Margin by Service Line: Finding the Work That Loses Money

A profitable business can contain a line of work that loses money on every job, and usually does. The whole-company profit figure hides it, because the profitable lines pay for it.

In my experience the loss-making line is rarely the one the owner suspects. It is often the newest offering, the one launched to win a particular client, or the one the owner personally enjoys most and therefore prices worst.

Why the income statement will not tell you

A standard income statement stacks revenue, cost of sales and expenses for the business as a whole. Read it for years and you learn the shape of the business without ever learning which part of it is carrying the rest. That limitation is structural, not a bookkeeping failure, and it is worth understanding before you go looking: see financial statements explained.

What you need instead is contribution margin by line: revenue for that line, minus the costs that would disappear if you stopped doing it.

That last clause is the whole test. Not costs allocated by some percentage. Costs that genuinely go away.

Splitting the cost

Most small businesses can do this in a spreadsheet, quarterly, from data they already have. You do not need a new accounting system. You need one decision per cost account.

Overhead stays out on purpose. The moment you allocate rent across five lines by revenue share you have built a model where a line looks unprofitable because another line grew. That produces confident, wrong decisions.

The two costs that decide the answer

The owner’s hours. In a business where the owner is also the most skilled technician, unpaid owner labour is a real cost being absorbed by whichever line the owner works on. It is the reason the owner’s favourite line so often looks like the best one. Cost it at what you would pay someone to do it, not at zero and not at what you bill.

Rework. Callbacks, warranty visits, the job that came back twice, the client who needed four rounds of revisions. Rework never has its own account, so it disappears into general labour and the line that generates it looks the same as the line that does not. Tag it for a single quarter. The result is usually unambiguous.

Where the numbers come from

If your accounting software supports classes, tracking categories, or job costing, turn it on and use it consistently from the start of a quarter. Retroactive tagging of a year of transactions is a project people abandon.

If it does not, do it manually for one quarter:

For each line of business:
  Revenue           from invoices, tagged by line
  Direct materials  from purchase invoices, tagged by job
  Direct labour     hours x loaded rate, from the schedule
  Owner hours       estimate honestly, x a real rate
  Rework            hours and materials, tracked separately
  = Contribution    what this line pays toward overhead

Loaded rate means wage plus employer CPP, EI, WSIB and any benefits, not the hourly wage. Using the bare wage understates delivery cost by a material margin in every province.

One timing error will wreck the exercise if you let it. If a job spans a period end, the revenue and the cost have to sit in the same quarter or the line looks brilliant in one and terrible in the next. Match them, either by billing on completion or by carrying work in progress. The same discipline applies to stock, which is why closing inventory has such leverage over reported profit: see inventory and cost of goods sold.

A line that contributes nothing is one the others subsidise

A line contributing nothing is a line the rest of the business is subsidising. That is a decision, and it can be a defensible one: a loss-leading service that reliably brings in profitable follow-on work is a marketing cost with a strange name. What is not defensible is subsidising a line without knowing you are.

The pricing environment makes this urgent rather than academic. In the second quarter of 2026, 64.3% of Canadian businesses expected cost-related obstacles over the next three months, and 28.4% named the cost of inputs specifically. Meanwhile Statistics Canada found that among businesses reporting lower revenue in 2025 than 2024, the average decline was 25.8%. Input costs rising into revenue like that is exactly the condition under which a thin line turns negative without anyone noticing.

What to do with a line that does not contribute

Line contributes nothing or negative
├── Does it reliably feed a profitable line?
│    ├── Yes → keep it, and cost it as marketing
│    └── No  → continue
├── Can the price move?
│    ├── Yes → raise it on new work only, measure
│    └── No, market will not bear it → continue
├── Can the delivery cost move?
│    ├── Yes → fix the process, remeasure next quarter
│    └── No → stop selling it
└── Stop selling it to new clients first,
    then wind down existing commitments

Note the order. Price before cost, cost before exit. Owners reach for cost cutting first because it feels within their control, and it is usually the harder of the two. Statistics Canada found that in the second quarter of 2026, 25.2% of Canadian businesses expected to raise their selling prices over the next three months, with accommodation and food services at 42.0%. Raising a price is not the unthinkable act it is often treated as, though it needs to be done deliberately: pricing your services covers the mechanics.

The scale problem, honestly

This analysis is harder in a very small business, and most Canadian businesses are very small. As of December 2024 there were 1,079,188 employer businesses with 1 to 99 employees in Canada, and 59.1% of all employer businesses had between one and four employees. At that size the owner is the labour, the timesheet does not exist, and every allocation is an estimate.

Do it anyway, and do it roughly. A rough split that identifies the line losing money beats a precise total that hides it. The point is not a number accurate to the dollar. The point is a ranking you trust enough to act on.

Then fold the result into the break-even model, because break-even for a business with three lines at different margins is three different numbers depending on the mix you sell.

If you want your last twelve months split by line and costed properly, including your own hours, send me the year end and a list of what you sell. The answer is usually one line, and it is usually not the one you would guess.