AI, where it does not belong.

An industry that never grew up

Looking after homes is one of the largest industries in the country and one of the least organized. Hundreds of billions of dollars a year are spread across millions of firms, and no company holds more than a sliver. Most are one owner, a few crews and a phone.

Customers do not love this arrangement. Calls go unanswered. Quotes take a week. The crew that did good work in March is a different crew in October. People put up with it because the alternative down the road is no better.

The usual explanation is that the work is local and manual, so it cannot be organized at scale. We think that explanation is half right.

The office is the limit

The manual half scales well enough. A second crew does about as much work as the first. What does not scale is the office.

In a small service company the office is the owner. The owner answers the phone, looks at the house, works out a price, books the crew, reorders materials, chases the invoice and remembers that the customer has a dog. All of it lives in one head. Each added crew makes that head's job harder, and at some point the owner stops growing because growing has stopped being worth it.

So the industry is fragmented for an old reason. A firm grows until organizing one more job inside it costs more than the job is worth. In these trades that limit arrives early.

What changed

Nearly everything in that office is reading, writing, looking and deciding. Measure a roofline from a photograph. Price it. Reply at nine at night, within a minute, in plain English. Put the right crew on the right street on the right morning. Ask for the review.

These were jobs for a capable person. They are now jobs for software, and the software improves every few months and costs less each time.

When organizing a job gets cheap, the limit on the size of the firm moves. A company can run ten crews, or a hundred, across several trades and several cities, on an office that does not grow with it. That has not been true before.

Where the profit goes

It would be easy to conclude that the money is in selling this software to contractors. We do not think so. A tool sold to everyone is an advantage to no one. A contractor can already rent an answering agent for less than a tank of fuel, and competition will hand most of the saving to the customer. In the long run, under competition, economic profit goes to zero.

What competition does not erase so easily is position. In every market there is one company in each trade that people call first. It has the reviews, the repeat customers, the trucks already on the street and the crews who stay. That position is slow to build and hard to take.

Intelligence is becoming abundant. Trust is not. We use the abundant thing to accumulate the scarce one.

Own the work

A firm that could reliably double a contractor's business should not charge that contractor a retainer. It should own the business and keep the gain. That is what we do. We do not sell software, advice or leads. We build and buy the operating companies and run them on a system of our own.

Owning has a second advantage. A vendor sees one slice of a customer's business. An owner sees every lead, every quote, every job and every complaint, and the system learns from all of it.

One household, many trades

The lights on the roofline, the gutters beneath them, the lawn and the pool belong to one household. That household would prefer not to find and vet a different company for each. Trust earned on one job carries to the next when the work is close enough that the customer believes you can do it, and the cost of winning that customer is paid once.

The same holds from market to market. Each town is a small prize, which is why large investors pass it by and why it can be won outright. Each one is also an experiment. What works in one is in all of them the next morning.

The order of things

The office comes first, and that is happening now. Selling and marketing come with it. Physical work comes last, and later than the forecasts say. A roof in the wind is a harder problem than a spreadsheet.

Over a long enough horizon we expect machines to do much of that work too. When they can, they will go to work for whoever already has the customers, the routes and the records of ten thousand jobs.

Until then the people on the ladder are the product. Our job is to give them better tools, fuller schedules and fewer wasted trips than any competitor can.

How we could be wrong

Three ways, at least.

The tools may spread so fast that every contractor gets the same office we have. Then the edge is execution, and we will have to be better operators than people who have done this for thirty years.

Labor may be the real limit. Software does not climb. If good crews cannot be hired and kept, nothing above matters.

And combining small companies has a poor record. Most attempts fail at the dull part, which is making many small firms behave like one good one.

Our answer to all three is the same. Go slower than the idea tempts us to, and finish one market properly before starting the next.