RateMyAgent Scraper — Real Estate Agent Leads (AU · NZ · US)

Turn RateMyAgent into a clean real-estate agent lead list. One row per agent: agentName, jobTitle, agencyName, suburb, state, postcode, rating, reviewCount, soldCount, forSaleCount, avgDaysOnMarket, avgSalePrice, medianSalePrice, totalSalesValue, specialties, phone, email, websiteUrl, socials, bio and up to 10 recent listings with addresses and sold prices.

Email and phone are each on ~99% of rows — email 1,354 of 1,366 agents (99.1%) and phone 1,356 of 1,366 (99.3%) across every validation run listed below, including a full 1,000-agent walk. Measured, and not rounded up to 100%: on the 1,000-agent run specifically, email was 98.8%. No login, no cookies, no CAPTCHA solving.

Scope: 178,789 agents — counted live off RateMyAgent's own sitemaps on 2026-08-16, not quoted from a marketing page:

{
  "regions": ["AU"],
  "maxItems": 50,
  "includeRecentSales": true,
  "includeBio": true,
  "maxConcurrency": 8
}

Click Try for free and hit Start — that block is literally the input this Actor ships with. It returns 50 Australian agents for $0.15 in under two minutes. Across the three separate runs of exactly that input recorded below, email was 100%, rating 100%, phone 98%, sold count 96–100% and suburb 94–96%.

Read this before you run it

Six things that would otherwise turn into a refund request. The first one is not a technical caveat — read it properly.

1. robots.txt — the actual lines, unedited

This Actor reads RateMyAgent's own API host, not the public HTML pages. Both files were fetched on 2026-08-16, and both are quoted here in full.

https://api.ratemyagent.com.au/robots.txt — HTTP 200, 25 bytes. That is the entire file:

User-agent: *
Disallow: /

Identical, byte for byte, on https://api.ratemyagent.com/robots.txt (US) and https://api.ratemyagent.co.nz/robots.txt (NZ) — all three 25 bytes, all three Disallow: /.

For contrast, https://www.ratemyagent.com.au/robots.txt — HTTP 200, 1,141 bytes — opens with:

user-agent: *
allow: /
disallow: /error/
disallow: /404
disallow: /500
disallow: /dashboard/
disallow: /promoter/
disallow: /account
disallow: /agency/
disallow: /agent/
disallow: /suburb/search/
disallow: /testimonial/
disallow: /arkadia

Sitemap: https://www.ratemyagent.com.au/sitemap.xml

So, stated plainly and without a justification wrapped around it: the agent records in this dataset come from api.ratemyagent.*, a host whose robots.txt disallows every path for every crawler. The only thing this Actor reads from the allow: / www host is the Sitemap: file that same robots.txt points at. It does not log in, does not use an API key, does not defeat any challenge and reads nothing behind authentication — but it is not robots-compliant on the API host, and RateMyAgent's terms of use prohibit automated collection regardless of technical access. Whether that is acceptable is your decision, in your jurisdiction, for your use case. Do not run this if the answer is no.

2. The public HTML profile pages are unreachable, and that is why the API is used

www.ratemyagent.*/real-estate-agent/… is behind DataDome tag v5.9.0. Measured through the Apify proxy: 0 of 60 requests returned HTTP 200 — 52 in the original probe across datacenter, residential, eight exit countries, four user-agents, eight host variants and three real Chromium modes (headless-shell, new-headless and fully headed with a warmed homepage session), plus a fresh 0 / 8 re-check on 2026-08-16. Every single one was HTTP 403 with a dd={…} body of 777–781 bytes. There is no HTML route to this data, and this Actor does not pretend there is.

3. Geography fields differ by country, because RateMyAgent stores them differently

RateMyAgent prints an agent's market areas as a single string, and the format is not the same in each directory:

New Zealand agents have no postcode and US agents have no suburb — the US directory keys markets on ZIP, and the town shows up in city instead. Those are source gaps, not scraping failures. primaryMarket always carries the raw string and marketAreas carries every one of them (a median of 8–10 per agent), so you can always re-parse yourself.

4. rating is 100% filled — including agents with zero reviews

Every profile carries a rating, and an agent nobody has reviewed carries 0. A minRating filter on its own therefore does not give you well-reviewed agents. Pair it with Minimum review count if that is what you mean.

5. Recent-listing prices are the agent's own free text

recentListings[].priceText is exactly what RateMyAgent printed — "$740,000 - $785,000", "Offers Over $1,049,000", "Auction", "Contact Agent", "Price Adjusted". priceMin / priceMax are that string parsed into numbers when it contains a real figure (≥ $10,000), and null when it does not. The reliable numeric price fields are medianSalePrice (94.6%), avgSalePrice (63.1%) and totalSalesValue (94.6%), which come straight from RateMyAgent's own sales statistics, not from text.

6. Everything except Regions is a post-fetch filter

Agent URLs carry no geography, rating or sales history, so state, suburb, rating, review count, sold count, email/phone and claimed-profile filters all run after a profile has been fetched. You are billed only for agents returned — a filtered-out profile costs run time, not money — but a rare suburb can cost hundreds of fetches per row kept.

What you get

One row per unique agent. scrapedAt is a full UTC timestamp; prices are plain numbers in the local currency (AUD / NZD / USD — see region).

The dataset ships with a saved table view — Agents — so you do not have to configure columns.

Field fill — measured on 149 agents

Every percentage below comes from one run on this Actor,

{"regions":["AU","NZ","US"], "maxItems":150}

Every *Url field is a real URL. RateMyAgent does not validate its own social fields, so a small share of agents typed a bare handle (uday_chandran), a bare host (compass.com) or a mistyped scheme (https//www.instagram.com/…, https:wwwinstagram.com.au/…) into one — 14 of the 1,917 social/website values across 1,060 delivered rows. All 14 are normalised to https://… before they reach your dataset, so nothing lands in your CRM as a dead link, and no value is ever dropped. Where the platform's path is ambiguous (LinkedIn /in/ vs /company/, YouTube /@ vs /channel/) a bare handle is passed through untouched rather than guessed into a wrong link.

email is likewise the agent's own free text. One agent in the same 1,048 addresses had typed two addresses into the single field (adam@… info@…), which fails a CRM import; the first valid address is returned rather than the raw pair.

Boolean flags are rates, not fills — the field is always present. On the same 149 rows: isClaimed 100% true, contactFormEnabled 100%, hasMlsConnection 60.4%, isPriceExpert 20.1%, isAwardWinner 6.0%, isHighlyRecommended 3.4%, isTrustedAgent 2.7%.

Every run prints its own measured fill in the log, so you can check any slice you pull against the table above rather than trusting it.

How to run it

Whole directories

{ "regions": ["AU", "NZ"], "maxItems": 5000 }

Reads each selected directory's sitemap-sales-agents.xml, queues its agents, and interleaves the regions round-robin — so ["AU","NZ","US"] with maxItems: 150 returns roughly 50 of each, not 150 Australians.

Paging across runs

{ "regions": ["AU"], "maxItems": 5000, "skip": 0 }
{ "regions": ["AU"], "maxItems": 5000, "skip": 5000 }

Sitemap order is stable between runs. With no filters set, a run consumes exactly maxItems candidates, so consecutive pages are disjoint — verified below. Advance skip by maxItems, not by the row count you received.

Specific agents

{
  "profileUrls": [
    "https://www.ratemyagent.com.au/real-estate-agent/troy-schultz-bz557/sales/overview",
    "https://www.ratemyagent.co.nz/real-estate-agent/sharon-larkins-bb682/sales/overview",
    "https://www.ratemyagent.com/real-estate-agent/elizabeth-ramirez-garza-b04l35/sales/overview",
    "bg295"
  ]
}

Refresh a list you already hold. Skips the sitemaps and overrides Regions and every filter. Each URL's own domain decides the directory; a bare agent code uses the first region selected. An entry that is not a RateMyAgent agent profile is skipped with a warning naming it and the rest are still scraped; only a list with nothing usable in it stops the run, and it stops cleanly with nothing charged.

Filtering to a lead list

{
  "regions": ["AU"],
  "states": ["NSW"],
  "minReviewCount": 5,
  "maxItems": 12
}

That exact input returned 12 NSW agents from 36 profiles fetched (run RFVX1oZYDPWlrKymv) — about 3 fetches per row kept, and the 18 discarded agents were not billed. Budget for that ratio: a geography filter multiplies run time, not price, so maxItems: 500 on the same filter is still $1.50 but will fetch roughly 1,500 profiles to find them.

• states — AU state codes (NSW, VIC, QLD, WA, SA, TAS, ACT, NT), NZ regions (Auckland, Otago, Canterbury, Wellington…), US state codes (FL, TX, CA…).
• suburbs — AU/NZ suburb names (Bondi, Takapuna) or US ZIPs (33126).
• minRating · minReviewCount · minSoldCount · withEmailOnly · withPhoneOnly · claimedOnly.

withEmailOnly and withPhoneOnly measured ~100% fill, so they discard almost nothing — they are there as a guarantee, not as a narrowing tool.

Lean, faster export

Set includeRecentSales: false and includeBio: false. Two API calls per agent instead of four, a much smaller dataset, and you keep every contact, reputation, geography and sales-statistics field. Price per agent is identical — this buys time, not money.

Scheduling and integrations

Save the input as a Task, then attach an Apify Schedule. The dataset is available over the REST API and through the standard Apify integrations (Zapier, Make, n8n, webhooks, MCP). Starting a run from the API:

curl -X POST "https://api.apify.com/v2/acts/scrapersdelight~ratemyagent-scraper/runs?token=YOUR_TOKEN" \
  -H 'Content-Type: application/json' \
  -d '{"regions":["AU"],"states":["VIC"],"maxItems":200}'

Sample row

A real row from an actual run, captured 2026-08-16 (bio truncated here for length, recentListings cut to 2 of 10).

{
  "agentName": "Adrian Jian-Sheng Wu",
  "jobTitle": "Sales Executive",
  "profileCode": "ce581",
  "profileUrl": "https://www.ratemyagent.com.au/real-estate-agent/adrian-jian-sheng-wu-ce581/sales/overview",
  "profileSlug": "adrian-jian-sheng-wu-ce581",
  "region": "AU",
  "country": "Australia",

  "agencyName": "Strathfield Partners",
  "agencyUrl": "https://www.ratemyagent.com.au/real-estate-agency/strathfield-partners-ao572/sales/overview",
  "officeName": "Strathfield Partners",
  "officeUrl": "https://www.ratemyagent.com.au/real-estate-agency/strathfield-partners-ao572/sales/overview",

  "phone": "0410 296 171",
  "email": "adrianw@strathfieldpartners.com.au",
  "contactFormEnabled": true,
  "websiteUrl": "https://www.strathfieldpartners.com.au/",
  "facebookUrl": "https://www.facebook.com/Adrian-Wu-491352231209294/",
  "instagramUrl": null, "linkedinUrl": null, "youtubeUrl": null, "twitterUrl": null, "tiktokUrl": null,

  "rating": 4.9,
  "reviewCount": 77,
  "isClaimed": true,
  "featureLevel": "Premium",
  "isAwardWinner": false, "isHighlyRecommended": false, "isPriceExpert": false, "isTrustedAgent": false,
  "hasMlsConnection": false,

  "suburb": "Homebush",
  "city": null,
  "state": "NSW",
  "postcode": "2140",
  "primaryMarket": "Homebush, NSW, 2140",
  "marketAreas": ["Homebush, NSW, 2140", "Strathfield, NSW, 2135", "Parramatta, NSW, 2150", "Newington, NSW, 2127", "Homebush West, NSW, 2140", "Campsie, NSW, 2194", "Carlingford, NSW, 2118", "Canterbury, NSW, 2193", "Lane Cove, NSW, 2066", "Hurstville, NSW, 2220"],
  "marketAreaCount": 10,

  "soldCount": 36,
  "forSaleCount": 16,
  "avgDaysOnMarket": 26,
  "avgSalePrice": 2080861.11,
  "medianSalePrice": 2080861.11,
  "totalSalesValue": 74911000,
  "specialties": ["Apartment", "House", "Land"],
  "propertyTypesSold": { "Apartment": 26, "House": 9, "Land": 1 },
  "medianPriceByPropertyType": { "Apartment": 874461.54, "House": 5569444.44, "Land": 2050000 },

  "avatarUrl": "https://cdn.ratemyagent.com.au/photo/2023-02-17-0321-1361-183289.png",
  "coverImageUrl": "https://cdn.ratemyagent.com.au/cover/2024-10-21-0923-2263-937689.png",
  "scrapedAt": "2026-08-16T18:54:31.238Z",
  "bio": "Committed to providing six star service to his clients, Adrian has a clear forward thinking vision for the future…",

  "activeListingCount": 16,
  "completedSaleCount": 36,
  "recentSoldPriceMin": 630000,
  "recentSoldPriceMax": 4500000,
  "recentListings": [
    {
      "status": "Active", "propertyType": "Apartment",
      "priceText": "Auction", "priceMin": null, "priceMax": null,
      "address": "56/22 Buchanan St", "suburb": "Balmain", "state": "NSW", "postcode": "2041",
      "bedrooms": 2, "bathrooms": 2, "carparks": 1, "resultDate": null,
      "listingUrl": "https://www.ratemyagent.com.au/real-estate-agency/strathfield-partners/property-listings/56-22-buchanan-st-balmain-ak2h05"
    },
    {
      "status": "Active", "propertyType": "Apartment",
      "priceText": "Just Listed", "priceMin": null, "priceMax": null,
      "address": "214/39 Cooper St", "suburb": "Strathfield", "state": "NSW", "postcode": "2135",
      "bedrooms": 2, "bathrooms": 2, "carparks": 1, "resultDate": null,
      "listingUrl": "https://www.ratemyagent.com.au/real-estate-agency/strathfield-partners/property-listings/214-39-cooper-st-strathfield-ak1mr2"
    }
  ]
}

Fields people misread:

• medianSalePrice and avgSalePrice are whole-career figures RateMyAgent publishes, not a 12-month median, and they can be equal on agents with few disclosed sales.
• specialties is derived from what the agent actually sold, ordered by count — it is not a self-declared badge. propertyTypesSold is the same data as raw counts.
• soldCount is RateMyAgent's Sold statistic; completedSaleCount is how many completed campaigns their listings feed carries. They usually match and occasionally do not.
• agencyName falls back to the office name when RateMyAgent files an agent under an office with no parent agency (common in the US directory).

Input

Pricing

$0.003 per agent returned — $3.00 per 1,000. Pay-per-event, one event (agent-scraped), nothing else.

• You are charged for rows delivered. An agent fetched and then removed by your filters is never charged. An agent that is queued twice is charged once — duplicates are dropped before billing.
• Rows are charged as they are pushed (Actor.pushData(items, 'agent-scraped')), so if you hit a budget cap you get whole rows and stop, never a half-billed dataset.
• maxItems is your hard cost cap. It counts unique, delivered agents.
• Turning includeRecentSales off halves the request count and costs exactly the same.

Honest limits

• The API host's robots.txt says Disallow: /. Quoted verbatim at the top of this page. Read it before you run anything.
• No licence numbers, and no review text. RateMyAgent publishes agent reviews on the site, but a review is a different record from an agent and merging them would give you a dirty dataset. This Actor returns agents. reviewCount and rating are the summary.
• NZ agents have no postcode; US agents have no suburb. Source gaps, documented in gotcha 3 with the per-country numbers.
• avgSalePrice is 63.1% and bio 59.1%. Those are optional fields RateMyAgent fills in or does not; nothing here can conjure them.
• email is ~99%, not 100%. It was 100% on every sample up to 149 agents and 98.8% on the 1,000-agent run. Use withEmailOnly if a row without one is useless to you.
• The sales record is RateMyAgent's, not the MLS's. It reflects campaigns RateMyAgent knows about. hasMlsConnection (60.4% across AU+NZ+US; AU 2%, NZ 80%, US 100%) tells you whether an MLS feed is attached.
• This is AU/NZ/US only. RateMyAgent runs three directories and there is no fourth.
• No login, no cookies, no CAPTCHA solving service, no browser automation.

How it works, and what it cost to make reliable

Enumeration reads the sitemaps the www robots.txt itself points at (/sitemap.xml → /sitemap-sales-agents.xml → shards). Extraction reads RateMyAgent's own JSON backend — up to four calls per agent (/Details, /Markets, /Stats, /Campaigns) — ordered so the cheapest filters run first: an agent removed by a rating or email filter costs one request, not four.

Transport ladder, re-measured through the Apify proxy on 2026-08-16. Each rung is single-shot against real agent codes with a fresh, hyphen-free session per call — no retries, so these are raw first-attempt rates:

Datacenter drops roughly one call in ten, and a dropped call on this Actor means a missing agent rather than a retry you can see, so RESIDENTIAL with the exit country matched to the region is the default. Misses are absorbed by a four-attempt ladder that mints a brand-new session each time rather than dropping to a cheaper rung. The one exception is the sitemap host, which is not walled and is read over the cheap datacenter rung — a shard is ~2.2 MB and reading it over residential would burn the run's proxy budget for nothing.

Sustained load, this Actor end-to-end on the Apify platform (2026-08-16). Every row is a real run you can look up by id:

One unreachable profile in 1,391 is the honest number. It is not 100%, and the Actor is built around that: a miss costs you a row, never the run.

Behaviour at the run time limit. Three deliberately unbounded runs ({"regions":["AU","NZ","US"],"maxItems":100000}) capped at a 120-second run timeout returned 34, 36 and 28 agents, each finishing Succeeded with a status message saying it stopped early — never timed out, and never discarded rows it had already fetched. The Actor reads the platform's own deadline, stops starting new work before it, and pushes what it has.

Duplicates — measured

The sitemap lists an agent under three URLs (/overview, /properties, /reviews) — 72,144 AU URLs collapse to 38,170 agents — so duplicates are dropped at enumeration, before a single request is spent. A run-wide set keyed on region:profileCode then drops anything that slips through before it is pushed or billed, and every run ends by checking that the number of rows equals the number of distinct codes and printing the result in the log. Across 1,366 delivered rows that check has never diverged.

Across runs, use skip — dedupe is per-run by design, and with no filters set the page boundaries are exact (verified above: zero overlap between skip: 0 and skip: 50).

When something goes wrong

This Actor is built not to end a run in the Failed state. A failed run tells you nothing useful, loses the rows already fetched, and counts against the Actor's health on the Store — so every problem below instead finishes Succeeded, keeps whatever it delivered, and puts the reason in the run's status message. Nothing partial is ever billed: you are charged per row delivered, so a run that delivers nothing charges nothing.

• 0 rows because everything was filtered out → succeeds, warns, charges nothing. A legitimately empty result is not a broken one.
• 0 rows because the profiles were unreachable → succeeds with a status message naming the counts (fetched, unreachable after retries, retired, filtered out). Re-run: the residential pool was having a bad minute.
• Hit the run time limit → stops starting new work, pushes what it has, says so in the status message. Raise the run timeout or lower Max results.
• Hit your Max total charge → stops at a whole row boundary and tells you. Never a half-billed dataset.
• Sitemap host down, or its layout changed → that region is skipped with a warning; the other regions still run.
• Bad entries in Agent profile URLs → skipped individually with a warning, naming the first one. The valid entries still run.
• A high block rate → warns with the measured percentage; the rows you got are real.
• Fewer rows than requested because the directory ran out is a warning, not an error.

Who buys this

• PropTech and real-estate SaaS (CRM, listing tools, digital marketing) selling to agents in AU, NZ and the US — soldCount, medianSalePrice and featureLevel rank the list by who can actually pay.
• Recruiters and franchise groups poaching high-performing agents: rating, review count, sold count and current agency, with an email on every row.
• Photography, staging, styling and signage suppliers targeting agents by suburb and by the property types they actually sell.
• Mortgage brokers and conveyancers building referral networks in specific postcodes.
• Market analysts — medianPriceByPropertyType and avgDaysOnMarket per agent per suburb is a view of a market that a listings scrape does not give you.

Our other real-estate agent Actors

That family is US-only. RateMyAgent is the one that covers Australia and New Zealand — 47,155 AU + NZ agents that none of the others can reach — and it carries the review-based reputation signal (rating, reviewCount, isHighlyRecommended) that the US listing portals do not publish. Buy it alongside them, not instead of them.

FAQ

Does this need an account, a login or cookies? No. No login, no cookie jar, no API key, no CAPTCHA solving service, no browser automation.

Do I really get email addresses? Yes — 1,354 of 1,366 agents across every validation run (99.1%), and they are real mailboxes, not contact-form links. It was 100% on the 149-agent three-country run and 98.8% on the 1,000-agent run, so plan on ~99% rather than all of them.

Why doesn't it scrape the normal RateMyAgent pages? Because they cannot be scraped. The public profile pages are behind DataDome v5.9.0 and returned HTTP 403 on every attempt across every proxy, country, user-agent and browser mode tried — including a fresh 0/8 re-check on 2026-08-16, each one a 781-byte dd={…} body. See gotcha 2.

Is it robots-compliant? No, and this page says so plainly. The API host's robots.txt is 25 bytes and reads User-agent: * / Disallow: /. It is quoted in full at the top. Read it and make your own call.

Can I get every agent in Australia? Yes — {"regions":["AU"],"maxItems":40000} queues all 38,170 for $114.51. Give it a run timeout to match: 1,000 agents took 29 minutes, so budget several hours, or page it with skip across several shorter runs.

Two runs — will I get duplicates? Within a run, never: dedupe is on region:profileCode, and every run checks and logs that rows equal distinct codes. Across runs with no filters set, skip: N after maxItems: N is exact — verified at 0.0% overlap between skip: 0 and skip: 50.

Do I get charged for rows a filter removed? No. Filtered-out agents are never pushed and never billed.

Will a run ever succeed with zero rows? Yes, in a few honest cases — filters that matched nothing, a run time limit hit before the first agent, or a bad minute on the residential pool. Each one says which in the run's status message, and each one charges nothing. See "When something goes wrong".

Do I need a residential proxy? Yes — it is the default and you should leave it alone. Re-measured 2026-08-16, single-shot: datacenter 18/20 (90.0%), RESIDENTIAL with a country-matched exit 20/20 (100%). One dropped call in ten is a missing agent, which is why the cheap rung is only used for the sitemaps.

Legal & fair use

This Actor reads RateMyAgent's public sitemaps and its own JSON backend. It does not log in, does not use an API key, does not defeat any anti-bot challenge, and collects nothing behind authentication.

The API host it reads publishes a robots.txt of User-agent: * / Disallow: / — quoted verbatim near the top of this page. This Actor is therefore not robots-compliant on that host, and RateMyAgent's terms of use prohibit automated collection regardless of technical access.

Records contain personal data — agent names, direct mobile numbers and personal email addresses. You are responsible for complying with RateMyAgent's terms of service and with how you use the data, including the Australian Privacy Act and the Spam Act, the New Zealand Privacy Act and Unsolicited Electronic Messages Act, and US CAN-SPAM and state privacy law.

RateMyAgent® is a trademark of its owner. This Actor is not affiliated with, endorsed by, or connected to RateMyAgent.

Feedback

Found a missing field or want a new filter? Open an issue on the Issues tab, and if the Actor earns it, a review on the Reviews tab helps other buyers find it.