Every builder shown here is scraped on a real schedule, region by region — their own website, plus a handful of public, terms-compliant off-site sources. Nothing is guessed or estimated; if a source can't be checked safely, it's recorded as a link for you to open yourself, not filled in with a guess.

Their own website — the main source

A real crawl of the builder's site: sitemap and navigation first, then every design/house page. From it we pull:

• Floor plans, facade images and renders — matched per design, not just the first image found on the page. During a Deep Dive these are downloaded as real files (marked 📐), not just linked — a link rots the moment the builder redesigns their site.
• Pricing — entry price, median price, $/m² where the site publishes it.
• Estates & display homes — which estates they build in, which have a display home.
• Promotions — current offers running on the site right now.
• Sales team & leadership — names, roles, and any individual LinkedIn profile link found on their "our team" page. A role matching Director/Founder/Owner/CEO/Managing gets a 👤 marker in the People list.
• Testimonials — real customer reviews as the builder themselves publishes them: name, buyer type, home size and the quote, from their own testimonials page. A different channel to ProductReview below — curated by the builder, but still first-party and attributable to a named buyer.
• Awards — their real award history, where they publish one: year, award name and awarding body (HIA, Master Builders, etc.).
• Social links — whichever of Facebook / Instagram / YouTube / LinkedIn the builder themselves links to (used to drive the checks below, not scraped directly for content).

Directors — two real sources, shown separately

"Directors — QBCC official register" is a real match against QBCC's own Key Personnel filings (directors/secretaries, with the date they were appointed) — the authoritative source, but only covers licensees already in that dataset. The 👤-marked entries in "People" are whoever the builder's OWN site names as a director/founder/owner, which can catch someone new the QBCC data hasn't caught up with yet, or miss someone the site simply doesn't publish. Neither is complete on its own, which is why both are shown rather than merged into one list.
Not built: ASIC's free company search only confirms a company exists (ACN, status, registered office) — actual officeholder names are behind their paid "Current Company Extract," so this pipeline can't pull them from ASIC for free without either paying per lookup or scraping a paywalled document, neither of which this does.

What's actually checked off-site, and how

Source	What we actually do
OpenLot	Real, independent check — fetches the builder's own OpenLot profile page: which estates it lists, and every house-and-land package currently advertised there with its real price (shown under "Where & how they sell"). This is the closest thing to "what's currently for sale" this app can reach without defeating real bot-protection — REA (Kasada) and Domain's search (Akamai) are both actively blocked; confirmed live, 25 Aug 2026.
Reviews	ProductReview.com.au — real rating and review count, shown as "Mentioned online". Google reviews specifically would need your own Google Places API key (real cost, real setup) — not built without that.
News	Google News' own public RSS feed — real, official, not a scrape of search results (which actively block automated access and change constantly).
Facebook	Only the public Meta Ad Library — what ads they're paying to run right now, and since when. We do not scrape their Facebook page or posts.
Instagram	Not scraped at all — against Instagram's terms. Recorded only as a link, if the builder's own site links to it.
YouTube	Real channel data (subscriber count, recent videos) — but only if the builder links their channel from their own site.
LinkedIn	Individual people only — a sales/team member's own profile link, as found on the builder's team page. The company itself isn't scraped.
SEEK	Real job ads currently listed — title, location, date, link. Narrower coverage than everything else here: only builders on the older Competitor Tracker watchlist are checked, not the whole region, so "not tracked" and "tracked, zero live ads" are shown as genuinely different things.
realestate.com.au / Domain / Homely / Landchecker	Not independently searched. Recorded as a link only if the builder's own website links to their profile there.
TikTok	Not used anywhere in this pipeline.

Where the data ends up on this page

• 📰 What changed — every real difference from the previous scrape run, ranked by significance.
• ⚖️ Compare — a sortable table across every builder: price, $/m², designs, storey mix, estates, displays, promotions, ads, sales staff, social platforms.
• 🏗️ Builders — one full card per builder.
• 💲 Pricing — the same builders, filtered to only those with a real published $/m².
• 🛍️ Where & how they sell — estates, displays, promotions, ad counts, and every channel/portal link found, plus which estates have more than one of these builders (direct competition).
• ★ Watchlist — builders you've starred, with a button to re-run the scraper on just them.

Who is in here (changed 4 Sep 2026)

A region sweep now targets real home builders, not every licensee with a website: QBCC licence class Builder Open / Low Rise / Medium Rise, financial Category 1 or above (so no Self-Certification tradies), and at least 10 residential jobs notified to QBCC across the last three financial years plus this one — which is what removes civil and commercial contractors whose class says "Builder" but whose home-warranty history is empty. Brisbane goes from 1,952 targets to about 450. The 🏠 Home builders only chip applies the same rule to regions scraped before the change; switch it off to see everyone that was scraped.

Every builder card also carries the QBCC register's own numbers — rank in region, jobs per financial year, region share, capture rate against ABS approvals, home-warranty claims per 100 builds — copied from the same builder-stats file the Builders tab uses, dated on the card. The register records jobs per financial year, not a rolling 12 months, so "jobs" is always the last complete FY plus this FY so far.

Every run is dated and archived — nothing here overwrites history, so "What changed" always has a real previous state to compare against.