Ok, fair point.
The number (94%) is of all the cases that were flagged.
The police in the UK records all the crimes and incidents that took place, and then decides whether to pursue the case or not, once the decision to pursue was made an official case file is openned and the crime is flagged.
Now, very well, I have went through the data again to be sure of the numbers:
Considering that flagged crimes are ones that the police have established official police files for, and that we're addressing 94% of them that were assigned an outcome, it comes to finding what this outcome is which we need to move out of this report and into the CPS data to know this information.
I did not find the full data for 2017-2018 report, but I found it for the 2015-2016 report.
From page 58-60: http://sro.sussex.ac.uk/id/eprint/70598 ... ROCESS.pdf
So, of those 94% cases with assigned outcome, roughly 78% goes to court, and then the conviction rate applies.
Noting, the reason why the conviction rates differ is that this is from 2015\2016, while the previously quote conviction rates are from 2017\2018.
Also noting that these percentages are across hate crimes in general, not just speech offenses, but they give an average that we can use for a clearer picture.
I'm not going to read a 40 page police report, 59 CPS report, and a 214 page final report all in one day to make a single post, so we'll use this average for hate crimes in general to get rough estimates.