This is based on the judgment in the relevant case. I hope that I haven't done any disservice to the arguments put forward or the proper facts. Apologies if that's the case...
Right. The case of Wilson. Here there were seven knife-point robberies over a period of four days by a masked man carrying a knife. A few days later Mr Wilson was arrested after police officers saw him and noted that his clothing matched the description given of the robber by the victims.
He was arrested and said that he was of no fixed abode. This wasn't true and when his flat was searched, property belonging to three of the victims were found. So far so good. It looks at this stage a pretty good case against him. At least there are some pertinent questions to be answered as to what he was doing with this property.
The defence seemed to have a few points; the description of him didn't match the victims descriptions, especially that Mr Wilson had a strong Scottish accent that the robber (or robbers presumably - this doesn't seem to have been explored) didn't seem to display. It seems then that the Prosecution felt that this wasn't enough, so seven months later got some new evidence:
"The issue on this appeal relates to the admissibility of the evidence of Miss Alexis Young, who was responsible for monitoring crime in Islington. The defence disputed the evidence could be adduced. It is the ruling of the Recorder at the trial that it could be adduced which is the subject matter of this appeal. The evidence of Alexis Young was that she had searched the police computer records for similar offences using two criteria, namely "robbery" and "mask" which brought up two other crimes, none of which were similar to the present offences. She also used a criteria "Islington", "lone white male 20 - 30" and "knife". These searches produced no results. She therefore concluded that no similar offences had been committed in the area between the date of the offence and 31st May 2007."
It isn't clear from the judgment whether the report from Ms Young was any more scientific than that, but it would appear not. And that’s it. That’s all she did.
The problem with the last sentence "She therefore concluded that no similar offences had been committed in the area between the date of the offence and 31st May 2007" is that this is a conclusion with absolutely no basis whatsoever in terms of analysis. The report seems to me to be valueless as evidence (before even looking at the maths, see how many cognitive biases there are). The faults with the methodology are so numerous you wonder how it got in front of the jury. If I get bored one day, I might go through all the ones I can see.
The basic point is that evidence of this character could be relevant to whether Mr Wilson was the robber, but only if it was done properly and rigorously (and isn't it really expert evidence?) Without this, it's the legal world's version of homeopathy, it may look good but it's without any substance.
Two other points. From a legal point of view a massive potential problem for the prosecution is that it opens the door to the defence to ask for reams of similar disclosure in a suitable case to prove the opposite (even cherrypicking the terms that produce the best results for them from a database?). This will involve a lot more work for the police (unless there turns out to be one rule for the Prosecution and one for the Defence? Surely not!).
Lastly, on one view there is a heuristic (it's got at least as much validity as the Court of Appeal's view) argument that this evidence actually makes it less likely that Mr Wilson was guilty, but that's for another day...