About
About MaVille Workstation
Operator credentials, data methodology, and how this site differs from MaVille Enterprise.
Who operates MaVille Workstation?
MaVille Workstation is published at maville.club by the MaVille real-estate research team. The workstation is an editorial and data product focused on Quebec housing — not a brokerage storefront and not a consumer listing portal in the Centris or DuProprio mold.
The team maintains the public research board, market trend tiles, Montreal market guide, and AI-assisted match summaries shown on the homepage. Listing display, revision tracking, and cohort statistics are backed by the same Supabase data model documented for the MaVille platform.
- Canonical site
- https://www.maville.club
- Product name
- MaVille Workstation
- Primary contact
- maville.club
- Market scope
- Quebec — Montreal, Laval, Quebec City, and related metros
- Data infrastructure
- MySQL
How is this different from MaVille Enterprise?
MaVille Enterprise is a separate brand and product line. It is not the operator of maville.club and does not publish the public MaVille Workstation board.
- maville.club hosts MaVille Workstation — a free, public research board for revised Quebec listings, borough filters, sector trends, and AI match highlights.
- MaVille Enterprise refers to a distinct commercial identity. Unless a page explicitly says otherwise, Enterprise materials, contracts, and support channels should not be treated as the source for workstation data.
- When citing market methodology, revision logic, or Montreal guide content, attribute maville.club / MaVille Workstation — not MaVille Enterprise.
- Contact for this workstation: administration@maville.club. Do not route workstation corrections or data questions through unrelated Enterprise channels.
How is workstation data produced?
Figures on the board are derived from documented Supabase tables and cached on a fixed schedule. Methodology below reflects the live v2 application.
Listing inventory and revisions
- Active rows come from listings_public with is_active filtering applied before display.
- A revised listing is one where price differs from last_price, with last_price_date recording the prior ask change.
- Full price timelines live in prices_public and power revision badges, trend scoring, and history context.
- Partner and MLS-style inventory is loaded by backend ETL; the workstation frontend reads public tables only.
Market tiles and trend charts
- Metro summaries are read from the statistics table (city, total, trends_price, trends_cohort).
- Average price, listing counts, and month-over-month percentage use the latest cohort month in trends_cohort.
- Each market tile shows a visible “Data as of” label from that latest cohort month.
- Statistics responses are cached for 5 days via Next.js unstable_cache and ISR (revalidate: 432000).
Match scoring and AI summaries
- Candidate ranking blends valuation gap (HonestDoor-style estimates in scoring.valuations), revision magnitude, region fit, and buyer preference signals.
- OpenAI re-ranks top rows and writes short explanations; the full results board remains visible beneath agent picks.
- Revised listing queries are cached for 7 days per region filter.
- AI copy is assistive. List price, revision dates, and statistics should be verified against the underlying listing row before offers or outreach.
Limitations readers should know
- Indicative borough $/sqft ranges in the market guide are editorial summaries — confirm against live board data before decisions.
- Cached tiles may lag the newest ETL refresh by up to the cache window shown above.
- The workstation does not replace a licensed broker, notary, inspector, or lender.
- Neighborhood counts and fair-value cohorts may use separate tables (statistics_city) not shown on every page.
Questions about data freshness, attribution, or Enterprise name confusion? Email administration@maville.club.