Systems before headcount

You turned work away last month, so the answer looks like a second installer. Before you post the job, count the jobs that never got far enough to turn down. The estimates you wrote at nine at night. The request buried in a text thread that never got a price at all.

A person is the most expensive and slowest thing you can add. If jobs are dying before anyone would have swung a hammer, hiring does not touch the loss. The process fix usually costs a weekend, and you still want it later, once you do hire.

Count last week before you count the crew

Split one real week into hours a customer pays for and hours nobody pays for. Not a typical week you remember — last week, off your calendar and your texts. The example below is a one-truck contractor, with invented numbers to show the method.

That is 32 billable hours out of 58, about 55%. The other 26 hours are the business running itself, and a second installer does not shrink them. He adds field hours and adds paperwork: another schedule to build, another set of hours to track, another person's questions mid-job.

There is no published benchmark for what a one-truck contractor's billable share should be, so do not go hunting for one. Use the direction instead. If nearly all of your week is already on site and you are still turning work away, hands are the constraint. If a big slice is office work, hiring does not reach it.

Find where the jobs stall

Count how many jobs entered each stage last month and how many came out. Same example, carried forward:

Eight inquiries never got an estimate — the biggest single loss in the month, and none of it cost field hours, because those customers were never on the schedule. Eight more estimates went out and died in silence at a median of eleven days, which is what "I meant to follow up" looks like in a table.

The bottom two rows are the same problem in the money. The eight-day gap before invoicing is self-inflicted, and a third still unpaid at 39 days is at least partly the same thing, because nothing was chasing them.

Run this on your own month. The stage with the biggest drop is your bottleneck.

What a hire actually commits you to

Wages and payroll setup are their own subject. What matters here is the shape of the commitment.

• The cost is not the wage. On top of gross pay you carry the employer share of Social Security and Medicare — 6.2% up to the annual Social Security wage base plus 1.45% on all wages, so 7.65% at an installer's pay level — plus federal and state unemployment tax and, in nearly every state, workers' compensation premiums priced by class code. Texas is the well-known exception, where most private employers may go without coverage and give up lawsuit protection in exchange. Your state labor department, your state's workers' comp regulator and your own insurance agent can price your case. Nobody online can.
• The ramp is months, not days. A capable installer still needs your material list, your standards and your callbacks explained, and those conversations come out of your billable hours.
• It does not switch off. Payroll runs in a slow February whether or not the phone rang, while a follow-up cadence costs one Sunday afternoon and then runs itself.

Four systems you can build in a weekend

None of these need software to start. They need a decision made once instead of every time.

A priced catalog. Your twenty most common line items with a unit and a price: linear foot of six-foot cedar privacy fence, post set in concrete, gate, mobilization, haul-away. Once these exist, an estimate is selection and quantity, not composition. It also gets the pricing out of your head, which matters most later.

One estimate format. Same sections, same order, same exclusions paragraph, same expiration window. Editing yesterday's document is not a template — it is how the last customer's exclusion ends up on this customer's job. If your state requires specific language on a residential contract, build it in once instead of recalling it each time; your licensing board publishes what has to appear.

A follow-up cadence with dates on it. Day 2 (confirm it arrived, offer to walk it), day 7 (one specific question), day 14 (last check before the price expires). Attached to the estimate, not left to your memory. Eleven days of silence is not a customer saying no. It is nobody saying anything.

One list of what is owed and what is scheduled. A spreadsheet is fine. Two columns matter: jobs approved but not yet on the calendar, and invoices sent but not paid, with the date each went out. A payment row past a month usually means no such list exists, and the owner finds out at the bank.

The catalog takes the longest, because you are writing down prices you have only carried in your head. Start with whichever system sits on your worst drop-off.

When the bottleneck really is hands

Sometimes it is. The signals are in the field, not the office.

• Your booked backlog — work already priced and won — keeps growing week over week instead of holding steady.
• You are declining jobs after the customer accepted, not before you responded.
• The work itself needs two people, so you are renting equipment or paying a sub to hold the other end on most jobs.
• Estimates go out fast and still convert, and the job still cannot be scheduled.

That is a capacity problem, and no template fixes it. Adding a person, or a sub for the overflow, is the right call.

Hire into a business that can run without you

An undocumented business does not delegate. If your prices live in your head, every estimate still routes back to you after the hire — now with payroll on top of the same evening paperwork.

Estimating was six of the twenty-six unbillable hours in the example week, and it is usually the shortest path to time back. In DoneQuote you describe the job in a few sentences, or forward the customer's email, and a draft comes back carrying your own catalog prices. You check the quantities and send. Follow-up runs on a schedule instead of on whether you remembered. That is the catalog and the cadence running themselves — the same two things you hand to someone else later.

Run the count for 30 days

Log two things for a month: hours by category, and how many jobs entered and left each stage. Fix the biggest drop, then measure again. If billable share climbs and the drop-off shrinks, you bought capacity without payroll. If neither moves, you have a real hiring case — and the numbers to make it.