How Do I Figure Out Which of My Services Actually Make Money?

Job-cost every service line: revenue for that line, minus the fully loaded labor it takes to deliver, minus its direct costs. Run one quarter of numbers. In the service firms I clean up, one or two lines usually carry most of the profit, at least one quietly loses money, and you have probably guessed wrong about which is which.

Your gut ranks services by revenue and affection. Margin ignores both. Even a firm with a healthy blended margin can hide a losing line inside the average, and the average is all your P&L shows you.

Why doesn't my P&L already show this?

Because it is organized by expense type, not by service. Payroll sits on one line. Rent on another. Software on a third. Nothing in that layout connects your senior tech's Tuesday to the retainer it was spent on, so the statement can tell you the firm made money in March but not which service made it. Blended numbers are the trap. A 15% net margin can be a 45% line and a money-losing line averaged together. Averages feel calm right up until you price a new engagement off one. The monthly reports worth reading tell you whether the whole firm is healthy. They were never built to tell you why.

The four-number test for every service line

No new software required. For every service you sell, pull four numbers out of one full quarter of activity, and be stubborn about the loaded part of number two:

1. Revenue by line. Tag every invoice to a service (QuickBooks classes or service items both work), then total the quarter.
2. Delivery labor, fully loaded. Hours spent delivering that line, and count the whole job: setup, handoffs, invoicing, and the account management each engagement drags along, not just hands-on-keyboard time. Price them at wage plus payroll taxes and benefits. In the payrolls I review, loaded cost typically runs 20% to 30% above the bare wage, higher if you pay for real health coverage.
3. Direct costs. Software seats, subcontractors, materials, anything that exists only because this line exists.
4. Gross margin. Revenue minus labor minus direct costs, divided by revenue.

Nobody logged hours last quarter? Do not reconstruct them from memory. Start logging now. Two weeks of honest logs builds a usable ratio for steady work like retainers. Project work is lumpy, kickoff-heavy up front and backloaded at QA, so cover at least one full project before you trust that number.

As a rough map from my own engagements, not a rate card: a healthy service line clears 50% to 60% gross margin before overhead. Under roughly 35%, a line usually cannot carry its share of rent, admin, and your own salary. The leaner your overhead, the more forgiveness there is.

Busy is not the same as profitable.

Play it out on an illustrative $2M firm, the shape I keep seeing:

One firm, three different businesses. The retainer line funds the company. The project line sits one bad scope from break-even. The ad-hoc line looks like easy money per invoice and eats the calendar in real life: thirty small jobs mean thirty setups, thirty handoffs, thirty invoices of admin nobody estimated. Nobody plans it that way. It accumulates while you are busy saying yes.

Gross margin is not the verdict, though. Now make each line carry overhead. Add up everything that is not delivery labor or direct costs: rent, admin staff, firm-wide software, your delivery team's bench time and PTO, and the salary you pay yourself to run the place. Divide that by total revenue, and every line owes that share. Say this firm carries $400K of overhead, 20 cents of every revenue dollar. The retainer line still nets 30%. Fixed-fee projects land at exactly break-even. And the ad-hoc line loses ten cents on every dollar it brings in. There is the line that quietly loses money, and your P&L never showed it.

The favorite is usually the suspect.

How do I count my own hours?

At market wage, all of them. A service that only makes money because you deliver it free is not profitable. It employs you. Price your own delivery hours at what you would pay a competent hire, the same logic as paying yourself a real salary, and watch which lines survive the correction.

This is where you flinch. Unbilled founder hours are the biggest lie in service-business math, and they hide inside the exact line you deliver personally because nobody else can do it yet. Cost those hours honestly and a 40% line can drop to 12% overnight. I have watched that single correction reorder an owner's entire ranking.

What do you do with a losing line?

Three options, in order: reprice, repackage, cut. Repricing goes first because it is the cheapest experiment you can run. Raise the rate 10% to 20% or set a project minimum, then let the clients who leave tell you what the line was actually worth. In my experience, the clients who walk over a 15% increase were the margin problem. Repackaging changes the shape of the work (tighter scope, fewer revisions, a productized deliverable) so delivery hours fall while the price holds. Cutting comes last, after the first two fail, and a quarter per experiment is enough to know.

A losing line is not automatically a dead line. Some genuinely feed the profitable work, the small job that turns into the retainer six months later. Keep one of those if the whole relationship pencils out, but decide it on purpose with the numbers in front of you. A loss leader you chose is a strategy. A loss leader you discovered is a leak.

The honest bottom line

You are not really asking which services make money. You already have a suspect. What you want is permission to stop selling it, and a quarter of tagged invoices is how you get that permission in writing.

Bottom line: tag revenue and hours by service for one quarter, load labor at wage plus 20% to 30%, and make every line carry its share of overhead. Lines clearing roughly 50% gross margin are funding you. Under about 35%, reprice, repackage, or cut.

Start with your worst suspect, not your whole catalog. Want the per-service view wired into your monthly numbers? Two-week setup. KLYVNT builds it as part of a normal cleanup, the same tagging and hour logging you would do anyway, just done for you. Either way, stop pricing on affection.