Methodology

    How the verdict is built.

    A buyer-side framework built on Dubai Land Department transaction records — backtested, calibrated, and honest when the evidence is thin. The free WhatsApp reply is one word: Above market, At market, or Below market.

    01

    You send a listing on WhatsApp

    Paste a Bayut or Property Finder URL into our WhatsApp. No account, no form. The listing URL is stored as a user-provided reference — we do not scrape or cache portal listings.

    02

    We confirm the unit before the verdict

    We parse what the URL contains, then ask you to confirm: building, bedrooms, size, asking price. Optionally floor, view, condition, and tenancy status. Every detail is confirmed back to you before we run the numbers.

    03

    Data sources

    Verdicts run exclusively on Dubai Land Department transaction records with documented provenance. Service charge context comes from the DLD Service Charge Index.

    No scraped listings, no asking-price averages, no cached portal data. Every input carries a provenance field.

    04

    Free verdict — one word back

    The free WhatsApp reply is a single label: Above market, At market, or Below market. That's the whole free reply — no range, no comps, no confidence tier.

    Time to verdict is a promise: we target minutes, not days.

    05

    Paid report — the numbers behind the label

    If you want the numbers behind the label, the €50 Negotiation Report includes the fair price range, three Dubai Land Department comparables, a calibrated confidence tier (high / medium / limited), one key caution, and a written negotiation script — opening offer, counter range, walk-away, broker message.

    06

    Calibrated confidence (inside the paid report)

    • High: deep, recent comp set with a strong match to the subject unit.
    • Medium: moderate match, adequate transaction depth.
    • Limited: thin evidence — the report returns market context, not a false-precision number.
    • Confidence labels are empirically calibrated against a backtest — an ‘80%’ label really lands ~80% of the time.

    06

    How we validate it

    • Pre-launch backtest: hidden historical transactions, predicted using only prior data. Median error and share of sales inside the predicted range are measured community-by-community.
    • Human review queue during alpha, with auto-send unlocking per confidence tier.
    • Blind sample review by an accredited valuer.
    • Day-30 outcome follow-up: did the property sell, at what price. Outcomes are what make the dataset defensible.

    07

    Insufficient evidence, when honest

    In thin-data communities we return market context and an ‘insufficient evidence’ note — never a fabricated verdict. A number you can't trust is worse than no number at all.

    08

    What we are not

    We are not a licensed valuer. We do not list properties. We do not represent sellers. We do not take broker commissions or transaction cuts. Buyer-side only, permanently.

    Read the methodology. Then get a verdict.

    Get a free verdict on WhatsApp