeks of data, but you can already see which channels are tagging correctly and which leads are converting. You spot a pattern: your Google Ads leads are booking at 70%, but only 40% are converting to invoices. Your referral leads are booking at 90% and converting at 80%. That tells you something about lead quality and pricing.

Week four, you have a full month of clean data. You can see cost-per-lead, lead-to-appointment rate, appointment-to-invoice rate, and total revenue by channel. You make your first budget decision based on real numbers. You kill one underperforming campaign and shift the budget to a winner.

Month two, the agents are running on autopilot. Your CSR isn’t tagging leads anymore. Your admin isn’t following up on estimates anymore. Your owner isn’t building spreadsheets at month-end anymore. The agents are doing it, and the data is cleaner than it’s ever been.

Month three, you’re optimizing. You see that leads from your truck wraps convert faster than leads from paid ads, so you add two more trucks to the wrap program. You see that follow-up on day five converts better than follow-up on day two for HVAC replacement estimates, so you adjust the agent’s timing. You see that your Yelp leads are high-ticket but low-volume, so you keep the profile active but don’t increase spend.

Six months in, you’ve shifted 30% of your marketing budget from guesswork to data-driven allocation. Your cost-per-acquired-job is down. Your revenue per lead is up. You’re not spending more on marketing. You’re spending smarter.

That’s what AI call tracking delivers. Not just better data. Better decisions. Better margins. More revenue from the same budget.

The next step

If you’re running a trades business and you can’t tell me which marketing channel delivered your three highest-margin jobs last month, you’re losing money every week. The fix isn’t more marketing spend. It’s better tracking.

AI call tracking closes the loop from ring to revenue attribution. The agents answer the calls, tag the sources, track the conversions, follow up on the estimates, and build the reports. You stop guessing and start knowing.

The AI audit for trades businesses is where we start. It’s 60 minutes. We map your current lead intake, identify the tracking gaps, and spec the agents that close the loop. You walk out with a process map, a gap analysis, and a one-page build spec. No deck. No sales pitch. Just a clear picture of what’s possible and what it takes to build it.

Book my Omni Audit and we’ll map the path from your current tracking workflow to a system that shows you true cost-per-acquired-job across every channel. If you want to see more about how we build AI systems for trades businesses, visit our insights library or explore the Omni platform that powers the agents.

The gap between what you’re spending on marketing and what you know about which channels work is costing you $50,000 to $200,000 a year. Let’s close it.