HVAC Software for 16+ Techs: When the System Becomes the Company

Sixteen technicians is a convenient marker, but the real threshold is structural. You have crossed it when no single person can answer, without opening a report, how a given department performed last month. At six trucks the owner knows. At twelve a good general manager still mostly knows. At twenty-five nobody knows, and the company starts making decisions off whatever the system happens to show.

Three specific things break at this size, and they break in a way that more effort will not fix. The first is that departments start subsidizing each other invisibly. Service, install, maintenance, and commercial have genuinely different cost structures and margins, but if they share one bucket of revenue and one bucket of cost, a strong service department will quietly fund an install department that is losing money on every job. Companies discover this after two years of growing revenue and flat profit.

The second is that your average stops describing anybody. Company-wide revenue per technician of $220,000 can be five people at $310,000 and five at $130,000, and those are two completely different management problems sitting inside one number that looks fine. The third is that tribal knowledge stops transferring. At six trucks the new hire learns by riding with the owner. At twenty-five he learns from whichever technician had room in the truck, and your quality becomes a function of who trained whom.

Every meaningful decision below follows from those three. You are not buying features at this size. You are buying separation, visibility, and repeatability.

For a 25-technician HVAC contractor, ServiceTitan typically lands between $9,000 and $14,000 a month, which is $108,000 to $168,000 a year on an annual contract. Per-technician pricing generally runs $350 to $500 depending on modules and term. That number is real, but it is also the smallest of the three costs you are signing up for, and the other two are what determine whether the investment works.

The second cost is implementation, which runs from $5,000 at the low end to $50,000 or more for a complex multi-department rollout, and takes three to six months of real calendar time. During that window your reporting is unreliable, because you are running two sets of habits at once. Budget for a dip. Companies that plan for a flat quarter through conversion come out fine; companies that planned for immediate gains panic in month two and start half-abandoning the rollout, which is the single most common way this money gets wasted.

The third cost is the one nobody quotes you: somebody has to own the system. Not use it - own it. Build the price book, maintain job types and profit centers, run the reports, retrain when configuration drifts. At sixteen-plus technicians this is a real role, either a dedicated operations or systems manager at $70,000 to $95,000, or a defined half of an existing manager job with the other half genuinely taken away. Companies that skip this do not get a bad platform. They get a good platform configured by nobody in particular, which produces reports that are wrong in ways that are hard to detect, which is worse than no reports at all.

Add it up honestly: for 25 technicians, roughly $110,000 to $170,000 in license, $10,000 to $50,000 amortized implementation, and $40,000 to $95,000 of internal ownership. Call it $170,000 to $300,000 a year in the first year against $6 million to $8 million of revenue - about 2 to 4 percent. That is defensible if the system produces a two-point margin improvement, and it usually can. It is indefensible if you buy the license and skip the ownership.

The single highest-value configuration decision at this size is how you separate the business into profit centers, because that separation is the limit of what you will ever be able to see. Get it right and you find out within a quarter which parts of the company make money. Get it wrong and you spend $150,000 a year on a platform that reports the same blur your old system did.

The minimum useful split for a residential-and-light-commercial contractor is four: service, replacement and install, maintenance agreements, and commercial. If you run new construction, that is a fifth and it should never be blended with retrofit. Each one needs its own revenue, its own direct labor, its own material cost, and its own gross margin - and the labor allocation has to be honest, which is where most setups quietly fail. A service technician pulled onto an install crew for two days has to land in install, or install margin will look better than it is and service will look worse.

What surfaces when this is done properly is consistent enough to predict. Service usually carries the highest gross margin, commonly 55 to 65 percent on a fully loaded basis. Replacement runs 40 to 50 percent and is far more sensitive to how tightly you buy equipment than most owners assume. Maintenance agreements look modest in isolation, often 30 to 45 percent, and are undervalued by that number, because their real return is the replacement pipeline and the off-season labor absorption they create. Commercial has the widest spread of all - excellent when scoped and terrible when a residential process gets applied to a building.

The test of whether your configuration is real is simple. Ask for last month gross margin for each department, and see whether you get four numbers within a minute or an argument about how labor gets counted. If it is the argument, the platform is not your problem.

At sixteen-plus trucks you have two to five CSRs handling somewhere between 800 and 2,000 inbound calls a month, and that room is one of the two or three largest uncontrolled variables in the business. Booking rate across a group of CSRs routinely varies fifteen to twenty points between the best and the worst performer, and it varies quietly, because the calls that do not book leave no trace unless something is deliberately recording them.

Run the number on your own volume. At 1,200 calls a month with a company booking rate of 55 percent, moving to 70 percent is 180 additional booked jobs a month. At a $450 average service ticket that is roughly $81,000 a month of revenue that already called you. Even discounting hard for calls that were never bookable, the recoverable portion of that dwarfs any software line item on your budget, and it dwarfs most of what you spend to generate new calls in the first place.

The move at this size is call scoring and coaching rather than more headcount. Tools such as Avoca run roughly $300 to $800 a month for a shop in the five to twenty-five technician range, and higher with volume, and what they buy you is every call scored on the things that decide bookings: was the appointment actually offered, was the objection handled, was the maintenance agreement mentioned, did the caller get put on hold and vanish. That turns your best CSR into a documented standard instead of a person other people vaguely try to imitate.

Keep after-hours separate in your head from daytime coverage. Your daytime problem is conversion quality. Your nights-and-weekends problem is coverage, and it is still a coverage problem at this size - shops in this trade miss 25 to 40 percent of inbound calls, and the misses cluster at exactly the hours homeowners call. An AI receptionist tier at $249 or more for the volumes you run costs a fraction of the after-hours calls it catches, and it is the correct answer for the overflow and the 7 p.m. no-heat call, not a replacement for the CSRs who handle your daytime conversion.

A dispatcher who is also answering phones saturates somewhere around six to ten technicians. A dispatcher doing nothing but dispatch handles fifteen to twenty. Past twenty technicians you are running either multiple dispatchers or a dispatcher plus a service manager, and the job changes character entirely: it stops being about filling today and starts being about protecting capacity across the week.

The concrete change is board separation. Service, install, and maintenance belong on separate boards with separate people, because they schedule on incompatible logic - service is same-day and unpredictable, install is multi-day and scheduled weeks out, maintenance is seasonal batch work that exists to absorb slow periods. Running them on one board means every emergency call cannibalizes install crews and every install week starves service. This is the single most common structural mistake at twenty trucks, and it costs far more than any software decision.

Drive time is where the money leaks and where the measurement is worth the trouble. Technicians average about 5.7 billable hours out of 8.8 paid, and drive time typically consumes around 28 percent of the day. Tightly zoned dispatch pulls that to 18 to 20 percent; ad-hoc assignment pushes it to 32 to 35. On twenty-five technicians, ten points of drive time is roughly 0.85 hours a day per technician, or 21 technician-hours a day - about two and a half additional technicians worth of capacity that you already employ and are currently spending on the highway.

On routing software specifically: a per-vehicle route optimizer at $49 to $149 per vehicle per month solves a different problem than yours. It is built for planned multi-stop routes, and your service board is reactive by nature. Get zoning and board separation right inside your main platform first. Standalone routing earns its place mainly on the maintenance side, where you actually do have a planned list of tune-ups to sequence, and that is the only place worth piloting it.

Second locations fail for a consistent reason, and it is almost never demand. It is that the second location is measured on consolidated numbers, so nobody can tell for eighteen months whether it works. The parent operation carries it, revenue grows, margin sags a point or two, and by the time the picture is clear you have built habits around a location that never earned its keep.

Before opening or acquiring, decide how the location is measured and configure it before day one, not after. Each location needs its own revenue, its own direct labor, its own allocated overhead, and its own marketing spend and lead attribution. Allocated overhead is the part people skip because it is arguable, and skipping it is exactly what makes a new location look profitable for two years. Pick a defensible basis - revenue share or headcount share - write it down, and hold it steady so the trend line means something even if the absolute number is debatable.

What genuinely centralizes is the call center, the price book, purchasing, and marketing. What has to stay local is dispatch and field management, because dispatch depends on knowing traffic, neighborhoods, and which technician handles which situation well, and none of that survives a two-hour drive. The common failure is doing the opposite - centralizing dispatch because the software makes it easy, while letting each location drift into its own pricing because nobody enforced the book.

On the platform side, verify how the system handles this before you sign, not after. Ask for a demo showing per-location P&L, technician transfer between locations mid-week, per-location price variation on a shared book, and consolidated reporting that still drills down. These behaviors vary more between platforms than the feature lists suggest, and they are extremely expensive to discover you cannot do in month eight.

At $8 million and up you are usually not short of leads in the abstract. You are short of knowing which spend produced booked, profitable revenue, and that gap gets expensive fast because the budget is now big enough to hide a lot of waste. Full-service home-services marketing agencies run $2,000 to $10,000 or more a month at this size, and that money is well or badly spent depending almost entirely on whether you measure the right thing.

Track cost per booked job, not cost per lead, and carry it through to gross profit by channel. A channel producing leads at $40 that book at 20 percent into $400 tickets is losing to a channel producing $120 leads that book at 60 percent into $1,100 tickets, and cost-per-lead reporting will tell you the opposite every single month. This requires the call center to tag source at the call and the platform to carry that tag through to the closed job, which is a configuration and discipline problem more than a software problem.

The most underused asset you own at this size is your existing customer database. A company at twenty-five technicians typically has 15,000 to 40,000 past customers, and the replacement candidates inside that list are the cheapest revenue available anywhere - they know you, and you already paid to acquire them. Database marketing platforms in the $800 to $4,000 a month range price off list size and exist to work exactly this: system age targeting, post-service follow-up sequences, maintenance conversion, and lapsed-customer reactivation. Run this before increasing any top-of-funnel spend, because the same dollar works several times harder against a warm list.

Reviews shift purpose here too. At five trucks reviews are how you get found. At twenty-five with a few hundred reviews you are already found, and the job becomes protecting the rating and covering every location and service area - multi-location review management typically runs $299 to $1,200 a month. The number to watch is not total reviews, it is your rolling ninety-day average, because that is what a homeowner comparing three companies actually sees, and a strong lifetime average will hide a department that started slipping four months ago.

At twenty-five technicians with normal industry turnover you need to replace six to eight people a year before you have added a single truck to grow. There is no labor market that supplies that reliably at the quality you want, which means at this size technician development stops being an HR nicety and becomes a production requirement. Companies that grow past this line without solving it end up capping out - not because they cannot sell the work, but because they cannot staff it without lowering the bar, and lowering the bar shows up as callbacks and margin loss within two seasons.

The structural version of the answer is an apprentice pipeline: a defined path from helper to installer to service technician with written competencies at each stage, so a person can see their next step and you can see who is ready. Structured training platforms in the $35 to $150 per user per month range make this practical by replacing ride-alongs with a curriculum and simulations, which matters most for the scenarios a new technician will otherwise meet for the first time in a customer basement.

Pair development with measurement or it drifts. A diagnostic standard at $30 to $120 per technician per month makes system performance a measurement rather than an opinion, which does two things at once: it protects quality as the bench gets younger, and it gives you an objective basis for advancement that everyone can see is fair. Both effects matter more at twenty-five technicians than at eight, because at eight the owner can just tell who is good.

The economics are not subtle. A technician producing $250,000 a year against one producing $150,000 is $100,000 of revenue and roughly $50,000 of gross profit, every year, from the same truck and the same overhead. Moving four people up one tier is worth more than any software purchase on this page, and training spend of $1,000 to $3,000 per technician per year is one of the few line items in this business with a return that obvious.

For a 25-technician residential-and-light-commercial contractor, a realistic all-in software budget is $13,000 to $22,000 a month. That is roughly platform $9,000 to $14,000, call scoring $500 to $1,000, after-hours AI receptionist $250 to $600, review management $300 to $800, database marketing $800 to $2,500, training $900 to $2,500, diagnostics $750 to $1,500, and photo documentation and accounting filling the remainder. On $8 million of revenue that is about 2 to 3.3 percent, which is the normal band at this size, and it excludes agency marketing spend, which belongs in the marketing budget rather than the software budget.

Keep the proportions in view. Card processing at 2.6 to 3.5 percent on $8 million of revenue is $208,000 to $280,000 a year, which is larger than your entire software stack. At this volume you have real negotiating leverage on effective rate, and a quarter point is $20,000 a year. That conversation is worth more than any comparison between platforms in the same tier.

Sequence the first year so the measurement comes before the spending. Months one through three: instrument what you have. Get four departmental gross margins, revenue and billable hours per technician, booking rate by CSR, and drive time percentage - even if you have to assemble them by hand. Almost every company doing this for the first time finds at least one department losing money and at least one CSR twenty points off the pace, and both are fixable before you buy anything.

Months four through six: fix structure, not software. Separate the boards, tighten zoning, name the system owner, and put the price book under one person. Months seven through nine: if the platform decision is still open, run it now with your own numbers in hand and configure profit centers before go-live rather than after, because retrofitting them means your first year of history is unusable. Months ten through twelve: layer the specialist tools onto a working foundation - call scoring, database marketing, training pipeline - in that order, because each one produces returns that are easier to see once the underlying reporting is trustworthy.

The version of this that fails is the reverse: buy the enterprise platform first, hope the structure follows, and discover in month nine that a $150,000 system is reproducing the old blur at higher resolution. The structure is what you are actually buying. The software just makes it enforceable.