Lab company Qlip is transforming into a data-driven company through big data analyses

Big data also creates new business models in the lab world. The Qlip company, active in the dairy chain, applies data science to the enormous amounts of data from milk samples. This provides the company with different insights and new business models.

By: Dimitri Reijerman

Qlip carries out audits and inspections within the dairy industry, but the company also carries out large-scale checks on milk production at dairy farmers and dairies. The company has set up the largest dairy laboratory in Europe for this purpose.

Erik Bos, ICT manager at Qlip, indicates that the large-scale checks generate an enormous amount of data: “We process almost 15 million samples per year, about 40,000 to 50,000 per day. These are analyzed completely automatically. We do this using infrared. We measure each milk sample for 1060 data points. Based on how these dots correspond with each other, information can be derived about numerous compositional aspects of the sample,” says Bos.

Qlip has now collected more than 80 billion database records, from samples checked in recent years. According to Bos, until recently the processing and analysis of such amounts of data – big data – was reserved for the Googles of this world. Data technology has now evolved to such an extent that medium-sized companies such as Qlip can also sink their teeth into it.

Collaboration with Wageningen

Qlip conducts research with Wageningen University, among others, into new applications based on big data. Bos gives a practical example: “The university has two experimental farms. They conducted a study with cows that stayed in a breathing room for two weeks. The exact emissions of greenhouse gases and phosphates were measured. The cows were also milked during this study. All this data came to us. We relate our data to that of the university.”

These big data analyzes potentially make it possible to calculate the phosphate standards no longer on an average of a group of dairy farmers, but per individual dairy farmer. Moreover, it is also technically possible to measure this data per cow, so that the farmer can optimize the feed per cow. This data-driven approach can potentially improve milk yields and reduce emissions of phosphate or other harmful substances.

The enormous mountain of data can also provide an answer to the question of how likely it is whether a farmer meets the requirements for grazing. This fact is important for Qlip, because the company also carries out inspections on farms.

Additional income from big data

Qlip now generates additional income with this method. It is also in the transformation from a classic lab company to a data-driven company, says Bos: “We are now a data company. We say this in our communications and it also influences which people we hire. Our data collection is still primarily lab work.”

Bos continues: “We have people who work full-time on data analysis and data science. We do this with tools that an SME company currently has available. Think of SQL databases, Microsoft Azure and 'R' as a statistical language. When we develop models to measure something, we take a limited data set from the past and models are tested based on a control set. Machine learning also plays a role in this. What are the biggest predictive factors coming out of this? What suggestions does machine learning make to improve our models?”

According to Bos, software around data science and machine learning has become indispensable for companies like Qlip: “This technology has now become indispensable. “

Advice to other companies

Based on his experiences at Qlip, Erik Bos also has advice for companies that want to get started with big data: “I would like to tell them: just start, but start small. Don't immediately look for the pot of gold, first record everything in a database. You can get a lot out of that.”