Your technician is on site fixing a leaking water heater. The customer mentions their AC hasn’t been serviced in three years. The tech nods, finishes the job, collects payment, and leaves. Two weeks later, that same customer calls a competitor for AC maintenance. You just lost $800 in work that was sitting right there.
This happens dozens of times a month in most trades businesses. Not because your crew doesn’t care, but because they’re focused on the job at hand. They don’t have the customer’s service history memorized. They don’t know the furnace is twelve years old or that the last plumber recommended a sump pump upgrade. They’re there to fix what’s broken, get paid, and move to the next call.
The cost adds up fast. A typical HVAC or plumbing business doing $2M annually leaves $50,000 to $120,000 on the table every year in missed upsells and add-ons. Electrical and roofing firms see similar leakage. It’s not dramatic, one-time losses. It’s a steady drip of revenue walking out the door because no one prompted the conversation at the right moment.
AI can fix this. Not with generic scripts or clunky checklists, but by analyzing every customer’s job history, equipment age, and service intervals in real time and delivering specific, contextual prompts to the technician’s phone while they’re still on site. This is what an Omni Ops agent does for trades businesses, and it’s one of the highest-return automations you can deploy.
The Manual Reality of Upselling in the Field
Most trades businesses try one of three approaches to drive upsells. None of them work reliably.
The first is the verbal reminder during dispatch. The owner or dispatcher tells the tech, “Hey, if you see anything else while you’re there, mention it.” The tech agrees, gets to the job, deals with a flooded basement or a tripped breaker, and forgets. There’s no system. It’s hope disguised as process.
The second is the printed checklist. The tech is supposed to walk through a list of common add-ons for every service call. In practice, the checklist stays in the truck or gets filled out in the parking lot after the fact. Customers can tell when someone is reading from a script, and it kills trust. The tech knows this, so they skip it.
The third approach is commission-based incentives. Pay the tech a percentage of any upsell they close. This works for some crews, but it creates uneven results. Your best technician might be terrible at sales. Your most persuasive tech might cherry-pick the easy upsells and ignore the ones that take explanation. You end up with a few winners and a lot of missed opportunities.
None of these approaches give the technician the information they actually need. They don’t know the water heater is nine years old. They don’t know the customer asked about a whole-home generator last spring. They don’t know the HVAC system is due for a tune-up in six weeks. That context lives in your dispatch software, your CRM, or the owner’s memory, and it never makes it to the field in a usable form.
What AI-Driven Upsell Prompts Look Like
An AI agent built for this use case sits between your dispatch system, your service history, and your technician’s phone. It watches every job that gets scheduled. When a tech arrives on site, the agent pulls the customer’s full history, cross-references equipment age and service intervals, and sends a short message with two or three specific recommendations.
The message might say, “Customer’s furnace is 11 years old, last serviced 14 months ago. Recommend scheduling a tune-up before winter. Water heater installed 2019, no issues. Ask about adding a maintenance plan.”
The tech reads it in ten seconds. They have the context they need to start a conversation that feels natural, not scripted. The customer hears, “I noticed your furnace is getting up there in age, and it’s been a while since the last service. Want me to take a quick look while I’m here?” That’s not a sales pitch. It’s helpful.
If the customer says yes, the tech completes the add-on work or schedules a follow-up. If the customer says no, the agent logs it and sets a reminder to follow up in three months. Either way, the conversation happened. You didn’t leave money on the table because the tech forgot or didn’t have the information.
This is what we build at Enterprise DNA when a trades business books the AI audit for trades businesses. We map out every customer touchpoint where upsell opportunities get missed, then design agents that surface the right information at the right time. For most firms, technician prompts are one of the top three revenue recovery levers.
The Three Layers of Upsell Intelligence
A good upsell agent doesn’t just remind the tech to ask questions. It builds recommendations from three layers of data, and it gets smarter over time.
The first layer is service history. Every completed job, every part replaced, every maintenance visit. The agent knows when equipment was installed, when it was last serviced, and what the technician noted during the last call. If a plumber recommended a sump pump replacement six months ago and the customer declined, the agent flags it again. If an HVAC tech installed a new furnace two years ago, the agent knows it’s time for the first tune-up.
The second layer is equipment age and lifecycle. The agent tracks manufacturer recommendations and typical replacement timelines for every type of equipment you service. A twelve-year-old water heater gets flagged differently than a three-year-old unit. A roof that’s eighteen years old gets a different prompt than one that’s eight. The agent doesn’t guess. It uses the data you already have and fills in gaps with industry-standard benchmarks.
The third layer is customer behavior. Did this customer buy a maintenance plan last year? Did they accept the last three upsell recommendations? Did they decline and then call a competitor? The agent learns which customers respond to proactive suggestions and which ones prefer to be left alone. Over time, it tunes the prompts so your techs aren’t wasting time on conversations that won’t convert.
Most trades businesses don’t have this intelligence accessible during the service call. It exists in fragments across three different systems, and no one has time to pull it together. The AI does it automatically, every time.
Real-World Impact on Revenue
Let’s walk through the numbers for a mid-sized HVAC company doing $3M in annual revenue. They run eight trucks, average 40 service calls a week, and close about 60% of their estimates. Before deploying an upsell agent, their add-on rate was around 8%. One in twelve service calls resulted in additional work beyond the original dispatch.
After three months with AI-driven prompts, their add-on rate climbed to 18%. The agent identified upsell opportunities on 35% of calls and prompted the tech with specific recommendations. The tech started the conversation on 28% of those calls. Half of those conversations converted to same-day add-on work or a scheduled follow-up.
The math: 40 calls a week, 35% flagged by the AI, 28% prompted, 50% conversion. That’s five additional upsells per week. Average ticket for an add-on in their market was $650. Five calls times $650 times 50 weeks is $162,500 in annual revenue that wasn’t there before. The agent cost them $1,200 a month to run. ROI in the first quarter.
The bigger win wasn’t just the revenue. It was the consistency. Every tech, regardless of experience or sales ability, got the same high-quality prompts. The rookies performed like veterans. The veterans stopped relying on gut feel and started using data. Customer satisfaction went up because the recommendations felt relevant, not random.
One of the HVAC owners in our network describes it this way: “We went from hoping our guys would remember to mention something, to knowing they had the right information every single time. It’s not about making them better salespeople. It’s about giving them the context to be better technicians.”
How This Fits Into the Broader Omni System
Technician upsell prompts are one agent in a larger system. Most trades businesses deploy three to five agents in the first 90 days, and they work together to close the revenue leaks that cost you six figures a year.
The 24/7 Dispatch Voice Agent answers every call, qualifies the job, and books it directly into your dispatch tool. That’s the front door. You stop missing calls, and you stop losing $500 to $3,000 per missed emergency job.
The Estimate Follow-Up Agent tracks every estimate that goes out and follows up on day two, day five, and day fourteen. Trades businesses typically see 15% to 25% of stale estimates convert when someone actually follows up. That’s $30,000 to $80,000 a year for a $2M firm.
The Review and Reactivation Agent asks every happy customer for a review the day after the job closes and reactivates past customers at the right service interval. More reviews mean more inbound calls. Reactivation brings back 10% to 15% of your old customer base without spending a dollar on ads.
The upsell agent sits in the middle. It makes sure that when your crew is on site, they don’t leave money on the table. These four agents together typically recover $80,000 to $200,000 in annual revenue for businesses in the $1M to $5M range. Larger firms see proportionally higher returns.
You can read more about how these agents connect on the Omni overview page, or dive into the specific automation categories in our insights library.
What an Omni Audit Looks Like for This Use Case
When a trades business books an Omni Audit, we spend 60 minutes mapping out where revenue is leaking and which agents will close the gaps. For technician upsells, we ask four questions.
First, how many service calls do you run per week, and what’s your current add-on rate? Most owners don’t track this precisely, so we estimate it from job counts and revenue mix. If you’re running 50 calls a week and closing five add-ons, that’s 10%. Industry range is 8% to 20%, so there’s room to move.
Second, what data do you have on customer history and equipment age? If it’s in your dispatch software or CRM, we can connect to it. If it’s in the owner’s head or scattered across spreadsheets, we build a lightweight data layer to capture it going forward. The agent gets smarter as it collects more history.
Third, how do your techs currently receive job information? Text, email, a dispatch app? We design the prompt delivery to fit their workflow. If they’re already checking their phone between calls, that’s where the prompts go. If they use tablets in the truck, we push it there. The goal is zero friction.
Fourth, what’s your average add-on ticket size, and what types of upsells convert best? Maintenance plans, equipment upgrades, same-day repairs? We tune the agent’s recommendations to prioritize the high-value, high-conversion opportunities first.
At the end of the audit, you get three outputs. A process map showing where upsells are getting missed. A one-page agent design specifying exactly what the AI will do, what data it needs, and how it integrates with your systems. A 90-day deployment plan with milestones, costs, and expected ROI.
No deck. No follow-up meeting to “discuss next steps.” You walk out with a blueprint you can execute, whether you build it with us or hand it to your internal team. Most trades businesses choose to move forward because the ROI is clear and the deployment is faster than trying to cobble it together in-house.
Practical Steps to Start Capturing Upsells Today
If you’re not ready to deploy an AI agent yet, there are three manual steps you can take this week to stop leaving money on the table.
First, pull a list of every customer who’s had service in the past 24 months and tag the ones with equipment over eight years old. Call them. Don’t wait for them to have a problem. Offer a pre-season tune-up or a free inspection. Half won’t answer, but 10% will book, and that’s found revenue.
Second, create a one-page cheat sheet for your techs with the five most common upsells in your trade and the trigger conditions for each. Water heater over ten years old? Recommend replacement. HVAC system with no maintenance plan? Offer one. Electrical panel from the 1980s? Flag it for an upgrade quote. Print it, laminate it, put it in every truck.
Third, set a weekly reminder to review completed jobs and identify the ones where an upsell was obvious but didn’t happen. Call the customer yourself. “Hey, our tech mentioned your furnace is getting up there. Want me to send someone out to give you a quote on a replacement before winter hits?” You’ll close 20% of those calls, and your techs will start noticing the pattern.
These three steps won’t scale, but they’ll prove the revenue is there. Once you see $5,000 or $10,000 in recovered upsells from a month of manual effort, the case for automating it becomes obvious.
We’ve also put together a worksheet that walks through after-hours call recovery, which ties into the same system that powers upsell prompts. You can grab the After-Hours Call Recovery Plan for Trades and use it to map out where you’re losing calls outside business hours. It’s a practical checklist, not a sales piece.
Why This Works Better Than Training or Incentives Alone
Every trades business tries to solve the upsell problem with better training or bigger commissions. Both help, but neither fixes the root issue, which is information asymmetry. Your tech doesn’t have the data they need when they need it.
Training teaches your crew what to look for and how to start the conversation. That’s valuable. But if they don’t know the customer’s history, they’re guessing. They might recommend a maintenance plan to someone who already bought one last year. They might skip mentioning a furnace tune-up because they assume someone else already did. Training gives them the skills. It doesn’t give them the context.
Incentives motivate your crew to look for opportunities. Also valuable. But if the tech has to dig through three systems to figure out what to recommend, they won’t do it on a busy day. Incentives work when the path to the commission is clear and fast. If it takes ten minutes of research to find an upsell, the tech will move on to the next call.
AI solves the information problem. It does the research in two seconds and hands the tech a ready-to-use recommendation. Training and incentives still matter, but now they’re working with good data instead of fighting against friction.
The best results come from combining all three. Train your crew on how to have the conversation. Incentivize them to close the upsell. Use AI to surface the opportunity in the first place. That’s the system we help trades businesses build during the AI audit for trades businesses, and it’s why the ROI shows up fast.
Common Objections and How to Think About Them
The most common pushback we hear is, “My guys won’t use it.” Fair concern. If the prompts are clunky or slow, they’ll ignore them. That’s why delivery method matters. We don’t build agents that require the tech to log into a new app or check a dashboard. The prompt comes as a text message or a push notification to the tool they’re already using. It takes five seconds to read. If it’s useful, they’ll use it. If it’s not, they won’t, and we tune it until it is.
The second objection is, “We don’t have clean data.” Also fair. Most trades businesses have customer history scattered across dispatch software, QuickBooks, and paper invoices in a filing cabinet. You don’t need perfect data to start. The agent can work with partial information and get smarter as you capture more. We’ve deployed upsell agents for firms that didn’t even have a CRM. We built a lightweight data layer as part of the deployment, and it paid for itself in six weeks.
The third objection is cost. A custom-built upsell agent typically runs $800 to $1,500 per month, depending on call volume and integration complexity. For a $2M trades business, recovering $80,000 to $120,000 in annual upsells, that’s a 5x to 10x return in the first year. The math works if you’re doing more than $1M in revenue. Below that, you’re better off with the manual steps outlined earlier.
The Next 90 Days
If you’re serious about stopping the upsell leakage, the next 90 days break into three phases.
Enterprise DNA put together a free field guide on exactly this: the full Claude ecosystem, Claude Code, and how to roll agents out without breaking things. Get the guide.
Phase two is deployment. Build the agent, connect it to your dispatch and service history systems, and test it with two or three techs for two weeks. Tune the prompts based on their feedback. Roll it out to the full crew once it’s working smoothly. This phase takes four to six weeks.
Phase three is optimization. Track add-on rates, revenue per call, and conversion rates by tech. Identify which prompts are working and which ones are getting ignored. Adjust the agent’s logic to prioritize high-conversion opportunities. This is ongoing, but the big gains show up in the first 60 days.
Most trades businesses see measurable revenue lift by week eight. The agent is live, the techs are using it, and the add-on rate is climbing. By week twelve, the ROI is clear, and you’re looking at which other agents to deploy next.
You can explore the full range of automation options on the Omni platform page, or browse case studies and how-to guides in our learning library. If you want to move fast, the audit is the best starting point. Sixty minutes, three outputs, no deck. Let’s map out where your revenue is leaking and how to close it.