US Unclaimed Property & Funds Scraper

Search unclaimed money and uncashed checks across 12 public US registers in one run. The actor reads statewide bulk files from California and Iowa plus municipal open-data registers, then returns owner, amount, holder, property type, dates and property IDs where the source discloses them.

What it does

• Three modes: search by owner name, browse a register, and byPropertyId for an exact property, warrant or check lookup.
• 12 sources in 7 states: California, Iowa, Massachusetts, Maryland, Minnesota, Rhode Island and Texas.
• Up to 50,000 records per run — the old 1,000-row ceiling is gone, and every source is paginated or streamed to its own limit.
• Useful filters: amount range, include/exclude records with no disclosed amount, date range, broad property-type groups, raw type text and property ID.
• Seven sort orders: amount, date, owner name and source order.
• Empty fields are omitted — the actor never invents a value for a field a source does not publish.

Services covered

State	Source	What it publishes
CA	California State Controller (SCO)	Statewide unclaimed property, weekly ZIP/CSV archive
CA	Marin County	Uncashed county checks
CA	City of Roseville	Stale-dated municipal checks
CA	Sonoma County	Stale-dated county checks
IA	Iowa Department of Administrative Services	Outdated state warrants by fiscal year
MA	Town of Framingham	Unclaimed payroll checks
MA	Town of Framingham	Unclaimed resident/vendor/AP checks
MD	Montgomery County	Unclaimed vendor payments
MN	Ramsey County	Unclaimed money
RI	City of Providence	Uncashed municipal checks
TX	City of Austin	Unclaimed property
TX	City of Corpus Christi	Unclaimed property

How it works

Municipal registers are queried through their public Socrata (SODA) JSON APIs with $limit, $offset, deterministic $order and server-side filters. California publishes four value-range ZIP archives (plus an all-records archive the actor does not download); the actor streams the selected range file to a temporary file with HTTP Range resume, scans the CSV parts line by line, and deletes the file. Iowa publishes one XLSX per fiscal year, read in openpyxl's read-only mode and deleted after the scan. All sources are anonymous: no account, cookie, API key or proxy setup is required.

The California statewide file has no date field. When a date filter is set, California statewide and the other date-less registers are skipped and the status message says so. The default California range is $500 and above (about 163 MB); choosing all value ranges scans four files one at a time, which takes much longer. A property-ID lookup always checks every California range, smallest archive first, and stops once the bucket containing the ID has been scanned, so an ID below $500 is never hidden by the default.

Input

Field	Type	Default	Notes
mode	select	search	search, browse or byPropertyId
state	select	MN	ALL, CA, IA, MA, MD, MN, RI, TX
jurisdiction	select	ALL	Narrow to one of the 12 sources
ownerName	text	—	Case-insensitive; word order and commas do not matter
propertyId	text	—	Exact property ID, warrant or check number
minAmountUsd	number	0	Amount floor; 0 means no floor
maxAmountUsd	number	—	Amount ceiling
includeUnknownAmount	boolean	false	Keep records with no amount when an amount filter is active
dateFrom	text	—	Earliest date, YYYY-MM-DD
dateTo	text	—	Latest date, inclusive, YYYY-MM-DD
propertyType	select	ALL	10 broad groups over raw NAUPA codes and source categories
propertyTypeContains	text	—	Raw type substring filter, e.g. Dividend
sortBy	select	amountDesc	Amount, date, owner name or source order
maxItems	integer	20	1–50,000
valueRange	select	500plus	California archive range; ignored for property-ID lookups, which scan all ranges
iaFiscalYear	select	latest	Iowa fiscal year, all, or a specific year

Examples

Search one name across every state:

{"mode": "search", "state": "ALL", "ownerName": "Smith", "maxItems": 50}

Browse California's largest balances:

{"mode": "browse", "state": "CA", "jurisdiction": "ca_sco", "minAmountUsd": 1000, "maxItems": 100}

Look up a California property ID:

{"mode": "byPropertyId", "state": "CA", "propertyId": "13755891"}

Browse Iowa warrants from FY2026 with dates:

{"mode": "browse", "state": "IA", "iaFiscalYear": "2026", "dateFrom": "2025-07-01", "sortBy": "dateDesc", "maxItems": 100}

Output

Every record contains the fields the source actually discloses. Existing fields are unchanged; the revamp adds richer fields beside them.

Field	Description
ownerName	Owner/payee name
amountUsd	Current cash balance or warrant/check amount
reportedDate	Check, warrant or payment date (YYYY-MM-DD)
address, city, postalCode	Owner last-known address, where disclosed
ownerState, ownerCountry	Owner state/country, where disclosed (California)
recordId	Source record/check/warrant ID
propertyId	California property ID
checkNumber, warrantNumber	Source check or warrant number
departmentCode	Issuing Iowa department
vendorId, invoiceDate, paymentStatus	Montgomery County vendor-payment fields
holderName	Entity holding the funds
holderAddress, holderCity, holderState, holderZip	Holder location, where disclosed (California)
propertyType	Raw source category or NAUPA property-type string
cashReported, sharesReported	Reported cash and share counts (California)
securitiesName, cusip	Security description and CUSIP (California)
numberOfOwners, pendingClaims, paidClaims	Ownership and claim counts (California)
valueRange	California archive range the row came from
state, jurisdiction	Two-letter state and source name
sourceKey, sourceName, sourceType	Source traceability
sourceUpdatedAt	Source file date (California/Iowa)
sourceUrl, claimUrl	Public dataset page and official claim/search page
recordType	Always unclaimedProperty
scrapedAt	UTC scrape timestamp

Usage cost

The actor runs on Apify's standard compute pricing. It uses no paid proxy and no external API key, so the only cost is the run's memory and compute time plus Apify's usual dataset/result charges. The default 2 GB memory is sized for the California ZIP streaming path. Start with a small maxItems and the default $500+ range for browsing; a full California all scan is intentionally expensive in time because it reads four multi-hundred-megabyte archives.

FAQ

Does this search every US state? No. It covers the 7 states listed above. Most other states publish their live search only through CAPTCHA- or WAF-gated portals, and this actor does not attempt to bypass those. The official state claim portals remain the right place to file a claim.

Why are some records missing an amount? A few registers publish names without a dollar value. When an amount filter is active, those records are excluded unless includeUnknownAmount is enabled.

Why does a date filter drop California results? The California statewide file has 25 columns but no date field. Date filters therefore skip it rather than guessing a date.

What is the difference between search and byPropertyId? search treats the name as words in any order. byPropertyId performs an exact match against property IDs, warrants and check numbers. Both accept the other field as an additional filter. A California property-ID lookup scans all four value-range archives because an ID lives in only one of them; this can take several minutes.

How fresh is the data? Municipal Socrata registers refresh on their own schedules, California posts new archives weekly (usually Thursday), and Iowa adds fiscal-year files as warrants are cancelled. The actor always reads the currently published source.

Can I get more than 5,000 rows from multiple sources at once? Yes, up to 50,000, but a multi-source run keeps at most 5,000 candidates per source so memory stays bounded. To take a single source deeper, select that jurisdiction alone.

How do I claim money I find? This actor only reports public records. Contact the jurisdiction/holderName office or use the claimUrl on the record.

Are the addresses and names personal data? Yes — they are public-record claimant data. Use them only with a lawful basis, and do not combine them with other datasets to re-identify people.

Notes

• Public-record data only: owner names and last-known addresses as published by government offices. No SSN, phone, email or claimant identity is collected.
• No CAPTCHA solving, no login, no credentials, and no proxy configuration are required or supported.
• California and Iowa archives are streamed and deleted per run; rows are never materialised in memory all at once.
• Duplicate rows are removed by source + record identity, while legitimate co-owners on one property ID are preserved.
• Invalid dates and inverted ranges produce a clear status message and a clean zero-record run rather than a crash.

Limitations

• 7 states, not 50. MissingMoney.com and most state live-search portals are behind CloudFront/WAF/Turnstile blocks. The actor links users to the official portals instead of bypassing them.
• No date for California statewide records. The source does not publish one.
• California property-ID lookups are slower. They must check every value-range archive (smallest first) and stop only after the matching bucket, because the default $500+ range would otherwise hide lower balances.
• The California all-records archive is large. The default $500+ file is about 163 MB compressed; the full all-records ZIP is 3.25 GB, which is why the actor scans the four smaller range files instead.
• Some municipal datasets change schema. Column names are verified against the live APIs; a renamed column causes that source to be skipped with a status note rather than returning wrong data.
• Amounts can be stale. A published balance is the holder's last report, not a live claim valuation.