Methodology & data sources

Every number, shown its working.

We don't ask you to take our word for it. Here is exactly where our data comes from, how each figure is calculated, and how you can check it yourself. The rent is an estimate — but the sold prices, the yield maths and the scoring rules are all things you can verify.

The proof

Where our data comes from

Three inputs feed every analysis. Two of them are hard data you can check independently; the third (rent) is an estimate we deliberately keep honest with a confidence score and a range.

HM Land Registry — Price Paid Data

Verifiable data

Actual sold prices for the property's postcode.

Official record of every completed sale in England & Wales. We pull the most recent transactions for the exact postcode and show them next to the asking price, so you can see what nearby homes really sold for — not what agents hope to get.

Crown copyright · Open Government Licence v3

The live listing itself

Verifiable data

Price, property type, tenure, lease length, EPC, description.

Every fact we score comes from the actual Rightmove/Zoopla listing you paste in — not a guess. Tenure, lease years remaining and red-flag phrases (‘cash buyers only’, ‘needs modernisation’) are read straight from the listing copy.

Sourced per-analysis from the listing you submit

AI rental model + regional benchmarks

Estimate

Estimated monthly rent, cross-checked against benchmarks.

A Claude model estimates market rent for the property type and location, then we clamp that estimate to within ±40% of a static bedroom-and-region benchmark. If the AI figure is pulled back significantly, we lower the confidence score accordingly.

Estimate — always shown with a confidence % and a range

Estimated rent

How we estimate rent

Rent is the one figure nobody can know with certainty until a property is let, so we're explicit about it. Our estimate is built in two steps:

  1. AI estimate. A Claude model prices the specific property type and location using current UK rental conditions.
  2. Benchmark sanity check. We compare that against a static benchmark built from bedroom count, region and property type, and clamp the AI figure to within ±40% of it. A figure that gets pulled back hard has its confidence reduced.

You always see the result as a range (not a false-precision single number) plus a confidence score. A full postcode gives a tighter, more confident estimate than a town name alone.

The maths

How we calculate yield

Gross yield is a plain, industry-standard formula — no black box:

gross yield % = (monthly rent × 12) ÷ asking price × 100

For example, £1,100/month on a £200,000 property is (1,100 × 12) ÷ 200,000 × 100 = 6.6% gross yield. Net yield goes further, subtracting running costs (management, insurance, maintenance, void periods) before dividing — which is why our net figure is always lower than gross. Both use the estimated rent above, so if you have a better rent figure, the yield updates with it.

The deal score

How the 0–100 score works

The headline score is deterministic. The AI explains the deal in plain English, but the number itself is decided by fixed rules — so the same property always scores the same, and you can reproduce it by hand. Every deal starts at 50, then:

Yield

Up to +30 for 9%+, scaling down to −15 for sub-3% yields — the single biggest driver.

Location demand

+8 in high-demand cities, −6 in the weakest rental markets.

Condition & features

Small positives for modern refurbs, gardens, parking, transport links, tenant in situ, HMO use.

Red flags

Sharp penalties for structural issues (−20), Japanese knotweed (−25), ‘cash buyers only’ (−20), damp, age restrictions.

Leasehold

Penalty scaled by lease length — heavy under 70 years, none at 90+.

The final score maps to a verdict at fixed thresholds:

BUY

68–100

MAYBE

48–67

AVOID

0–47

On every result we also list the individual score drivers — each with the delta it contributed and where it came from — so nothing about the number is hidden.

Honesty

What the confidence % means

Confidence reflects how much the inputs agree. Above 85% means clear, consistent numbers; below 70% means conflicting signals — a thin listing description, a town-level location instead of a full postcode, or a rent estimate that needed a big correction. We show it rather than hide it because a confident-looking number built on weak inputs is worse than an honest one.

Check us

How to verify it yourself

  • Compare our sold prices against HM Land Registry's own Price Paid tool — the exact same public dataset.
  • Re-do the yield by hand with the formula above — it will always match.
  • Sanity-check our rent estimate against current listings on Rightmove or SpareRoom for the same area and property type.
  • Before you offer, get contractor quotes for any refurb and commission a full survey — our score flags risks, it doesn't replace due diligence.

Not financial advice. DealFlow AI is an analysis tool, not a regulated adviser. Estimates are indicative and rental/sale values can change. Always conduct full due diligence before investing.

Ready to see it on a real listing?

Analyse a property free →