Which organisations actually say no: a league table, not a vibe
Reputation for openness and actual measured behaviour aren't always the same thing. Now that enough releases have real archived text behind them, we can rank organisations directly - not by how much they publish, but by how often what they publish turns out to be a refusal, an exemption, or "we don't hold that."
The headline contrast
The Serious Fraud Office cites a refusal or exemption in 90.3% of the releases we could read (31 of them) - the highest rate of any organisation with a meaningful sample in this tracker. The Home Office, by contrast, sits at 1.8% - and it isn't a small sample either: 1,660 releases, by far the largest of any organisation here. The Home Office publishes roughly 54 times as much as the SFO does in our dataset, and refuses at roughly a fiftieth of the rate. Volume and openness are clearly two different axes, not proxies for each other.
The full picture
| Organisation | Releases read | Refusal / exemption rate |
|---|---|---|
| Serious Fraud Office | 31 | 90.3% |
| Civil Nuclear Constabulary | 1,131 | 49.2% |
| Civil Nuclear Police Authority | 1,108 | 49.0% |
| Foreign & Commonwealth Office | 22 | 27.3% |
| Metropolitan Police Service | 70 | 18.6% |
| British Transport Police | 129 | 4.7% |
| Ministry of Justice | 89 | 2.2% |
| Home Office | 1,660 | 1.8% |
| Youth Justice Board for England and Wales | 27 | 0.0% |
| UK Visas and Immigration | 182 | 0.0% |
| Immigration Enforcement | 79 | 0.0% |
| Border Force | 15 | 0.0% |
The Civil Nuclear Constabulary and its authority sit in the middle at just under half - notable because we've written before about how disproportionately visible CNC is on GOV.UK's feed compared with territorial forces. High visibility and a high refusal rate can coexist: CNC isn't just publishing a lot, it's also saying no to close to half of it.
The four organisations at a clean 0% are worth a caveat rather than a celebration: Border Force's sample is only 15 releases, right at our minimum threshold, so treat that one especially as provisional rather than a track record.
Speed only tells a story for two of these
We can only measure response speed where a release states a date reliably, and for most organisations here that leaves too small a sample to rank fairly - only the Civil Nuclear Constabulary (42 working-day median, 432 dated releases) and its authority (45 working days, 411 releases) have enough. Both sit close together and both are, per our response-time piece, well outside the 20-day statutory limit on a typical release.
Reading this responsibly
This ranks organisations with real archived text - GOV.UK releases and, where relevant, police-portal content - not everything this site tracks. Most territorial police forces come to us through WhatDoTheyKnow, which we deliberately don't pull full text from (see our about page for why), so they simply aren't measurable here yet, not absent because they're clean. "Refusal or exemption rate" is a text-presence signal, not a legal audit - a document can mention an exemption while explaining it doesn't apply. We set a minimum of 15 releases to appear at all, but smaller samples within that (the four 0% organisations, the Foreign & Commonwealth Office's 22) deserve more caution than the ones running into the hundreds or thousands.