Advanced chemical sensors are beginning to become abundant in the industry. The last decades of advances in sensor development have produced relatively cheap and very robust chemical sensors that can ensure that a given production always runs at an optimum with no scraps or rejects through on-line measurements and process analytical technology. Sensors can be put at key locations through the production process and generate large amounts of very complex data, giving near-instantaneous fingerprints of the chemical composition of a process stream. All these sensors will generate massive amounts of data that needs to be stored and handled correctly in order to reveal its full value. Producing a lot of data quickly is in and by itself not very useful – the data must be relevant and of a very high quality as well. In this talk I will demonstrate how the careful processing of data generated from complex sensors can be combined with multivariate data analysis and machine learning to generate relevant knowledge under the premise: to make measurements only where and when it matters.