Research study
How many ADUs are there in America? The honest answer
Nobody knows. That is not a dodge — it is a finding with a mechanism, and this page publishes the mechanism, the closest defensible estimate, and the error bars that estimate’s own authors printed.
Written by HyreADU Research Desk Primary-source research and data analysis
Audited by HyreADU Research Desk Evidence-ladder and retrieval-date audit
The finding
No federal statistical agency publishes a count of accessory dwelling units in the United States, and none of the federal instruments that would produce one collects the category. The Census Building Permits Survey classifies buildings into four structure types and accessory is not among them.
The words “accessory”, “accessory dwelling”, “ADU”, “granny” and “secondary unit” do not appear anywhere in the 2023 American Housing Survey definitions document. The American Community Survey’s units-in-structure table inherits the same definitions.
The only national estimate with a published method is Freddie Mac’s, and it does not measure the housing stock. In July 2020 Freddie Mac text-mined a licensed CoreLogic dataset of roughly 600 million MLS transactions and reported that “[a] total of 1.4 million distinct single-family properties with accessory dwellings were identified in our final population.” That is a count of properties whose listing text matched ADU key-phrases at some point since the 1990s — not a count of accessory units standing today, and Freddie Mac names error directions that run both ways.
The one genuinely hard number is a permitted flow in one state. California jurisdictions reported 178,495 ADU permit units and 102,356 ADU completions to HCD for 2018–2025, deduplicated by tracking identifier — a floor for California over eight years, and the strongest ADU count in the country because California is the only state that collects one.
So the defensible statement is bounded and awkward, and we publish it in that form. Anyone publishing a single round national number should be asked which instrument it came from.
HyreADU makes no national claim from its own store, which is 735 CSLB-verified companies and five California permit jurisdictions and appears nowhere in this analysis.
Read this first
- No national number appears on this page, on purpose
If you need one figure for a slide, the least-wrong thing to cite is Freddie Mac’s 1.4 million properties with the sentence “properties identified from MLS listing text since the late 1990s, not a stock count” attached to it.
Detaching that sentence turns a careful estimate into a fabrication, and it is the single most common way this number gets misused.
- A property is not a unit and a flow is not a stock
Three different quantities appear on this page and they are not interchangeable. Freddie Mac counts properties ever associated with an accessory unit in listing text.
HCD counts permit events and completions — a flow, over a defined window. A stock count would be the number of accessory units standing on a given date, and nobody publishes one.
- We did not build our own national estimate, and we will not
We could take the California permitted flow, divide by California’s share of national housing units and multiply up.
That would produce a plausible-looking number and it would be worthless, because it assumes the rest of the country permits ADUs at California’s rate under California’s statutes. It does not. Where we have not measured something, we say so.
- Unpermitted units are a separate, larger unknown
Everything permit-based on this page excludes units built without a permit, and Freddie Mac notes explicitly that “illegal units — also called ‘shadow housing’ — will not be represented in” permit data.
Prior field research put the informal share very high in specific California neighborhoods. We treat that question separately in how many unpermitted ADUs are there.
Why the number does not exist
The reason is not that nobody has got round to it. It is that the American housing statistical system has no field for the concept.
Accessory dwelling units became a policy category in the 2010s; the instruments that count American housing were designed around a different question — how many units are in this building — and that question does not distinguish a subordinate unit from a primary one.
The consequence is very specific. A detached backyard ADU is, in the Census structure definition, “detached if it has open space on all four sides” — which is to say, it is its own single-unit detached structure, identical in the data to the house in front of it.
An internal or attached ADU makes the host property a structure with two or more units — identical in the data to a duplex.
There is no third state in the classification for “a smaller unit accessory to a larger one on the same lot.”
That is why our companion study, why the national permit data cannot see ADUs, treats this as an instrument problem rather than a data-availability problem. You cannot query your way out of a missing category.
HyreADU analysis: the practical effect is that the national ADU numbers in circulation are almost all downstream of one of three things — Freddie Mac’s text-mining study, a state or city series being described as if it were national, or a chain of citation with no origin at all.
The third is the most common. When we could not trace a number to a document, we left it out of this page rather than repeating it.
The evidence ladder: what each source can and cannot bound
Read the third column carefully. Several of these sources are excellent instruments that simply do not answer this question, and one of them answers a question that is easy to mistake for it.
| Source | What it publishes | What it can bound | Direction of error | Retrieved |
|---|---|---|---|---|
| Freddie Mac text-mining study (July 2020) | 1.4 million distinct single-family properties with accessory dwellings, identified from about 600 million MLS transactions since the late 1990s. Also: ADU share of active for-sale listings rising from 1.6% (2000) to 6.8% (2019). | The number of properties that have appeared on the MLS with accessory-unit language. Nothing about units that never listed. | Both ways, unquantified. Freddie Mac names understatement (owners who hold, off-MLS transactions, non-matching descriptions) and overstatement (older units listed for the first time), plus no adjustment for demolition or permanent loss. | 2026-09-05 |
| California HCD Annual Progress Report | 178,495 ADU permit units and 102,356 completions, 2018–2025, deduplicated, across 511 reporting jurisdictions. | A firm lower bound on permitted ADU production in California over 2018–2025. | Undercount only, within its scope: it misses unpermitted units, units permitted before 2018, and jurisdictions that failed to report. Four jurisdictions were unusable in our extract. | 2026-09-05 |
| Census Building Permits Survey | Permits by four structure categories at about 19,900 permit-issuing places. | Nothing about ADUs. There is no accessory item, conversions carry no unit count, and detached ADUs are merged into single-family. | Not applicable — the quantity is absent, not mismeasured. | 2026-09-05 |
| American Housing Survey (2023) | Detailed characteristics of a national housing sample. | Nothing about ADUs. None of “accessory”, “accessory dwelling”, “ADU”, “granny” or “secondary unit” appears in the published definitions document. | Not applicable. | 2026-09-05 |
| ACS table B25024, units in structure | Units in structure for all housing units, to tract level. | Nothing about ADUs directly. It inherits the AHS structure definitions, so detached ADUs are 1-unit detached and internal ADUs are 2-unit. | Not applicable. | 2026-09-05 |
| City ADU annual reports | Local permit counts. Seattle’s report gives 7,029 ADUs permitted citywide 1994–2024 (4,062 attached, 2,967 detached). | A firm count for that city over that window. | Undercount of the city’s stock (unpermitted units excluded); says nothing about anywhere else. | 2026-09-05 |
HyreADU note: the second column is the honest one. If somebody hands you a national ADU figure, the useful question is not “does that sound right” but “which row of this table did it come from, and did they carry the fourth column with it.”
The Freddie Mac estimate, in Freddie Mac’s own terms
The Freddie Mac figure is the only national ADU estimate with a documented, reproducible method, and it deserves to be described accurately rather than either dismissed or rounded off into something it is not.
What they did
Freddie Mac applied text mining to the unstructured description field of MLS listings in a CoreLogic-licensed dataset described as “a national-level dataset of 600 million MLS transactions dating back to the late 1990s,” covering both for-sale and for-rent listings, and matched descriptions against accessory-dwelling key-phrases — the granny flats, garage apartments and in-law suites of the paper’s title.
The headline output, verbatim: “A total of 1.4 million distinct single-family properties with accessory dwellings were identified in our final population.” Half of those 1.4 million properties are in four states: California, Florida, Texas and Georgia.
What the method can see
It can see a property that transacted, or was offered, through the MLS, and whose agent chose to describe the accessory unit in words the classifier recognized.
That is a real and useful population — and, importantly, a population no permit dataset can reach, because it includes units that were never permitted but were later advertised.
It is also a growth series, and a good one. The ADU share of active for-sale listings rose from 1.6 per cent in 2000 to 6.8 per cent in 2019; the share of for-rent listings from 1.8 per cent in 2003 to 4.1 per cent in 2019.
First-time ADU listings grew on average 8.6 per cent a year between 2009 and 2019.
What the method cannot see — in their words
“Understatement bias will occur in the following events: owners hold on to their units; rental or sale transactions occur outside of MLS—includes pocket listings, for-sale-by-owner, online rental sites, and/or referrals; or listing descriptions do not match ADU key-phrases.”
“Overstatement bias will occur when existing, older units are listed for the first time on MLS.” And: “A source of error in our estimates of accessory housing stock is we do not account for the demolition or permanent loss of older units.”
HyreADU analysis: those three sentences are why the 1.4 million cannot be treated as a stock estimate. The first pushes the true stock above it; the second and third push the identified population above the true current stock.
Without a magnitude on either, the figure is not a bound in either direction — which is exactly what Freddie Mac says, and exactly what almost every subsequent citation of it drops.
Why it is still the best national thing available
Because it names its data, names its method, names its biases, and can be argued with. Every other national figure we could trace either lacks one of those or lacks all four.
It also lands somewhere useful qualitatively. A method that only sees listed properties still finds 1.4 million of them, which is a strong indication that the national stock of accessory units is large — larger than the permitted flow of the one state that counts them. The honest form of the sentence is comparative, not numeric.
The one hard number: California’s permitted flow
California is the only state with a mandatory, jurisdiction-level ADU permit return. This is what it reported, deduplicated by tracking identifier so that a unit appearing in two annual returns is counted once.
| Year | ADU permit units | ADU completions | Completions as % of that year’s permits |
|---|---|---|---|
| 2018 | 12,050 | 3,186 | 26.4% |
| 2019 | 12,992 | 5,957 | 45.9% |
| 2020 | 13,116 | 8,065 | 61.5% |
| 2021 | 21,163 | 10,147 | 47.9% |
| 2022 | 26,224 | 14,043 | 53.6% |
| 2023 | 29,309 | 17,676 | 60.3% |
| 2024 | 31,515 | 20,652 | 65.5% |
| 2025 | 32,126 | 22,630 | 70.4% |
| 2018–2025 | 178,495 | 102,356 | 57.3% cumulative |
California ADU permits and completions by year, from HCD Annual Progress Report Table A2, extracted 2026-09-05. Percentages are HyreADU calculations from the two published counts and compare same-year figures, which is not a cohort completion rate.
Read the last column with care: it divides completions in a year by permits in the same year, and completions overwhelmingly belong to earlier permit cohorts.
It is a pipeline-pressure indicator, not a completion rate. The cohort analysis lives in permits versus completions, and the permit-to-occupancy clock in how long an ADU takes to build.
How to cite an ADU number without being wrong
A HyreADU recommendation, and the rule we hold ourselves to. Four steps, in order of how often they are skipped.
- 1 Name the instrument, not just the number
“1.4 million” is unusable on its own. “1.4 million single-family properties identified from MLS listing text by Freddie Mac in July 2020” is a citable claim that a reader can check and, if they wish, dispute. The clause is not padding; it is the difference between a fact and a rumour.
- 2 Say whether it is a stock or a flow, and over what window
California’s 178,495 is permits over eight years. Seattle’s 7,029 is permits over thirty. Freddie Mac’s 1.4 million is properties ever listed since roughly 1997. Three numbers, three windows, three quantities — and stacking them without saying so is how bad national figures get built.
- 3 Carry the author’s own caveat with the figure
If the source published a bias statement, the citation inherits it. Freddie Mac published three. Dropping them is not simplification; it converts a two-sided uncertain estimate into a false point estimate.
- 4 Do not scale one state to the nation
California has statutes no other state has, in the form of a ministerial approval duty, a size floor, a parking prohibition and a permanent bar on local owner-occupancy requirements.
Multiplying its production rate by a population ratio assumes those away. If you need a national figure and only have California, the correct output is “we do not know”, and this page is what that looks like written down.
What we could not verify
We could not locate a national ADU stock estimate published by any federal statistical agency, and we do not believe one exists as at the retrieval date on this page.
We could not obtain state-level counts behind Freddie Mac’s top-ten state ranking; the study publishes the rank order and the four-state concentration, not the counts. The rank chart above is labeled accordingly and no counts have been imputed.
We could not locate a statewide ADU permit series for any state other than California, despite specifically checking the states with broad enabling statutes. Those states are reported as unmeasured in our growth study, never as zero.
We did not attempt to update the Freddie Mac estimate. It is a July 2020 publication using data to 2019, and six further years of ADU production have happened since. Nobody has published a successor with a comparable method that we could find, and we have not built one.
Questions
What is the number of ADUs in the United States?
Is 1.4 million an underestimate or an overestimate?
Why can you not just use California’s number and scale it up?
How many ADUs are permitted each year in the US?
Does the American Community Survey have an ADU variable?
Are unpermitted ADUs included in any of these numbers?
Which states have the most ADUs?
Will there ever be a proper national count?
Written and audited by
HyreADU Research Desk
Primary-source research, data analysis and fact checking
We are a research desk, not a builder. We read the permit extract, the statute, the HCD return or the fee schedule ourselves, and publish each figure with its source and retrieval date.
Where a number cannot be traced to a primary source, we leave it out and say what we could not verify. Our store-based claims cover California only.
- CA
- the only state this desk will make store-based claims about
- 5
- jurisdictions with extracted ADU permit evidence
- 735
- CSLB-verified companies in the California store
- 0
- national claims from a one-state store
How this desk works
- Primary sources only. Permit counts come from the city or county that issued the permit. Production counts come from HCD’s Annual Progress Report. Rents come from HUD or the Census. We do not cite an article that cites a source; we download the source and compute the figure ourselves.
- This is a California site. The company store is 734 California firms and one New Mexico firm. Permit evidence exists for five named jurisdictions: Los Angeles, San Francisco, Sacramento, San José and unincorporated Marin. A number from that store is titled to those places, never to the United States.
- A permit is not a completion, and a license is not an ADU grade. California licenses no ADU classification. Being named on an ADU permit is evidence of engagement in that jurisdiction, not of quality, completion, or work anywhere else. Owner-builder permits are excluded from contractor counts.
- Calculation is labeled as calculation. Figures we derive are never presented as something HCD, HUD, the Census or a city published. Terner Center research is cited as Terner’s, never restated as ours.
- We do not design, permit or build ADUs, and we take no payment for placement, ranking or a favorable mention. Pages that look like rankings are not: they publish public-record counts and let the reader decide.
- Nothing here is legal, tax or financial advice. Zoning, underwriting and appraisal practice vary by jurisdiction, lender and appraiser. The useful next step on a specific lot is the planning counter and a licensed professional.
Data as of Freddie Mac Economic & Housing Research Insight (July 2020) retrieved 2026-09-05; Census BPS methodology and item definitions and 2023 AHS definitions retrieved 2026-09-05; California HCD Annual Progress Report Table A2 extract, HCD file last updated 4 September 2026. Authorship on this site is organizational: the analysis belongs to the desk rather than to a named individual, and we do not publish credentials we do not hold.
Our editorial policy sets out how we source, date and correct what we publish.
Sources & retrieval dates
- Freddie Mac, “Granny Flats, Garage Apartments, In-Law Suites: Identifying Accessory Dwelling Units from Real Estate Listing Descriptions Using Text Mining” , Economic & Housing Research Insight, July 2020. Source of the 1.4 million distinct single-family properties figure, the method description, the listing-share series (for-sale 1.6% in 2000 to 6.8% in 2019; for-rent 1.8% in 2003 to 4.1% in 2019), the four-state concentration, the top-ten state ranking, the fastest-growing metros, and the verbatim understatement, overstatement and demolition bias statements quoted on this page. Retrieved 2026-09-05.
- California Department of Housing and Community Development, Housing Element Annual Progress Report — Table A2 , Full Table A2 extract, HCD file last updated 4 September 2026. 922,102 rows, 281,321 with UNIT_CAT = ADU. Deduplicated by tracking identifier: 178,495 ADU permit units and 102,356 completions, 2018–2025, across 511 reporting jurisdictions, with four jurisdictions excluded as unusable. The naive year-by-year sum without deduplication is 213,732 permits, 16.5% higher. Retrieved 2026-09-05.
- U.S. Census Bureau, Building Permits Survey — Methodology , Four structure categories: “(1) single family houses (attached and detached combined), (2) two-unit buildings, (3) three- and four-unit buildings, and (4) residential buildings with five or more units.” Universe of about 19,900 permit-issuing places as of January 2023. No accessory category. Retrieved 2026-09-05.
- U.S. Census Bureau, Building Permits Survey — item definitions (items.txt) , Item 434, additions and alterations to housekeeping residential buildings, carries permits and valuation but no units field. Items 101–105 classify buildings by number of units. No accessory item. Retrieved 2026-09-05.
- U.S. Census Bureau, 2023 American Housing Survey — Definitions , Searched for “accessory”, “accessory dwelling”, “ADU”, “granny” and “secondary unit”: none appears. Housing unit and units-in-structure definitions quoted on this page and in the companion study. Retrieved 2026-09-05.
- City of Seattle Office of Planning and Community Development, “Accessory Dwelling Units 2024 Annual Report” , October 2025. Source of the 7,029 ADUs permitted citywide 1994–2024 (4,062 attached, 2,967 detached) used in the evidence ladder as an example of a firm local count with a defined window. Retrieved 2026-09-05.
- U.S. Census Bureau, Building Permits Survey — Annual History by State and Territory , Annual 2025 release, 14 May 2026. Used for the California one-unit series in the companion instrument study; cited here for the coverage of the national permit series. Retrieved 2026-09-05.
A number without an instrument is not evidence
A number with no instrument behind it is why this page has no headline figure. Everything else we publish carries the document it came from and the date we retrieved it — including the parts where the honest answer is that we could not verify it.
HyreADU does not design, permit or build accessory dwelling units. This page is informational and is not legal, tax or financial advice. It deliberately publishes no national ADU count.
Freddie Mac’s findings are Freddie Mac’s and are cited as theirs; California figures are the state’s own Annual Progress Report returns, with the deduplication method stated.
HyreADU’s company and permit store covers one state and five permit jurisdictions and is used nowhere on this page, because a one-state store cannot support a national claim.