This paper presents a comprehensive survey and how-to guide for NLP research on counterspeech against hate. Counterspeech is defined as non-aggressive, evidence-based responses that challenge hate speech, aiming to reduce online and offline violence. The paper discusses the growing interest in NLP for analyzing, collecting, classifying, and generating counterspeech to reduce the burden of manual production. It outlines three key steps for conducting NLP research on counterspeech: task design, data selection and evaluation, and evaluation. The paper reviews existing NLP studies and resources, proposes common concepts and best practices, and highlights open challenges in the field. It also discusses the importance of ethical considerations, data collection biases, and the need for diverse annotators. The paper emphasizes the importance of understanding the strategies used in counterspeech, such as fact-based responses, empathy, and constructive tone, and the challenges in automatically detecting and generating effective counterspeech. It also highlights the need for evaluation metrics that can accurately assess the effectiveness of counterspeech generation. The paper concludes that while NLP has the potential to make counterspeech more scalable, researchers must be aware of the consequences of their choices to avoid spreading further harm.This paper presents a comprehensive survey and how-to guide for NLP research on counterspeech against hate. Counterspeech is defined as non-aggressive, evidence-based responses that challenge hate speech, aiming to reduce online and offline violence. The paper discusses the growing interest in NLP for analyzing, collecting, classifying, and generating counterspeech to reduce the burden of manual production. It outlines three key steps for conducting NLP research on counterspeech: task design, data selection and evaluation, and evaluation. The paper reviews existing NLP studies and resources, proposes common concepts and best practices, and highlights open challenges in the field. It also discusses the importance of ethical considerations, data collection biases, and the need for diverse annotators. The paper emphasizes the importance of understanding the strategies used in counterspeech, such as fact-based responses, empathy, and constructive tone, and the challenges in automatically detecting and generating effective counterspeech. It also highlights the need for evaluation metrics that can accurately assess the effectiveness of counterspeech generation. The paper concludes that while NLP has the potential to make counterspeech more scalable, researchers must be aware of the consequences of their choices to avoid spreading further harm.