HiringContractorsTrades
|
August 11, 2026
|
8 min read

Hiring in the Trades: How to Stop Losing Good Applicants Before You Call Them Back

Good techs apply to three companies at lunch and take whoever texts back first. A simple setup that replies in two minutes, screens by text, and books interviews people actually show up for.

You posted the job three weeks ago. Eleven people applied. You called four of them back, two never picked up, one had already taken another job, and the fourth showed up to the interview, seemed great, and then stopped answering his phone. Meanwhile you're turning down work because you don't have the hands to do it.

If that sounds familiar, you're in good company. Associated Builders and Contractors projects the industry needs roughly 349,000 net new workers in 2026, and around 78% of construction firms say they have trouble filling hourly craft positions. But here's the part most owners miss: a big chunk of the problem isn't that good people don't exist. It's that the good ones applied, waited two days to hear from you, and took the other offer.

Speed is the whole game now. And speed is exactly the kind of thing you can hand to software.

Why good applicants disappear

Think about who you're trying to hire. A solid HVAC tech with five years under his belt is not sitting at a desk refreshing his email. He's on a roof. He applied from his phone during lunch, probably to three or four companies at once, and whoever texts him first gets the conversation.

Recruiters who work the trades all say the same thing: the companies with the worst ghosting rates are the ones with the longest gaps between "applied" and "heard something back." Candidate ghosting searches have jumped enormously in the last year, and more than half of job seekers say they got ghosted by an employer. It cuts both ways, but in a tight labor market the applicant has the leverage.

Then there's the format problem. If your application process needs a laptop, a resume file, and a quiet hour, you have screened out most of your target pool before you ever looked at anybody. Electricians and welders and equipment operators work ten hour days. Email is not where they live. Text is.

So you have two things working against you at once: you're slow, and you're asking people to communicate in a channel they don't use. Fixing both is not complicated.

What "AI for hiring" actually means here

Set aside the buzzwords. For a shop with 8 or 30 or 60 people, AI hiring means a few small, boring automations that run whether or not you're free:

An applicant hits submit. Within a couple of minutes, they get a friendly text from your company. Not a generic auto-reply, a real message that names the role and asks a few questions. They answer by text, on the roof, between calls. If the answers line up with what you need, they get offered three interview slots that match your actual calendar. They pick one. They get a reminder the day before and the morning of.

You show up to an interview with someone who is qualified, still interested, and expecting you.

That's it. There's no robot deciding who to hire. The software is doing the part you keep meaning to do at 9pm and never get to.

The four pieces worth building

1. Instant first contact, by text

This is the single highest-value change, and it's the same idea behind missed-call text-back for contractors, just pointed at applicants instead of customers.

When someone applies, they get a text inside two minutes:

"Hi Marcus, this is Dana at Ridgeline Mechanical. Thanks for applying for the HVAC service tech role. Got 2 minutes to answer 4 quick questions by text?"

Two things happen. You've jumped the line ahead of every other company that will email him tomorrow. And you've signaled that you're organized, which matters more than owners think. A candidate reads a fast, clear response as a preview of what working for you is like.

2. Screening questions that filter on facts, not vibes

Keep it to four or five questions with clear right answers. For a service tech role that might be:

  • Do you have a valid driver's license with fewer than two points?
  • How many years have you worked on residential systems on your own?
  • Do you hold an EPA 608 certification, and which type?
  • Can you be on site in the northwest suburbs by 7am?
  • What's your target hourly rate?

The AI asks these one at a time by text, waits for real answers, and handles the normal messiness of people replying "yeah I got the 608 universal" instead of ticking a box. Then it sorts the responses into two buckets and gives you a short summary of each person.

Important: keep the screening to things you can defend, like licenses, certifications, availability, and years of hands-on experience. Do not let a tool rank candidates on personality, tone, or anything it inferred about them. Several states now regulate automated hiring decisions, and the safe design is simple. The software gathers facts and organizes them. You decide who to interview.

3. Interviews that book themselves

Once someone clears the questions, the system offers real time slots pulled from your calendar. Candidate picks one, it goes on your schedule, and both of you get a confirmation.

This kills the phone tag that costs you three days. It also protects the shop owner who is genuinely busy. You do not have to be the bottleneck between a qualified guy and a Tuesday morning conversation.

Add a reminder the day before and one the morning of. Interview no-shows drop noticeably when someone gets a short, human sounding text the night before: "See you tomorrow at 9 at the Elmwood shop, ask for Dana at the front. Reply here if anything changes."

4. Follow up with the ones you didn't hire

Every applicant who was close but not right goes into a list. Six weeks later, when you land a big job and need two more bodies, you text that list first. Most shops throw this away and start from scratch every time.

That's the same instinct behind winning back past customers. You already paid to get these people's attention once. Don't let it evaporate.

What this is worth in dollars

Do the math on your own shop, but here's a version that's typical for a 20 person mechanical contractor.

An open service tech seat costs you roughly $1,200 to $2,000 a week in work you can't take or have to sub out. If your average time to fill is six weeks and you cut it to three, that's somewhere around $4,000 to $6,000 recovered per hire. Fill four seats a year and you're looking at real money.

Then there's your time. Owners and office managers routinely burn four to six hours per open role on phone tag, reading applications, and rescheduling interviews. Automating first contact and booking takes most of that back.

The tooling itself is not expensive. Most shops can put this together for somewhere between $100 and $400 a month using a texting platform, a scheduling tool, and an AI layer that handles the conversation. If you want a fuller picture of what these builds run, we broke that down in how much AI automation costs for a small business.

How to start this month

You don't need a system. You need one week of setup.

Week 1. Write down the four or five questions that actually determine whether someone can do the job. Not nice-to-haves. The things that make it a no. Then write the first text message you'd want an applicant to receive. Read it out loud. If it sounds like a form letter, rewrite it.

Week 2. Pick a texting tool your team will actually check, connect it to wherever applications land, and set the auto-text to fire on submission. Do this part manually first if you have to. Even a person copy pasting the same message within ten minutes beats what you're doing now.

Week 3. Add the screening conversation and the calendar link. Test it on yourself and on a buddy. Fix the spots where it sounds like a robot.

Week 4. Turn on the reminders and start your "close but not now" list.

Track two numbers the whole time: how many minutes pass between an application and your first reply, and what percentage of scheduled interviews actually show up. If the first number drops under five and the second climbs above 70%, the thing is working.

A word on what not to automate

Do not automate the offer. Do not automate reference calls. Do not send a rejection that reads like it came from a machine, because the trades are a small world and the guy you brushed off is friends with the guy you want next year.

And be honest in the job post. Real pay range, real hours, real truck situation. The fastest response in the world doesn't help if the person shows up and finds out the job isn't what you advertised.

The point of all this is not to remove yourself from hiring. It's the opposite. You get to spend your judgment on the ten minute conversation that actually matters, instead of on voicemail tag.

If you'd like help setting this up for your shop, that's the kind of thing I do with small businesses. Happy to look at how you're hiring now and tell you where the leaks are.

See it

If you want the whole thing on one page, I sketched the flow from new application to booked interview here: Hiring flow for contractors.

Tom Karels

I help companies turn AI into measurable financial impact. For SMBs, that means automating real workflows, saving real hours, and freeing up teams to grow. For enterprise teams, it means embedding AI into sales, operations, and delivery so the value shows up in lower costs, higher productivity, and revenue growth.

Related Posts