Sam Kaufman

AI and the Remodeling Industry

AI Is Not Your Business Model

I’m extremely bullish on AI. I use it every day. I build with it every day. I think it’s the biggest leverage shift small business owners will see in our lifetime.

And I’m watching it make a lot of contractors worse at business.

Not because the technology is bad. Because they’ve confused the tool with the model. They think AI is the strategy. They think if they build enough agents, automations and dashboards, a profitable company will show up on the other side.

It won’t. AI is not your business model. It never will be.

What a business actually runs on

Every remodeling, construction and home improvement company I’ve ever seen that consistently makes money runs on the same five things:

  1. A model. How the company makes money, on paper, with real numbers.
  2. Disciplines. The departments and repeatable processes that execute that model.
  3. KPIs. The handful of numbers that tell you whether each discipline is working.
  4. Targets. What those numbers have to be for the model to produce profit.
  5. Variable compensation. Pay that rewards people for hitting the targets that matter.

That’s the engine. Model, disciplines, KPIs, targets, comp.

AI doesn’t show up anywhere on that list. It’s not supposed to. AI is what you use to make those five things run faster, cleaner and with less friction. It’s a power tool. And a power tool in the hands of someone who doesn’t know what they’re building just makes mistakes faster.

1. The model comes first

A business model is a math problem. Revenue target. Average job size. Number of jobs. Close rate. Number of appointments it takes to get there. Number of leads it takes to get those appointments. Cost per lead. Gross margin. Overhead. Net profit.

If you can’t write that out on one page, you don’t have a model. You have a hope.

And here’s the uncomfortable part. Most companies chasing AI right now haven’t written it out. They don’t know their real cost per lead. They don’t know gross margin by job type. They don’t know what overhead is as a percentage of revenue. But they’ve got three AI subscriptions and a custom GPT.

If your business isn’t producing a net profit right now, a 0.01% technology efficiency gain is not valuable. It’s a distraction. Fix the math first.

I go deeper on building the model in Your Business Model Is a Math Problem.

2. Disciplines execute the model

A model on paper doesn’t do anything. People and processes do.

Every remodeling and home improvement company runs on five disciplines, and each one has something sacred:

  • Marketing: the lead is sacred.
  • Sales: the appointment is sacred.
  • Production: the install timeline and scope are sacred.
  • Finance: your profit is sacred.
  • In line with it all: your customer is sacred.

That’s the supply chain of a contracting company. Marketing creates leads. Leads fuel sales. Sales create production. Production generates income. Finance tells you whether you kept any of it. And the customer runs through every step. Money flows in one direction, and every discipline either speeds that flow up or slows it down.

A discipline isn’t a department name on an org chart. It’s a set of written processes, owned by a person, done the same way every time. If your sales process lives in your top rep’s head, you don’t have a sales discipline. You have a guy.

AI can’t build your disciplines for you. It can help you document them. It can help you train them. It can take repetitive work off them. But someone has to decide how the work should be done first.

More in The 5 Disciplines Every Home Improvement Company Needs Before AI.

3. KPIs tell you if it’s working

Every discipline needs a scoreboard. Not 40 metrics. A handful that actually move the needle.

Marketing: cost per lead, lead volume, speed to lead. Sales: set rate, close rate, average sale, gross margin sold. Production: gross margin delivered vs. gross margin sold, cycle time, callbacks. Finance: net profit, overhead percentage, cash position.

If you don’t measure it, you can’t manage it. And if you can’t manage it, you definitely can’t improve it with AI. You can’t prove an AI project worked if you never knew the number before it started.

4. Targets turn KPIs into accountability

A KPI without a target is trivia.

“Here’s our close rate.” Okay. Is that good? Is that what the model needs? If the model needs a higher close rate to hit net profit, you have a problem. If it needs a lower one, you’re ahead. Without the target, the number is just a number.

Targets come straight from the model. Work backward from the net profit you want and you’ll know exactly what every number has to be. Now every KPI has a line. Above the line, you’re building. Below it, you’re bleeding.

This is where most owners skip ahead. They track numbers but never decide what the numbers should be. So nobody knows if they’re winning.

I break down how to set them in KPIs Without Targets Are Trivia.

5. Variable comp makes people care about the right numbers

You get the behavior you pay for. Every time.

If you pay salespeople on revenue, they’ll sell revenue, including the bad jobs, the discounted jobs and the jobs production loses money on. If you pay production on completed jobs, they’ll complete jobs, whether or not margin holds.

Variable compensation ties pay to the targets that protect the model. Sales gets paid on gross profit, not top-line price. Production gets rewarded for delivering the margin that was sold. Office staff gets rewarded for the numbers they actually control.

When comp is built right, you don’t have to chase people about KPIs. They chase them for you.

More in You Get the Behavior You Pay For.

Where AI actually fits

Now we can talk about AI. Because now there’s something for it to improve.

Once the model, disciplines, KPIs, targets and comp are in place, AI does four things extremely well:

It improves the client experience. Faster response. Clearer communication. Better expectations set before the job starts. Proactive updates during the job. Fewer moments where the homeowner is wondering what’s going on.

It increases productivity. Your people get more of the right work done in the same hours. Proposals in minutes instead of an hour. Training built in an afternoon instead of a month.

It increases efficiency. Overhead that used to grow with every job stops growing. Data gets entered once. Reports build themselves.

It reduces friction. Every handoff in your supply chain is a place where money leaks: marketing to sales, sales to production, production to finance. AI smooths those handoffs so nothing gets dropped.

Two jobs for every AI project

Here’s the filter I run every AI idea through. It has to do one of two jobs.

Grow what the business can do. Respond faster. Hand off cleaner. Produce more with the same overhead. Give the customer a better experience at every step. This is capability that lives in your systems, and it stays when people come and go.

Or grow what your team can do. Train a new rep in weeks instead of months. Prepare every salesperson before every appointment. Turn your best people’s knowledge into SOPs everyone can run. Give your team leverage so their own thinking gets sharper, not replaced.

Most owners only think about the first job. The second one is where the long-term advantage is. Tools can be copied. A team that keeps getting better can’t.

Each path has its own disciplines and its own scoreboard. I lay out both in Grow the Business. Grow the Team.

Either way, it has to be measurable. Every one of those should show up in a KPI. Speed to lead goes down. Close rate goes up. Admin hours per job go down. Gross margin delivered gets closer to gross margin sold. Net profit goes up.

If the AI project doesn’t move a KPI that’s tied to a target that’s tied to the model, it’s a hobby.

The order matters

Here’s the sequence I use with every company:

  1. Write the model.
  2. Build the disciplines that execute it.
  3. Pick the KPIs for each discipline.
  4. Set targets from the model.
  5. Align comp to the targets.
  6. Then find the constraint and point AI at it.

Most owners are trying to start at step 6. That’s why they’re busy, excited and not any more profitable than they were a year ago.

AI will amplify whatever you already are. A disciplined company with a clear model gets dramatically better. A chaotic company gets chaos at scale.

The question to ask before every AI project

Before you build anything, ask: which KPI will this move, by how much, and how will I know?

If you can answer that, build it. If you can’t, you’re not ready for that tool yet. Go back up the list until you can.

The model is the business. The disciplines run it. The KPIs measure it. The targets define winning. The comp makes people care.

AI makes it all faster. That’s it. That’s enough.

Systems beat trends. The net profit doesn’t lie.