Methodology

The AgentScore is a single 0–100 number. There is no editorial input, no paid placement, and no agent can change their position by contacting us.

The four pillars

How each pillar is normalised

• Sale-to-list is mapped linearly: 93% of asking scores 0, 102% of asking scores 100.
• Days on market is inverted: 7 days scores 100, 75 days scores 0.
• Volume uses a logarithmic curve so an agent with 300 closings a year doesn't flatten everyone else. Recent 12-month volume counts for 70%, career total for 30%.
• Reputation maps a 3.5-star rating to 0 and a 5.0 to 100, then discounts it by review confidence — a perfect 5.0 from four reviews lands well below a 4.8 from nine hundred. When we can also match the agent's brokerage office on Google, that office rating blends in at a lower weight, so a single platform's review culture can't decide the pillar on its own.

Confidence grades

A score is only as good as the evidence under it, so every agent carries a separate A–D confidence grade. The grade never changes the score — it tells you how much to lean on it.

Ranking order also respects evidence: when two agents land within a few points of each other, the one with more verified closings behind it sorts first, so a thin sample can't leapfrog a well-documented record on noise.

Missing data

Public directories don't publish every metric for every agent. When a pillar has no underlying data we drop it entirely and redistribute its weight across the remaining pillars. We never impute a value, and we never treat "unknown" as "average" without saying so — every agent page shows exactly which pillars were counted, and explains in plain English why a metric is blank (a buyer-side specialist, for instance, has few listings of their own to measure).

Where the data comes from

Agent figures are compiled from publicly visible real estate directory and profile pages. Sale-to-list ratio and median days on market are not taken on trust: we read the price history of an agent's recent closings and calculate them ourselves, anchoring on the price the home was actually marketed at rather than a relist. Brokerage office details, ratings and photos come live from Google. Data is cached and refreshed on a rolling schedule, so a figure may lag its source. Verify anything you plan to act on, and treat this as a shortlisting tool rather than a substitute for talking to an agent.

Why sale-to-list and days on market are samples

These two figures are medians over the closings an agent's public profile actually itemises — which is usually a handful, not their full book. An agent with 150 sales in a year may publicly disclose five, and we can only measure what is disclosed. So wherever we show these metrics we also show the sample they came from ("n=5"), and the confidence grade drops when the sample is thin. A thin sample is not a criticism of the agent and a strong number over few closings is not proof of a pattern: it is an indication, and we would rather label it honestly than round it up into a verdict.

What we don't do

• — Ask for your email, phone number, or address.
• — Sell or broker leads to agents.
• — Accept payment for ranking position or placement.
• — Hide any part of a ranking behind a signup.