Short answer / Lab
An experiment costs the materials it used, priced at what each lot cost, plus the bench time it took at each person's labour rate. One cell culture passage in the example comes to £61.42. A programme's cost per run differs, because it divides all money spent, including stock not yet used, by every run, drafts included.
The cheapest serum-free medium our example lab tested turned out to be the dearest one to grow cells in. It's £212 a litre against £238 for the one it chose, and £51.71 per billion cells against £38.39.
The number that matters is rarely the one on the invoice. That's the whole case for costing experiments properly, down to the run. Below is how I'd cost one run of routine cell culture, what the averages on a programme page actually mean, and the places both mislead. The example lab is a small biotech adapting HEK293 cells to grow in suspension without serum, passaging them twice a week all summer.
What goes into the cost of one experiment?
Two things: the materials it used, each priced at what its lot cost, and the bench time it took, at the rate of the person who did it. Everything else is overhead, and that's a separate conversation.
One passage of the adaptation carries three shake-flask lines forward:
| Item | Used | Price | Cost |
|---|---|---|---|
| Serum-free medium B | 90 mL | £0.238 a mL |
£21.42 |
| Shake flasks, 125 mL | 3 | £4.70 each |
£14.10 |
| Counting slides | 2 | £0.95 each |
£1.90 |
| Materials | £37.42 |
||
| Bench time | 30 min | £48 an hour |
£24.00 |
| The run | £61.42 |
Labour is 39% of it. Leave it out and every comparison between a manual method and an automated one is decided before you start.
The serum for the 5% to 0.5% steps is on each run with its lot, but it came from a bottle entered with no unit cost, so it adds nothing to the figure. That's the most common way a run looks cheaper than it was, and it's the first thing below under where it goes wrong.
How should bench time be costed?
At the person's rate, for the time the step was planned to take or actually took, and the choice between those two is worth making on purpose. CCA Labs sets it once for the organisation, under Labour Costing in Settings: Pro-rata charges every minute the step was open, Pro-rata, capped charges the time open up to the step's planned time, and Fixed charges the planned time however long it took.
The example lab uses capped, the default. A passage that runs long because someone was called away shouldn't cost more than one that didn't, and one that runs short should get the credit. It's the clamp on a controller's integrator: past a limit, more time stops adding to the total. A step left open overnight bills nothing extra, which is kinder to the budget than to whoever forgot to close it. Fixed suits a lab that prices its protocols and wants every run of one to cost the same.
Each person has a labour rate with a date it takes effect from, and the rate is saved onto the step when it runs. Give someone a pay rise in October and September's passages keep September's cost.
Is the material charged twice, once when bought and once when used?
No. The medium was charged to the adaptation project when the order arrived. Using 90 mL of it in a run puts that part of the cost on the experiment, and nothing new is charged to the budget. The money moved once, when the goods arrived. The run just says where it went.
Labour works differently, because time has no purchase to charge it to. The run's labour is charged to the project's fund when the step completes, but only if that fund pays for bench time. The adaptation runs on a grant that does, so each passage charges £24.00. The media screen before it ran on the lab's internal R&D fund, which doesn't: salaries come from payroll. Its runs have a labour cost on the experiment and none in the budget.
Why does the programme page say £138.28 a run?
Because it answers a different question: what has this programme spent per run so far? It takes the money the programme has actually spent and divides it by every run on its projects that isn't archived or deleted, drafts included.
On 31 August the programme has spent £6,499.00 and has 47 runs: 18 from the media screen, 27 passages (one of them running that morning) and 2 draft runs for a 2 L bioreactor. £6,499.00 / 47 = £138.28.
That's more than double a passage, and it's not wrong. The spend includes £3,390.00 of bioreactor vessels and probes that arrived on 27 August for runs that haven't happened, and medium bought in bulk that's still on the shelf. The count includes two drafts.
The same 47 runs were planned, from their own estimates, at £168.37 each on average. Hover over the cost per run and the page says so: "Cost per run, £30.1 under model". Labour more than explains it. Each media screen run was estimated with 170 minutes of bench time, £136.00, which the estimate counts and the internal fund never charges: £2,448.00 across 18 runs. Stock bought ahead of the runs that will use it pulls the other way. Under model can mean a thrifty lab, or a fund that doesn't pay for time. You have to look.
What is the cost per result?
Cost per result is the cost of the work divided by what it produced, and it's the figure that should decide between methods. The media screen, three media in 18 small cultures from March to May, shows why. It ran on a suspension-adapted HEK293 line the lab keeps for comparisons like this, because its own cells wouldn't grow without serum yet.
| Medium | Price a litre | Peak density | Per 10⁹ cells |
|---|---|---|---|
| A | £212 |
4.1 × 10⁶ cells/mL | £51.71 |
| B | £238 |
6.2 × 10⁶ cells/mL | £38.39 |
| C | £265 |
6.0 × 10⁶ cells/mL | £44.17 |
The arithmetic is price per litre over cells per litre. A litre of B at peak holds 6.2 × 10⁹ cells, so £238 / 6.2 = £38.39 per 10⁹. Medium B costs 12% more per litre than A and 26% less per cell. The screen cost £922.80 to find that out, before the grant paid for a single 2 L run.
What did the slip cost?
About three weeks and £371.10. At 0.5% serum the cells clumped and viability fell to 82%, and the lab held them there with an anti-clumping agent until they recovered: five passages more than planned at £61.42, and a £64.00 bottle of agent.
On a £20,000 grant, that money is 1.9%. The three weeks are the expensive part, because the grant ends on a fixed date and the bioreactor work that follows has to fit in whatever is left. No cost column shows that.
Where it goes wrong
Materials without a price. A run's materials cost is only as good as the unit costs on the lots. An item nobody priced costs nothing, and the forecast warns when that happens: "Some materials lack a unit cost in inventory; estimate is partial."
Overheads aren't on the run. A run in CCA Labs carries materials and labour (and a service cost, if it's outsourced), not instrument time. Utilities, calibration and equipment servicing post to the budget as scheduled overhead, against a fund or a project, not onto each run. If you're pricing work for a customer, add them yourself.
Averages across funds. Two projects on funds with different labour settings aren't comparable run for run. Compare runs on the same basis, or compare their costs on the experiment, which always include the time.
Estimates age. A labour estimate is planned minutes at the rate in force when the run is due to start, and CCA Labs re-prices runs that haven't started when a rate changes. Methods get faster too, and nothing re-prices that. The experiments list shows each run's cost against its estimate, and a steady gap in one direction means the estimate needs updating, not the lab.
How CCA Labs costs a run
Every run has a materials cost and a labour cost, and the same figure shows on the experiment, the project and the programme, to the people you give cost access. Each person's rate, with its history, sits on the Team page; the organisation's default rate and charging basis sit under Labour Costing in Settings; and a workflow step can set its own rate.
The programme page's Spend card shows the cost per run, with its gap to the modelled cost on hover, and the Projects tile shows how many runs are in flight. How the programme's money is split in the first place is in how to split one budget across programmes and projects, and how an order becomes spend is in how to track a lab budget across grants and projects.
You can see a run's cost build up step by step when you create your lab. The costing tools are on the all-in-one lab page.
£61.42 a passage, twice a week, is a burn rate. Add a £3,390.00 order for bioreactor vessels in the same month and the burn rate starts saying something quite different. That's the last part: how to work out a lab's burn rate and runway.
The whole story / one floor at a time
One programme's year, from the budget split to the August review
One example programme, followed through a small biotech: getting HEK293 cells to grow in suspension without serum, then in a 2 L bioreactor. Leadership splits two funds across three projects and sets the goals, one project slips three weeks, the bench counts what each run costs, and the August review shows what the slip did to a grant with a fixed end date. Each part is written for the people on that floor.
FAQ / asked about this
Frequently asked questions
- Should the cost of an experiment include labour?
- Yes, if you want to compare methods or price work, because bench time is often the larger share. Whether that time is charged to a grant is a separate question, decided by what the grant pays for. CCA Labs costs every run's time at the person's rate and charges it to a fund only where the fund covers labour.
- Why is a programme's cost per run different from the cost of one experiment?
- The programme figure divides everything the programme has spent by its number of runs. The spend includes stock bought and not yet used, and equipment bought for runs still to come; the count includes drafts. One experiment's cost is only what that run used and the time it took.
- What is cost per result, and when does it matter?
- It is cost divided by what the work produced: per vial banked, per billion cells, per sample tested. It matters when choosing between methods or media, because the cheapest input per litre is not always the cheapest per result. In the example, the cheapest medium cost the most per cell.