If that survey were properly randomly sampled across the UK population, the 95% confidence interval for that sample size would be around +/- 200,000 people: i.e. we are 95% sure that between 800,000 and 1.2M people canceled broadband (or whatever the question is in the survey).

Even if the 6000 people surveyed are selected from those who seek help from Citizens Advice, there are ways to make reasonable inferences from that.