In summary - and much more details below: there is massive variation between police regions in how well rape cases are handled, but if every force followed current "best practice", then over 40% of reported rapes could result in the rapist being convicted.
The BBC's angle in the reporting is in the massive variance in "no crime" classification of reports. This is an update of very similar research they did, which I wrote about briefly two years ago.
As in the BBC's previous study, and as in earlier similar studies carried out by the Fawcett Society, the main finding is of massive variation from region to region.
Reporting rates by region do not vary significantly - though there is a little variation between Cheshire (0.17 per 1,000 population) and London (0.43 per 1,000 population).
On marking reports as "no crime", the rate varies between 2.4% in Gloucestershire and 30.0% in Kent. A small number of "no crimes" is expected - Kelly, Lovett and Regan's 2005 study gives several good reasons (other than the stereotypical "false accusation") why this might occur. However, it's clear from the discrepancies that a lot of police forces are heavily overusing this method of closing a case.
The "sanction detection" rate also shows significant variation between regions. This is the rate at which reported cases result in either a charge against a suspect, or the police formally cautioning a suspect. (The latter is relatively unusual in rape cases, though still uncomfortably common).
In Lincolnshire and Bedfordshire, only 11.1% and 11.8% of cases include a "sanction detection". In Durham, the figure is 60.8%, and in South Wales, 49.8%. Again, there's clearly a major difference in procedures.
Once the case has moved to the CPS for prosecution, there are again regional variations. Ignoring Lincolnshire as an outlier, as the police are clearly only passing the most obvious of cases to the CPS there, Nottinghamshire CPS drop 36.3% of cases before court. Dorset CPS only drop 5.4%.
Then, in court, the conviction rates again vary - Surrey, Dyfed-Powys, and Hampshire have conviction rates in court less than 60%. (Dyfed-Powys occasionally get confused and prosecute the victim instead). Meanwhile Warwickshire, Devon and Cornwall, and Leicestershire are able to secure a conviction in around 85% of court cases (not always for rape, though - sometimes for lesser offences).
Multiplying the various attrition rates together isn't completely valid (the figures for prosecutions are not for the same cases as the sanction detections) but gives a rough indication of the combined effectiveness of the local police and CPS, without having to do the massive longitudinal studies that Kelly, Lovett and Regan did, where there is almost a five-fold difference between the most and least effective forces. (Since this is only a rough estimate, I won't name the 'best' and 'worst' regions)
Baroness Stern's report on rape prosecutions strongly noted that investigation and prosecution would be considerably more effective if everyone just followed the guidelines that had already been written.
Cross-tabulating the various figures against each other suggests that - with the exception of a few outliers such as Nottinghamshire - the effectiveness of the CPS in prosecuting cases doesn't depend much (or even at all) on what proportion of cases the police pass to them. In other words it is generally not the case that the police referring more cases to the CPS will just lead to the CPS either dropping those cases or being unable to secure a conviction.
We can therefore also look at a theoretical "best" region, which improves its "sanction detection" and "conviction" rates to the best found in real regions, by application of best practice.
The result of this is not particularly surprising, perhaps. The theoretical "best" region - merely on current best practice in policing, investigations, prosecutions, and court cases - would be able to secure convictions, if not for rape then at least for some sexual offence1, in around 40% of all reported cases.
Furthermore, best practice is continuing to improve - that "theoretical" figure is itself noticeably increased from even two years ago.
It wouldn't be impossible - police forces have made significant improvements in only a few years before - for that 40% figure to be achieved by the end of this decade. If that doesn't happen, it will be solely because it wasn't considered a high priority.
I'll be writing to the Ministry of Justice (responsible for the CPS) and the Home Office (responsible for the police) soon to ask what plans they have to ensure that under-performing police forces are rapidly brought up to the level of the best.
1 In practice, around half will be for lesser offences.