Read the Progressive Dairy Canada digital edition

Milking robots: Where efficiency meets profitability

Gervais Bisson for Progressive Dairy Published on 17 August 2021

When a dairy farm’s objective is to maximize milk production in the bulk tank, the calculation is simple: multiply average production by the number of cows in milk. With robotic milking systems, however, the robots operate continuously.

This means that, to maximize production, cows must have additional aptitudes. Not only do they have to be good producers, but they also need to be more efficient to maximize the 1,180 minutes of milking available every day; this is the approximate time available for a robot that has 10% free time after considering wash time and other nonavailable time.



To sum it up, milking-robot efficiency is the number of kilograms of milk per minute in the robot. Efficiency takes into account the time a cow spends in the robot during two phases: the preparation and milking phases. For a cow to be efficient, she must, therefore, have a good temperament, good mammary gland conformation and an above-average milking speed. This is the ideal cow for robot milking.

Does productivity equal efficiency?

We often think, the higher the cow’s productivity, the more efficient the animal must be. This is the question we looked at in 264 Quebec Holstein herds that subscribed to milk recording in March 2021. It can be true, but it is not necessarily always the case. The correlation is relatively low at the herd level and practically nonexistent when looking at individual cows.

As you can see in Figure 1, there is significant variation among herds.

Variation in efficiency in kilogram of milk per minute in the robot

We used the average of the last 12 months for both efficiency and production. In order to illustrate the impact of the variation in efficiency on productivity more clearly, a red and a green dot were added to the graph. These points represent two farms that both have the same production average of 10,000 kilograms. The red farm has an efficiency of 1.4 kilograms of milk per minute in the robot, whereas the green farm has an efficiency of 2 kilograms per minute in the robot. Let’s look at each farm’s potential productivity when robots are full and there are 1,180 minutes available for milking.


Red Farm: 1.4 kilogram per minute x 1,180 milking minutes = 1,650 kilograms of milk

Green Farm: 2 kilograms per minute x 1,180 milking minutes = 2,360 kilograms of milk

The result is 710 kilograms more milk per day for the green herd and milk revenues of close to $500.

Yet these two herds both have the same average milk production over 12 months. Greater efficiency in robotic milking therefore makes it possible to increase dairy income with the same assets.

Evaluate the herd as a whole

Since March 2020, our robot report has provided four new calculated values that can help monitor herd performance. These new values make it possible for producers to compare with their herd’s 12-month average, the provincial average and the top 20% of their breed.

In the example shown in Table 1, the herd has good productivity per cow and a robot that is close to maximum capacity.


Key robot indicators

Even though their production per cow is very good, their efficiency is below the provincial average. The two lines project the permitted production of milk and fat when the robot is at 10% free time. For this producer, we can conclude that the robot is full with only 52 cows using the robot. The question that arises is: Which cows are penalizing them the most?

Which cows are the least efficient?

We are working on a project to identify the least efficient cows in the herd. To validate the direction of the project, 11 producers with DeLaval and Lely robots shared their evaluation of their current situation and what they need:

  • Classifying cows by efficiency is not easy and is a long and tedious process.

  • The methodology is not standard, according to the manufacturers.

  • The tool has to gather all the data in one place.

  • It must have an efficiency index expressed in dollars.

Data from five herds with Lely robots have been uploaded on a daily basis. This data is being used to develop a model for a tool that will try to rank cows according to efficiency.

The ultimate objective of the project is to provide a list that ranks all the cows in the herd to see which are impeding herd efficiency and which are the best in the herd. The 96-hour International Committee for Animal Recording (ICAR) standard is being used for the average calculation range.

For the purposes of this article, several rows in Table 2 have been hidden.

Example of cow ranking based on milk per minute in the robot

Click here or on the image above to view it at full size in a new window.

Orange cells indicate that the value for this cow is below 80% of the herd average, and green cells indicate that the value for this cow is above 120% of the herd average. Ranking is based on the milk value per minute in the robot. We can see the six cows with the worst and best performances according to this value.

The six parameters we focused on include:

1. Efficiency in kilogram of milk per minute in the robot. It is also possible to make the same calculation based on successful milkings, in order to highlight cows that have had milking failures.

2. Milk value per minute in the robot. Cell 1 is multiplied by milk components and price of each of the milk components.

3. Fat value per minute in the robot. Many producers wanted to have a ranking based on fat. This is the same calculation as the previous one but using kilograms of fat per minute, the fat test and the price of fat per kilogram.

4. Time spent in the robot per day. This allows you to see that although some cows are very productive, they use far too much of the robot’s time.

5. and  6. The preparation time and milking speed based on a 96-hour average. These are two important elements that can explain why a cow has a low rating. Preparation time is primarily connected to a cow’s temperament and udder conformation, whereas milking speed is more related to a cow’s genetic potential. In the case of cow number 95, a long preparation time and a slow milking speed explain her ranking. However, in the case of cow number 48, only her milking speed is a problem.

What should you remember about this new ranking?

When you look at individual cows, you can see that it is very variable. There are some very productive cows that are not very efficient and others that are much more so. To better understand the impact of efficiency, take, for example, a cow in Table 2. Cow number 48 produces 48 kilograms of milk and uses the robot for 47 minutes per day. If all the cows were like her, the robot would be full with only 25 cows, and production per robot would be below 1,200 kilograms.

The herd average is 21.9 minutes per cow in the robot per day, and the average production is 38.5 kilograms of milk. If this cow was sold, it would free up 47 minutes that could be filled by animals closer to the herd average: 2.15 cows per 38.5 kilograms of milk, therefore a total of 83 kilograms of milk, amounting to gains of 36 kilograms of milk in the same number of minutes. Even better, if you consider the best cows; cow number 105 gives 49.5 kilograms of milk in 17.2 minutes.

We are currently working on the automation of this process to provide a tool for our clients that will allow them to rank cows according to their efficiency. Producers will be able to immediately see whether milking speed, preparation time or both are connected to an animal’s performance. end mark

Gervais Bisson
  • Gervais Bisson

  • Dairy Production Expert
  • Robot Milking and Strategic Adviser
  • Lactanet
  • Email Gervais Bisson