You’re browsing publicly. Statutes and search stay open. A free account adds bookmarks, document upload, side-by-side compare, and forum posting.

Industry data

Methodology

Exactly how each figure on the industry data page is calculated: the source table, the SQL logic, inclusion thresholds, and refresh cadence. No AI interpretation is applied at any step in this pipeline. Every number is a direct aggregate of public-record data.

Pipeline overview

Four of the five metrics on /insights (association counts, SIRS exposure, FHA approval, property values) read from a single Postgres materialized view, association_report_card_metrics. That view joins the organizations table to per-association public-record facts: parcel-derived property metrics, elevator certificates, pool inspections, SIRS filings, HUD FHA project status, building permits, FEMA county risk, and IRS Form 990 filings. The fifth metric, insurance licensee stats, reads from a separate table, insurance_licensees, populated directly from state department-of-insurance bulk downloads.

The five insights_* objects the page queries (defined in migration 20260626140000_insights_aggregate_views.sql) are plain SQL views, not materialized views: they push a GROUP BY into Postgres and recompute on every read. That means the aggregation logic itself runs live against whatever is currently in the underlying tables. The population those views aggregate over is only as fresh as the last refresh of association_report_card_metrics (see the note under each metric below) or the last write to insurance_licensees.

Every metric excludes user-created stub organizations: the base population is restricted to org_type = 'association' with a non-null directory_source that is not 'user_created'. If a state or subtype cohort doesn't appear on the page, it either has zero rows meeting that filter or falls below a metric's minimum group-size threshold (documented per metric below).

Associations by state and type

insights_association_counts

Calculation: count(*) from association_report_card_metrics, grouped by state and subtype. Rows with a null state or subtype are excluded before grouping.
Population: Every organization with org_type = 'association' that has a non-null directory_source other than 'user_created': every directory-linked association Common Elements has matched to a state registration or public record, not associations a user manually added to the platform.
Displayed subset: The page requests the top 40 state-subtype cohorts ordered by count descending, then renders the top 15 in the table.
Source: State Secretary of State / corporation registries, county recorder filings, and state-specific HOA/condo registries (coverage varies by state, see the per-state ingestion jobs in apps/web/vercel.json), matched into the organizations table.
Refresh: Recomputed on every page load, but only over the current snapshot of association_report_card_metrics. That materialized view is refreshed by the service-role-only function public.refresh_association_report_card(); the codebase does not register a fixed recurring schedule for calling it, so treat the count as updated as new data is ingested and the view is refreshed, not on a fixed clock.

SIRS structural-inspection exposure

insights_sirs_exposure

Calculation: For each state+subtype group: round(100.0 * count(*) FILTER (WHERE has_sirs IS TRUE) / count(*), 1). has_sirs is a boolean computed in association_report_card_metrics as true if at least one row for that organization exists in fl_sirs_projects, and false otherwise. It records whether an association has a Structural Integrity Reserve Study on file at all. It does not distinguish a study that found the structure compliant from one that flagged deficiencies, and it does not confirm the study is current.
Inclusion threshold: Groups with fewer than 10 associations are dropped (HAVING count(*) >= 10).
Displayed subset: The industry data page further filters the table to Florida rows with a non-zero rate, since Florida is the only state SIRS data currently covers. The underlying view itself is not FL-scoped: if SIRS data for another state were ingested, it would appear in the view without a migration change.
Applicability: Florida's SIRS requirement (SB 4-D, the post-Surfside milestone inspection law) legally applies only to condominium and cooperative buildings three stories or taller. The denominator in this metric is every association in the cohort regardless of height or building type, so the exposure rate understates compliance among buildings that are actually subject to the law and overstates it as a share of all associations.
Source: FL DBPR SIRS reporting (SB 4-D), ingested into fl_sirs_projects.
Refresh: SIRS records are pulled by a standalone ingestion script, not a registered recurring cron in this codebase. Combined with the materialized-view refresh dependency described above, treat this rate as updated as new SIRS data is ingested, not on a fixed schedule.

Median assessed value per unit

insights_property_values

Calculation: For each state, subtype, and unit-count band: percentile_cont(0.5) WITHIN GROUP (ORDER BY just_value_per_unit), i.e. the true statistical median (linear interpolation between the two middle values), not an average. avg_jvpu (the mean) is also computed but not shown on the industry data page.
What just_value_per_unit means: Computed once per association in association_property_metrics, from every parcel matched to that association in association_parcels: sum(just_value) / sum(num_res_units)across the association's parcels, where "just value" is the county property appraiser's statutory market-value estimate (not sale price, not assessed/taxable value). If a parcel roll doesn't report residential unit counts, the calculation falls back to averaging just value across condo-unit parcels only. The median shown on the industry data page is the median of this per-association figure across every association in the cohort, not a median of individual parcel values.
Unit band: Associations are bucketed by unit count: <25, 25-49, 50-99, 100-199, 200-499, 500+, or unknownif unit count isn't on file.
Inclusion threshold: Groups with fewer than 10 associations reporting a value are dropped. Rows with a null state, subtype, or just_value_per_unit are excluded before grouping.
Displayed subset: The page requests the top 40 cohorts by median value descending, then renders the top 12.
Source: County property appraiser cadastral rolls (statewide parcel data, e.g. Florida's DOR NAL file), ingested into association_parcels and matched to associations by address/situs.
Refresh: Parcel rolls are pulled by standalone per-state ingestion scripts, not a registered recurring cron. Same materialized-view refresh dependency as above applies: updated as new parcel data is ingested and the view is refreshed.

Active FHA condo approval by state

insights_fha_approval

Calculation: For each state+subtype group, has_fha_data counts associations with any HUD FHA project status on file (fha_status IS NOT NULL). pct_fha_approved is round(100.0 * count(*) FILTER (WHERE fha_status = 'Approved') / has_fha_data, 1). The percentage is a share of associations WITH HUD data, not a share of the full cohort. fha_status in the underlying materialized view is the most recent status by status_date when an association has more than one HUD FHA project record.
Inclusion threshold: Groups with fewer than 10 associations that have HUD FHA data are dropped. The page additionally filters the displayed rows to groups with 10 or more (redundant with the view's own threshold) and shows only the top 10 by group size.
Source: HUD FHA-approved condominium project data, ingested into hud_fha_condo_projects and matched to associations.
Refresh: Pulled by a standalone ingestion script, not a registered recurring cron. Updated as new HUD data is ingested and the materialized view is refreshed.

Insurance producers by state

insights_insurance_licensees

Calculation: count(*) from insurance_licensees, grouped by state and entity_kind (individual, business, or unknown). active_count is the same count filtered to license_status = 'Active'.
What it does not filter on: This table is not scoped to community-association-relevant coverage lines. It is every licensed insurance producer record in the source state's bulk download, regardless of whether that producer writes HOA/condo master policies, auto, life, or any other line.
Coverage today: Every row currently on file carries license_state = 'FL' (the schema supports other states, but only Florida has been ingested as of this writing). Treat this metric as Florida-only until other states appear in the data.
Displayed subset: The page requests the top 30 groups by licensee count, then renders the top 12.
Source: FL Department of Financial Services bulk licensee download (licenseesearch.fldfs.com/BulkDownload, public record, no login required).
Refresh: Weekly, Mondays at 03:00 UTC (/api/cron/ingest-fl-dfs-licensees, registered in apps/web/vercel.json). Unlike the other four metrics, this table is queried directly. There is no materialized-view refresh step in between, so a successful weekly cron run is reflected on the next page load.

A note on the "Active FHA approval" summary card

The summary stat card reading "< 1%" on the industry data page is a fixed descriptive statement about the Florida condo cohort in insights_fha_approval, not a live-computed value like the other summary cards. To see the live per-state, per-subtype figure this describes, use the FHA table on the compliance tab, which reads pct_fha_approved directly from the view on every load.

What this measures, precisely: the share of condo projects that currently hold an activeapproval status on HUD's FHA-approved condominium list. It is a measure of approved stock, not an application approval rate. HUD's public data does not distinguish projects whose approval expired from projects that never applied, so no denial or application-rate statistic can be derived from it.

Data policy

Every figure on the industry data page is a direct aggregate (count, percentage, or median) over public-record data. No AI summarization, estimation, or interpretation is applied to any of the five metrics. The numbers are exactly what the SQL documented above returns. No per-association rows or personally identifiable information are exposed through these views; all five are grants to anon because they return only grouped, thresholded aggregates.

If a number on this page looks wrong or you need a definition not covered here, the source of truth is the SQL itself: the view definitions in supabase/migrations/20260626140000_insights_aggregate_views.sql and the materialized view in supabase/migrations/20260608011000_association_report_card.sql.

Back to industry data