How property sourcers analyse deals at volume

Sourcing economics only work if the top of the funnel is wide. If a patch produces 300 new listings a month and roughly 2 percent of them can be bought at a price that leaves margin, that is six possible deals, of which perhaps two survive viewing, negotiation and legal work. A sourcer charging a typical fee of 2,000 to 5,000 pounds per packaged deal is therefore living on two transactions from three hundred candidates. Miss half the listings and the month produces one deal, which is the difference between a business and a hobby.

The corollary is that time spent deeply analysing a property that was never a candidate is the most expensive mistake in the workflow. The first pass exists to say no quickly, hundreds of times, so the hours go into the handful of properties where the answer might be yes.

The first pass asks only three questions. Is the asking price plausibly below what comparable sold prices support? Does the gross yield clear the local floor if it were rented as it stands? And is there anything in the listing that kills it outright: a short lease, cash buyers only wording, an auction guide price pretending to be an asking price, or a retirement restriction?

Screening by hand means reading the listing, pulling two or three sold prices from Land Registry data, and doing one division. Done briskly that is five to ten minutes per property, which at 300 listings is 25 to 50 hours a month before a single real appraisal has happened. This is the stage automation eats: software can apply the same three questions to every new listing the moment it appears and discard the obvious noes without a human reading anything.

A property that survives screening gets the full treatment, and the full treatment has a fixed shape: end value from sold comparables, refurbishment cost from the property's actual condition, rent from lettings comparables matched on bedroom count and type, then finance costs, buying costs and the exit modelled properly. Each figure needs evidence attached, because the investor buying the deal will check.

The end value is the number most sourcers get wrong, and the usual failure is valuing off asking prices or a single flattering comparable. A defensible figure comes from a pool of sold comparables filtered for type, size and distance, with the outliers excluded for stated reasons and a price per square foot cross-check. The refurbishment estimate fails in the opposite direction: a flat rate per square metre applied without looking at the photographs, which prices a tired but sound house as if it were a wreck, or a wreck as if it needed carpets.

Manually, a proper appraisal of this shape takes one to three hours. Software that reads the listing, the photographs, the floor plan and the EPC can produce the same structure in minutes, which changes the economics of the shortlist: appraising twenty candidates properly stops being a week of evenings.

A deal pack that says trust me is worth nothing. A deal pack that shows its working is what gets a fee paid: the comparable table with dates, distances and floor areas, the room-by-room refurbishment schedule, the rent evidence, every cost in the stack, and the return on each exit that makes sense for the property. Investors who buy repeatedly from the same sourcer describe the same reason: the numbers survived their own checking.

The pack should also state what would change the answer. If the deal works at a 92,000 pound purchase but not at the 105,000 pound asking price, the pack should say so and say why, because that number is the negotiating instruction.

Automation is strongest exactly where the volume is: watching every new listing in a patch continuously, applying the screening rules identically every time, and producing the full appraisal structure on the shortlist with the evidence attached. A tool like PropDetect runs that loop end to end: its Deal Radar watches Rightmove around the clock, every match gets the full appraisal automatically, room-by-room refurbishment costs are estimated from the listing photographs, the end value comes from up to 100 sold comparables with exclusions explained, and only deals that clear the sourcer's own threshold arrive in the inbox. The offer price on failed deals comes with it, which is the negotiating instruction stage three needs.

What automation cannot do is the part the fee is actually for. It cannot view the property, smell the damp, read the seller's real motivation from a conversation with the agent, negotiate the price, or take responsibility for the recommendation. Sourcers who treat software output as a finished deal pack get found out at the first viewing; the ones who treat it as a screened, evidenced starting point get their evenings back and their volume up. Whatever tool does the screening, the professional obligations do not move: sourcing legally in the UK still requires redress scheme membership, client money protection where money is held, ICO registration and anti money laundering supervision.