[Side note: current coverage is London, Manchester, Liverpool, Birmingham, Nottingham, Bristol, Newcastle, Leeds, Edinburgh and Glasgow... any idea when Cambridge might be added to the mix?]
One thing I've noticed in scanning different sites is that the data at the source is often quite limited, possibly intentionally. The different sites have the same limited info; and I'm fairly sure the reason you don't have searches based on "children allowed" or "pets allowed" is because this isn't recorded reliably (or in any way consistently)... it's just occasionally in the text description, worded however (possibly with spelling errors).
As long as you're aggregating the same limited data, there's only so far you can go.
Input from searchers could be the way out -- like the idea of letting users flag scam properties, or add other info after talking with the agent -- but this may get hard to protect from abuse, frustratingly.
1) We are indeed trying to solve this problem, our approach is collaboration amongst house hunters. Users who found scam/expired properties can report them, then other won't find it. We also intend to integrate AI in this problem. Pls Keep an eye on this ;)