The unit cost is the most practical measure, reducing translation expenditure to a single number. The same simplicity makes comparison between periods misleading.
When the mix changes, the number changes
More technical texts may have been translated in one period and more correspondence in another. The difference in the unit cost comes not from a difference in quality or price but from the mix of jobs.
Set-up expenses are separated
Memory alignment and the set-up of the termbase are done once. When they are mixed into the ordinary job expenses, the unit cost of that period looks higher than it really is.
The share of rush jobs has an effect
When the share of rush work rises, the unit cost rises too. The source of this increase is not the supplier but the flow of requests.
The repetition rate works in the downward direction
As the memory grows, the match rate rises and the unit cost falls. This fall is the return on the investment and should be shown separately.
The review layer spoils the comparison
When jobs produced with different layers are added up in the same average, a low unit cost may hide a scope that is incomplete.
The language direction is broken down
In directions where the pool of resources is narrow, the unit cost is naturally high. A combined average does not show which direction is expensive.
The trend is worth more than a single measurement
The number of one period is an observation. Placing several periods side by side shows whether the change is real or arises from the mix.
Counter-view: a single number is enough for management
It is true that senior management should not go into detail. But a single number, too, is meaningful only if its definition is fixed; when the mix and the layer are not broken down, the same number measures a different thing in two periods.
The total cost of ownership is calculated separately
The correction and coordination effort inside the organisation does not enter the unit cost. A comparison that looks only at the invoice remains incomplete.
Setting a target can distort behaviour
When a target is attached to the unit cost, narrowing the scope becomes the easiest route. The target is set together with quality indicators.
A change in volume has an effect
In periods of small volume, minimum volume practices can raise the unit cost. This effect may be taken for a price increase.
Exchange rate and seasonal effects are noted
If there are cost items of external origin, these effects are stated separately in the comparison between periods.
The definition of the measurement is written down
Which items are included, which unit of volume is used and how the set-up expenses are handled are defined.
It is read together with quality indicators
A falling unit cost is not good news if the density of findings is rising. The two trends are assessed together.
Takeaway
The unit cost is not a performance indicator on its own. Breaking down the mix of jobs, the review layers and the language directions, showing the set-up expenses separately and reading the trend together with quality indicators: without these four, a falling number may tell you nothing other than that the scope has narrowed.