I've been reading through the ipsea website which gives insight into this. As ilikemysleep says, LAs do in fact use broad criteria (moderated by focus groups, not sure if that's what they're actually called - I forget) to achieve some consistency in agreeing/refusing SA. This is different from having blanket policies. That's the theory according to ipsea, but of course there is a fine line, and there is no accounting for how the 'broad criteria' might be interpreted by short-sighted paper-pushing office monkeys at the LA who might use them as blanket policy. The SENCOP, however, is clear on this.
Ipsea also suggests, like ilikemysleep that requests for SA are likely to be turned down when there is limited evidence that the school has made a decent attempt or had a chance to make provision to meet the child's needs. The more evidence the better. For this reason I am biding my time and allowing IEPs to be set up and reviewed before applying for SA because this will strengthen my case that the school is unable to provide for dd2 from their devolved budget.
However, a balance has to be met for those circumstances where parents have been banging their head against a brick wall and no provision is forthcoming, in which case parents may have no option but to go ahead and apply for SA. In those instances, setting up a papertrail would be vital.
Oh and sweettea we had a very similar situation with dd1 (Dyslexia) who failed to make any sublevel NC progress 2011-2012. School tried to pretend I was wrong (despite having the NC levels in black and white on two school reports) and then when I requested a meeting with SENCO/CT to discuss, lo and behold, suddenly some sublevel progress had happened, magically. Reading between the lines I think dd1's NC levels had been somewhat inflated from KS1 SATs. Schools try their hardest not to let NC levels go backwards, because that makes them look..... er..... shite. It is frustrating, to put it politely.