Overhead Recovery Rate for Trades: Stop Under-Pricing

Your overhead recovery rate for trades is the amount of running cost you add to every chargeable hour so that the costs which never appear on a job card — unit rent, vans, insurance, software, the owner's admin evenings, training, compliance paperwork — get paid for by the work. You calculate it in two steps: add up twelve months of overhead, then divide by the chargeable hours you actually sell (not the hours you pay for). For a typical five-person firm that lands somewhere near 20-30 per hour in your currency, roughly double what most owners guess. Get the number wrong and you lose the same amount on every single hour, quietly, all year.

Why the number is wrong in nearly every small firm

Two mistakes stack, and they compound. The first is a missing cost list: nobody forgets the rent, almost everybody forgets the owner's unbilled hours, the quotes that never converted, the callbacks you swallowed and the eleven subscriptions that arrived one at a time. The second is an inflated hour count — dividing by the hours you wish your team sold rather than the hours they sold.

Both errors push the same way: down. That is why the fix is nearly always a price rise, and why it feels uncomfortable before you check the arithmetic.

Step 1: find every cost that never lands on a job

Open last year's profit and loss and go line by line. For each line ask one question: can I name the job this belongs to? If yes, it is a direct cost and stays out. If no, it is overhead and goes in the pot.

Here is a worked list for a five-person electrical firm — four engineers in vans, one owner running the office. Annual figures, in whatever currency you work in. The categories matter far more than my numbers.

The largest line is the one with no invoice behind it: your own time.

The lines almost everyone misses

• Your own unbilled time. If you spend 25 hours a week quoting, chasing, ordering and invoicing, that is a full-time office manager you are not paying — but the business still consumes the cost. Put your market salary in the pot, or you are subsidising your own customers.
• Quoting that goes nowhere. At a 35% win rate, roughly two thirds of your estimating hours produce no invoice. That cost has to be recovered from the third that converts.
• Callbacks and warranty returns. A revisit is a paid hour with no invoice attached. Track them for one quarter; they are not rare.
• Holiday, sick and training pay. You pay for the hour; nobody bills it. Either strip it out of the chargeable hours or carry it as overhead — do one, never both, and never neither.
• Software. Job management, accounting, payroll, storage, email, price files, certification portal, phone system. Each looks trivial monthly. Sum the annual figures once and the total tends to surprise.

Step 2: divide by the hours you actually sell

This is where the honest number comes from. Work down from paid hours to chargeable hours per engineer:

Four engineers gives 5,540 chargeable hours a year. Not 8,320. If you have never measured this, measure before you argue with it — a time clock plus a job card that records start and stop on site will tell you the real split within a month. Guessing this number is what makes the whole calculation collapse.

Method A: overhead recovery per chargeable hour

134,700 of overhead divided by 5,540 chargeable hours = 24.30 per chargeable hour.

Now build the price from the floor up:

• Direct employment cost per engineer, including employer contributions, pension and holiday accrual: 48,000
• Divided by 1,385 chargeable hours = 34.70 of labour per chargeable hour
• Plus overhead recovery of 24.30
• Break-even = 59.00 per hour. At this price the business makes exactly nothing.
• For a 20% net margin, divide by 0.8: 73.75, call it 74 per hour.

Note the divide, not the add. Adding 20% to 59 gives 70.80 and a margin of 16.7%, not 20%. Margin and mark-up are different animals, and mixing them up costs you a fifth of your profit.

Now the scale of the error. If you have been carrying 12 an hour of overhead in your head instead of 24.30, you are short by 12.30 across 5,540 hours: 68,100 a year. That is not a pricing tweak. It is most of a wage.

Method B: a percentage on contribution

The second method loads overhead as a percentage on top of direct cost, so it rides on materials as well as labour. In Swedish costing this is the omkostnadspålägg applied against täckningsbidrag (contribution); the English equivalent is an overhead absorption percentage on prime cost.

Same firm. Direct costs are labour 192,000 plus materials at cost 160,000 = 352,000. Overhead of 134,700 divided by 352,000 = 38%. So every job gets direct cost plus 38%, then margin on top.

It is quick and it survives a mixed job book. But it charges overhead on price rather than on effort. A material-heavy supply-and-fit job gets loaded with overhead it never caused; a labour-only day gets away light. Buy a 30,000 switchboard and the maths adds 11,400 of overhead to a purchase that took one phone call.

Which one to pick

The dual rate, which is what most firms should actually run

Do not choose. Split it:

• Put the time-driven overhead — vans, rent, insurance, software, unbilled admin, training — on the chargeable hour.
• Put the purchase-driven overhead — ordering, delivery handling, storage, warranty exposure, financing the stock, returns — on materials as a percentage, typically 8-12%.

Re-run the example: pull 15,000 of procurement-related overhead out of the 134,700 and charge it as 9% on 160,000 of materials. The remaining 119,700 over 5,540 hours gives 21.60 an hour. Every job now pays for what it actually consumes, and a big kit purchase stops looking like a windfall it never was.

Why this is worse in 2026 than when you set your rate

Overhead per employee has been climbing faster than headline inflation, in three quiet directions:

• Software. Ten years ago a small firm ran on a phone and a spreadsheet. Now there is a subscription for scheduling, one for accounts, one for payroll, one for storage, one for certificates, one for price files. Fragmentation, not extravagance, is what costs the money — and it is the cheapest overhead to cut, because consolidating tools removes licences and the admin hours between them at once.
• Insurance. Liability and tools premiums have moved several years running. Very few rate cards moved with them.
• Compliance admin. Certification, self-inspection records, documented handovers, safety paperwork. Every hour is real and none of it is billable. In Sweden that means Elsäkerhetsverket notifications and egenkontroll; in the UK it is scheme notification and certification. Either way the hours are overhead.

Meanwhile most firms still use an uplift set in a year when none of that existed. The recovery rate is a snapshot with an expiry date, and it expired.

How to keep the number honest

• Recalculate every 12 months, from the actual accounts, not from last year's rate plus a bit.
• Recalculate on any headcount change. Hiring engineer number five spreads overhead across 1,385 more hours and can genuinely lower your rate. Losing one raises it.
• Use last year's real chargeable hours, not this year's target. Optimism in the denominator is the same as a price cut.
• Watch the overhead spiral. When work drops, hours fall, per-hour overhead rises, and the arithmetic tells you to raise prices into a slow market. That signal is real but you cannot always follow it. Cut overhead or accept a thinner margin deliberately — do not pretend the hours are still there.
• Never treat a customer subsidy as headroom. Sweden's ROT deduction takes 30% of labour off what the household pays, within an annual per-person ceiling shared with RUT (check the current amount at Skatteverket). It changes the customer's net price. It does not pay a penny of your rent, and it is not a reason to discount.

The measuring is the hard part, not the maths. You need real hours on site rather than hours guessed at the end of the week, and job costs landed against the job while the van is still on the drive. That is what a time clock and a job card give you: attended hours on one side, chargeable hours on the other, and the gap between them is the denominator this whole calculation stands on. OdinTask records both.

Do the exercise once. It takes an afternoon with your accounts and a calculator. Then look at your current rate and decide, on purpose, whether you are running a business or a very busy hobby. If you would rather have the hours counted than estimated, start a free trial and let the field app measure them for a month before you set next year's rate.