We conclude this report with a system design and proof-of-concept to show how an adaptable hybrid sentiment classification system is able to improve sentiment analysis for organisations. GreenOnline, a service company in the field of customer services, wants to be able to quantify sentiment for organisations precisely, to create new services for organisations. To start with, this sentiment analysis will be based on Twitter messages. The main challenge during this research was that Tweets, short WOM (Word-Of-Mouth) messages that contain only little words, are highly abbreviated and sentiment is expressed in subtle ways with irony, sarcasm, slang and other linguistic shades of grey [9]. Therefore, the focus of this thesis project was to design a system that is able to combine different sentiment analysis techniques to find sentiment. Also, not only existing algorithms were combined, but also information from the message (message attributes) are regarded as a way to determine the sentiment or which algorithm will classify the sentiment of that message best. Overall, it was regarded that all these different elements leave room for optimisation, for what algorithms and attributes to use and for what messages to select from Twitter for an organisation. To support a process of optimisation for a campaign or organisation another goal was to embrace the ability of system optimisation by (GreenOnline) customer service experts. The result is a design and proof-of-concept implementation of a hybrid and adaptable sentiment analysis system design, which is using implementations of three sub classifier algorithms and message properties, that are combined by a hybrid sentiment classifier in a sentiment value of negative, positive or neutral. This proof-of-concept implementation showed a performance of 71,2% which is a great improvement with respect to the single sub classifications of which the best performance was only 58,2%. By improvement of customer service experts this performance can even grow further.