The Roofing Tech Stack, by Company Size
- Under 20 employees: get off paper first (Jobber, Housecall Pro, or AccuLynx), then AI call answering, review automation, and financing at the point of sale.
- Skip route optimization and anything with the word predictive in it until you're a much bigger company. Those tools solve problems you don't have yet.
- At 50+ employees the math flips: dispatch AI, job costing, and lead scoring start earning their keep.
- Four governance rules regardless of size: disclose AI on calls, no customer details in public tools, treat job photos as sensitive, a human approves every price.
- Lesson 5 of 5. Some of this you buy off the shelf, some is cheaper built custom, and that call comes down to volume.
Every roofing software vendor will tell you their product is for companies like yours. It isn’t. A tool built for a 40-crew storm chaser is a different animal than one built for a shop where the owner still climbs ladders, and buying the wrong size wastes more than money. It wastes the office’s patience for the next tool, which might have been the right one.
So this lesson sorts by size, not by feature list. What to buy at 1-20 employees, what to skip, and what starts making sense at 50 and up. If you haven’t run the rollout plan in lesson 4, read it first: the order you adopt these matters as much as what you pick.
Get off paper first.
Before any AI tool, the business needs one system of record. If jobs live on a whiteboard, quotes in a truck, and customer history in somebody’s text messages, no AI purchase will fix that. AI on top of a paper filing system is a robot filing clerk in a burning building: very fast, very organized, still on fire.
Every tool below plugs into the FSM. The text-back needs to know which number called about which job. The review ask fires when a job closes, which means jobs have to close somewhere. Build the base layer first and everything after it is a bolt-on instead of a rebuild.
Under 20 employees: buy these four, in this order.
1. The FSM base layer
Field service management software: the one place where jobs, quotes, schedules, and customers live. Jobber and Housecall Pro are the general-purpose picks and the fastest to learn. AccuLynx is roofing-specific, built around measurements, material orders, and insurance workflows. Pick one, put everything in it, and stop keeping the schedule in three places.
2. AI call answering
The phone is where roofing money dies. Everyone's on a roof at 2pm, the call rings out, and the homeowner dials the next name on the list. Call answering plus missed-call text-back means every caller gets a response even when nobody can pick up. This is the first AI purchase that pays for itself, and usually the fastest.
3. Review automation
A tool that asks every customer for a review after the job closes, automatically, and tells you fast when someone's unhappy so you can fix the job instead of reading about it later. The company outranking you on Google Maps probably doesn't do better work. They ask more consistently.
4. Financing at the point of sale
A 2,000 sq ft roof runs $10,500-$14,500 in the Milwaukee market. Very few families have that sitting in checking, and the ones who don't aren't bad leads, they're unfinanced ones. A financing option presented with the estimate turns the five-figure number into a monthly one, on the spot, without you playing bank.
And skip these, without guilt.
Route optimization
You have three trucks and a foreman who's driven every road in Waukesha County for fifteen years. The software will not beat him. This is a tool for fleets, and three trucks isn't a fleet, it's three trucks.
Anything predictive
Predictive maintenance, predictive demand, predictive whatever. These need sensor data and job volume you don't have. Roofs don't phone home. Wait until the word predictive stops being the most exciting thing about the product.
Dashboards you won't read
If nobody in the office will open it on a Tuesday, it's not reporting, it's decoration with a subscription fee. Lesson 4's three baseline numbers on a whiteboard beat an unread analytics suite.
At 50+ employees, the skipped tools come back.
The tools that were overkill for three trucks become load-bearing at a dozen crews. Same categories, different math:
Dispatch AI
With a dozen crews, subcontractors, and Wisconsin weather rearranging the week every Thursday, scheduling stops being a whiteboard problem. Dispatch tools that reshuffle crews around rain and reroute around a stuck job earn their fee at this scale, and not before.
Job costing automation
Somewhere around the fourth crew, owners stop knowing which jobs actually made money. Materials, labor hours, dump fees, and callbacks tracked per job, automatically, is how you find out that your most impressive-looking contracts are your worst margins.
Lead scoring
When leads outrun your capacity to chase them, scoring decides who gets called first: roof age, neighborhood, storm exposure. The local inputs for that model are sitting in this site's market data and storm history pages, free.
The lead-scoring inputs mentioned above live on this site: the market data for housing age by town, and the storm history for verified hail and wind events by location.
Four rules, whatever your size.
Disclose the robot
If an AI answers your phone, it says so, and the caller can always reach a human. People don't mind talking to a machine nearly as much as they mind being fooled by one.
No customer details in public tools
Names, addresses, phone numbers, and claim numbers stay out of free consumer AI tools. Use business plans with training on your data turned off, or draft with placeholders the way lesson 3's prompts are built.
Treat job photos as sensitive
Roof photos catch more than roofs: interiors through windows, kids in yards, license plates, paperwork on the kitchen table. Don't feed them to third-party tools without consent, and be picky about which platforms hold them.
A human approves every price
The AI can measure, draft, and suggest. It doesn't set the number. Every price that reaches a customer went past a person with the authority to own it. This rule has no exceptions and no expiration date.
The honest ending: some of this you shouldn’t buy.
The FSM, call answering, and review tools are solved problems. Buy them off the shelf, because building your own is a hobby disguised as a business decision. But the subscriptions stack. A follow-up sequence, a storm-event mailing list, a reporting layer that pulls your numbers into one page: at a certain volume, renting each of those monthly costs more than building a thing you own once.
Where that line sits depends on your volume, and it’s a math conversation, not a sales one. The services page shows what we build custom for roofing companies, and 15 minutes on the calendar is enough to tell you honestly which side of the line you’re on. Sometimes the answer is “keep the subscriptions,” and you’ll hear that too.
Write down every place job information lives right now: the paper folder, the whiteboard, the text threads, the foreman’s head. If the list is longer than one item, that list is your buying guide, and the FSM comes before any AI tool on this page.
Asked before you had to ask.
Do I need AccuLynx, or is Jobber enough?
Depends on your mix. If insurance and storm work is a big share of revenue, the roofing-specific tool earns its premium with measurement integrations and supplement workflows that general tools don't have. If you're mostly retail re-roofs and repairs, Jobber or Housecall Pro does the job for less money and less setup. The wrong answers are running both, or neither.
What does this stack cost per month?
Published prices change too often to print here, but the pattern matters more than the number: every tool in the stack should tie to a number you track, like calls recovered, reviews added, or quotes closed. Lesson 4's baseline exists so you can tell which subscriptions are earning and which are furniture.
Can't I just do all this with ChatGPT and a spreadsheet?
The drafting, yes, and lesson 3 shows you how. The things that must happen every single time without a human remembering, like the missed-call text or the review ask after every closed job, no. Reliability is the actual product you're buying, and a spreadsheet doesn't wake up at 7am on a Saturday because a hailstorm did.
That’s the whole curriculum. If you want the build-or-buy math run on your actual numbers, book 15 minutes and bring last month’s phone log.