Well - QuickBooks does HALF a job of entering the journal into the bookkeeping automatically - you still have to do the other half yourself - i.e. the net pay debited to payroll clearing and credited to the bank.
I have a couple of clients that don't use the bank feeds so this additional net pay manual journal is necessary - but also, even if the bank feeds are used, I find it better to manually create the net pay journal, so that the bank feeds then 'match' rather than 'add' the figures in. This way, I pick up much quicker if there is a discrepancy on net pay and the payroll clearing is balanced back to zero every pay run.
With regards to the 'why I don't like it' - it's very dis-jointed in how you get around it - and not very obvious as to whether things have been carried out or not. The reports are just OK rather than good - and its easy to get things wrong and not realise it.
That said - once set up, my clients seem to get on with it OK for everyday payroll, its just when there is an issue, that they can't even start to sort it themselves.
While its free, it's perhaps worth consideration, but I certainly wouldn't like to try and handle auto enrolment with it! (but I have no experience of this yet with QuickBooks payroll, so may be being unfair) and I feel it would be expensive if they brought in the £1 per month per employee that they're suggesting its likely to be, if there were more than half a dozen employees. For just a couple of employees, it's probably fine.
My personal opinion is that Moneysoft Payroll Manager beats it hands down - and if your payroll is for a large number of employees (more than 20), using an importer such as Business Importer is reasonably easy to get it from Moneysoft into QBO - though this I'm saying for a bookkeeper, rather than a non-bookkeeper client.