Although underestimated, Kosu is the most relevant indicator. It allows to take into account all the possible losses of the company and to have a real vision of the level of productivity.
Introduction
Kosu, meaning “ Manual Time”, is a key indicator in production control. It defines the level of productivity by identifying the true production time needed to make a part. In its concept, the generic formula is as follows :
Exemple : We are an SME of 150 people (human resources, management, production …) and 10 temporary workers. Over a week, we accumulate a total of 5,900 hours of work (some managers working more than 35hrs) and we produced 29,500 pcs good. Our Kosu is 5,900 * 60 / 29,500 = 12 minutes. Each piece represents 12 minutes of working time for the entire company.
One can easily identify the level of productivity of the company and compare itself with other companies in the sector. This will tell us :
 the ratio between the number of support persons and the number of people actually productive.
 La productivité d’une ligne et ces améliorations.
 Will make the direct link between the cost of labor per product.
1 – Calculate the standard cycle time
The standard time represents the optimal time to produce a good piece. This is a theoretical time that we will compare with the Kosu. The standard time is calculated according to the type of line we have, that is to say with an operator isolated or not.
An isolated post is considered to be a post whose duties can not be combined with others. This for reasons of distances, technologies, skills…
To ensure the correctness of the indicator, the standard time must be updated every 6 months, and is measured on the average of thirty cycles.
Case 1: with isolated operator but not the bottleneck
In this case, we have a production line where an operator has a apart from position, it is not in the feed with the other operators. The cycle time of his task is not the bottleneck of the line. The standard cycle time will then be the addition of all times of the line.
In our example this gives : 20 + 16 + 22 + 14 = 72 secs
Case 2: with isolated operator that is the bottleneck
We are in the same configuration as before, but this time the isolated post is the bottleneck of the line. In this case, we will then take a standard time equal to the cycle time of the isolated operator multiplied by the number of operators in the cell.
In our example, this gives : 4 * 22 = 88 secs
Case 3: no isolated operator
All the operators are put together, and the operations can be combined. We will then add the cycle times of each station.
2 – Calculation of Kosu
Kosu is the ratio between the production time and the number of good pieces produced over a given period. The generic formula is the following :
Kosu = Time worked * Number of operators / Quantity of good parts normalized
With :
 Time worked : Actual production time by counting breaks, meetings, serial changes and preventive maintenance intervention.
 Number of operators : Number of people dedicated to the production of parts.
 Number of standard parts : Number of good parts produced during the time worked by all operators. As we will see below, the number of parts to be taken into account is “normalized ».
Case 1: The range of cycle times is less than 10% of the average value
For example, on a line composed of 4 operators, we have 3 types of product references but with cycle times for each of the references close :
 Product 1 : 30 sec
 Product 2 : 31 sec
 Product 3 : 29 sec
In this example, we have an extent of 2, less than the maximum extent, which in this case is equal to 3.
The calculation is then very simple: it is necessary to calculate the number of hours, minutes or seconds, which one divides by the number of good parts produced on this same week.
For example, if over a week we produced 9000 pieces in 2 teams of 4 people over 5 days, we get the following Kosu :
Kosu = 105 000 * 4 / 9000 = 46,66 secs
We need 46.66 effective working hours to produce 1 piece good.
Case 2: The extended cycle times are greater than 10% of the average
If on the same previous line, we had the following cycle times :
 Product 1 : 30 sec
 Product 2 : 60 sec
 Product 3 : 40 sec
In this case, the range is 30 seconds, which is considerably greater than 10% of the average, which is in this case 4.3 seconds..
We will use a coefficient of normalization allowing to take into account this difference. The procedure is as follows :
 The reference time is the cycle time of the weakest product.
 The time of another product is divided by the reference cycle time to obtain its coefficient.
 We multiply the coefficient by the number of products of this manufactured reference.
 We calculate the overall Kosu by taking the sum of the product number normalized by reference as the quantity of good products.
Using the above example, for a 1 hour production calculation to 4 people we get the following table :
Product 1 
Product 2 
Product 3 

Cycle time 
30 sec 
60 sec 
40 sec 
Quantity of good produced parts 
102 
76 
84 
Coefficient of time 
1 
60/30 = 2 
40/30 = 1,33 
Kosu of the process 
3000 * 4 / (102 + 152 + 112) = 32,78 sec 
We need on average 32.78 secs of effective labor to produce 1 piece good.
3 – Manage Kosu
The Kosu will be compared to the Kosu goal. It is calculated according to the standard time calculated above and the objective of the company. Generally, the Kosu goal is set between 10 and 25% of the standard time. For example, we will take a Kosu objective of 12 secs for a standard time of 10 secs.
Then, we will define alert thresholds by respecting the following rules :
 – de 15% : the team leader is warned with a mission to understand the reason for the variability.
 15 to 25% : The workshop manager is notified and will have to decide what to do.
 + 25% : the problem is traced back to the direction of the site which has the mission to go to the Gemba to follow the investigations.
Source
M. Imaï (1997) – Gemba Kaizen
M. Jaccard (2010) – Objectif qualité
T. Nishiguchi (1994) – Strategic industrial sourcing : the japanese advantage