Survey data on actual past and expected future output movements are often available in a more timely manner than official statistics. We quantify qualitative survey data from the UK’s Confederation of British Industry to generate nowcasts on output growth of the whole economy and the sectors of Manufacturing and Services. Implementing model weighting techniques, we find that a weighted model that combines the survey responses, expected and realized, with an AR(1) is the most accurate one for the period before the pandemic. Survey-based data can improve the accuracy, relatively to an AR(1), of out of-sample nowcasts of the aggregate monthly growth of the GVA. At the sectoral level, this is the case only for Manufacturing. During the pandemic we cannot however draw safe conclusions. Finally, we make suggestions about future use and improvement of the survey-based qualitative data.