Initial solution used a ranked list of unique counts joined back to the original data set and filtered for top 5; then realised that the same thing could be achieved by simply sorting the data set, using the mighty multi-row formula tool to set the ranking and then filtering for top 5.