Original research
From permit to final inspection: how long an ADU really takes
Every published ADU timeline is a builder describing their own construction phase. This one is measured from the City of Los Angeles permit record, from the day the permit issued to the day the certificate of occupancy did — and it keeps the unfinished projects in the denominator.
Written by HyreADU Research Desk Primary-source research and survival analysis
Audited by HyreADU Research Desk
The finding
On 39,278 non-supplemental City of Los Angeles building permits that add an accessory dwelling unit, issued between 2020 and 31 August 2026, the estimated median time from permit issuance to certificate of occupancy is 499 calendar days — about 16.4 months — with an interquartile range of 292–1,066 days.
That figure is a Kaplan-Meier estimate, which is to say it keeps the 14,306 permits that have not completed inside the calculation. Take the median of only the 24,972 that finished and you get 367 days, roughly four months shorter.
The shorter number is the one a builder’s own records would produce, and it is wrong in a specific direction: it can only see the projects that ended. Only 35% of issued ADU permits reach a certificate of occupancy within a year, and 65% within two.
A detached new-build ADU runs longest (median 564 days); a conversion coded as alteration runs shortest (419).
Separately, and this is the finding we did not expect: 15% of the ADU permits Los Angeles issued in 2020 still have no certificate of occupancy and no permit final, 6.2 years on. For the 2022 cohort it is 21%. Those permits are not slow.
On the evidence available they are the permit-to-completion gap, visible for once at the level of the individual project. This is Los Angeles, not California and not the United States. It is not a compliance finding, not a construction schedule, and not advice. HyreADU does not design, permit or build ADUs.
Read this before you use the number
- One city is the entire frame: not California, and not a national ADU build time
Los Angeles is the only jurisdiction this desk has found that publishes an issuance date and a certificate-of-occupancy date on the same permit row.
San Francisco, Sacramento, San José, unincorporated Marin and San Diego were all opened; none of their published extracts carries a usable completion date paired to an ADU flag.
So the frame is the City of Los Angeles, and a figure from the City of Los Angeles is a figure from the City of Los Angeles.
The "3 to 6 months of construction" band that circulates on builder sites is not a measurement we can source, here or anywhere.
- A certificate of occupancy is not the day the drywall went up
The clock measured here ends when the department closes the file — 24,876 of the 24,972 completions are a certificate of occupancy date, the remaining 96 a "Permit Finaled" status.
It therefore contains everything between issuance and sign-off: mobilisation, financing that fell through and was rebuilt, inspection scheduling, correction notices, the owner who stopped for a year, and the paperwork lag between the last inspection and the certificate.
It is the clock a homeowner lives through. It is not a construction-crew productivity measure and must not be quoted as one.
- The headline is censored-aware on purpose, and the naive figure is published beside it
14,306 of 39,278 permits in the frame had not completed by 2026-08-31. Dropping them produces a median of 367 days; keeping them, via Kaplan-Meier, produces 499.
We publish both, and we lead with the second, because the first systematically over-samples the projects that were quick enough to finish inside the window. Read the method section before reusing either.
- Permits issued in 2025 and 2026 cannot yet have a median
A permit issued in March 2026 has had five months to complete. Its cohort’s upper quartile is unreachable and its median is an extrapolation from a small number of fast finishers.
Where a percentile is not reached inside the observation window we print "not reached" rather than a number. Do not read the year-by-year table as a trend in construction speed; read it as a censoring diagram.
- n is on every figure, and the areas are not ranked
Every cut on this page carries its permit count and its observed-completion count.
The seven area planning commissions differ by 121 days at the median, which is real, but the composition of permit types differs across them too and we have not decomposed that.
It is a description of where projects sit, not a league table of neighborhoods.
The ADU clock has two halves and the market only ever quotes one
Ask "how long does an ADU take" and you will be answered about whichever half the answerer controls. A city talks about review. A builder talks about construction. Nobody adds them.
- Half one: application to issuance
- The review clock. Statutory in California — 60 days from a completed application under Government Code § 66317 — and measured, for three jurisdictions, in our companion study how long an ADU takes to approve. In Los Angeles the measured median on the published clock is 162 days from submitted to issued.
- Half two: issuance to certificate of occupancy
- The build clock, which is what this page measures. No statute governs it. In Los Angeles the censored-aware median is 499 days. It is roughly three times the review clock, and it is the half every timeline article treats as the easy part.
- What a homeowner experiences
- Both, end to end, plus the design and engineering time that precedes any application and appears in no permit record at all. We do not add our two medians together and present a total: they are different jurisdictional frames, different cohorts and different classifiers, and the sum would be an invented number with a false precision.
- What "never finalled" means
- A permit that issued and then stopped producing records. Not a delay measurement — an outcome. 14,306 permits in this frame sit there today; 517 of them were issued in 2020.
The build clock in four numbers
(the quick quarter) n = 39,278
(about 16.4 months) Kaplan-Meier
(nearly three years) 24,972 completions observed
three years after issuance Estimated from the same curve
Why the honest number is four months longer than the obvious one
The obvious way to answer this question is to take every ADU permit that has a certificate of occupancy, subtract the issue date, and report the median.
Do that here and you get 367 days (n = 24,972, IQR 238–563, mean 446). It is a real calculation on real records and it is the figure almost any analyst would ship.
It is also biased, and biased in a knowable direction. A permit issued in 2024 can only appear in that calculation if it finished before 2026-08-31.
The slow ones from 2024 are still running, so they are not in the numerator; they are simply absent. The faster a project was, the more likely it is to be counted.
Statisticians call this right-censoring, and the standard correction is more than a century old.
The Kaplan-Meier product-limit estimator handles it by asking a different question at each day: of the permits still unfinished and still under observation on day t, what share completed on day t? An unfinished permit contributes to the denominator for every day it was observed and then leaves without ever being counted as a completion.
The resulting survival curve is unbiased under the assumption that censoring is independent of completion time — which here means the assumption that a permit’s issue date does not itself predict how long the build takes, only how much of it we have been able to watch.
On this frame the correction is worth 132 days: 367 becomes 499. The upper quartile moves further, from 563 days to 1,066, because the slow tail is exactly what censoring hides. HYRE calculation, on LADBS’s records; the method is standard, the application to ADU permits is ours.
This matters commercially and not only academically. A homeowner planning around 367 days is planning around a survivor.
A lender sizing a construction-loan interest reserve around twelve months is sizing it for the fastest third of projects.
The timeline planner on this site defaults to editorial placeholders for exactly this reason: we would rather it be obviously a placeholder than quietly a survivor’s number.
The same permits, two estimators
| Estimator | Permits in the calculation | Median days | Upper quartile | What it answers |
|---|---|---|---|---|
| Completed permits only (naive) | 24,972 of 39,278 | 367 | 563 | How long did the ADUs that have already finished take? Useful, and not the question a homeowner is asking. |
| Kaplan-Meier (censoring-aware) | 39,278, all of them | 499 | 1,066 | For a permit issued today, how long until the certificate of occupancy? This is the headline. |
City of Los Angeles ADU building permits issued 2020 through 2026-08-31, supplementals excluded. HYRE calculation on LADBS resource pi9x-tg5x, retrieved 2026-09-05. The mean of the completed set is 446 days and is published only as a contrast; it is never the headline on a skewed distribution.
A detached ADU takes about five months longer than a conversion
| Permit type | Permits | Completions observed | Median days | IQR | Within 1 year | Within 2 years |
|---|---|---|---|---|---|---|
| Alteration/repair (conversion) | 16,438 | 11,282 | 419 | 230–999 | 45% | 68% |
| Addition (attached ADU) | 12,227 | 7,544 | 528 | 325–1,121 | 31% | 63% |
| New construction (detached ADU) | 10,612 | 6,145 | 564 | 365–1,058 | 25% | 62% |
LADBS permit_type, which is the department’s classification of the work, not ours. Kaplan-Meier medians. The gap between an alteration and a new detached structure is 145 days at the median.
What the permit-type gap does and does not tell you
HYRE analysis, clearly labeled as analysis: the 145-day spread between a conversion coded as alteration and a detached new build is the shape you would expect if foundation, utility trenching, separate service and a full inspection sequence were the difference — but this dataset cannot prove that mechanism, because it does not record what happened between inspections.
What it does show is the ordering, and the ordering is stable across every cut we ran.
The within-one-year column is the more useful one for planning. 45% of conversion-coded permits reach a certificate of occupancy inside a year against 25% of detached new builds. Put the other way: three-quarters of detached ADU permits are still open on their first anniversary.
Junior ADUs — units carved out of the existing house under Government Code § 66335, capped at 500 square feet — are faster again.
On 3,778 JADU-flagged permits the median is 387 days against 509 for standard ADUs, and 47% finish inside a year. That is consistent with a smaller scope of work and no new structure.
It is not an argument that a JADU is the right unit for any particular lot; a JADU shares the house’s kitchen wall, its entrance requirements and its rentability profile, and none of that is a schedule question.
The finding we went looking for something else and found instead
15% of the ADU permits Los Angeles issued in 2020 — 517 of 3,373 — still carry no certificate of occupancy and no permit final on 2026-08-31, 6.2 years after issuance. The 2021 cohort is at 18% and the 2022 cohort at 21%.
Some of that is genuinely slow construction still under way. Some of it is not: 304 permits across the frame carry a "Permit Expired" status, and 12,735 sit at plain "Issued" with no subsequent record.
We cannot tell from these fields which permits were abandoned, which were superseded, which were built and never signed off, and which are still running. So we report the cohort, not a cause.
What we can say is that this is the permit-to-completion gap seen from underneath.
That study takes HCD’s Annual Progress Report and shows that California’s permit counts and its completion counts are different numbers that the market quotes interchangeably.
This page shows the same divergence in the individual permit records of one city, and it is not small.
Permits with no certificate of occupancy, by year of issue
| Issued in | Permits | Completed by 31 Aug 2026 | Still no final | Share with no final | Years elapsed |
|---|---|---|---|---|---|
| 2020 | 3,373 | 2,856 | 517 | 15.3% | 6.2 yr |
| 2021 | 5,071 | 4,174 | 897 | 17.7% | 5.2 yr |
| 2022 | 7,120 | 5,625 | 1,495 | 21.0% | 4.2 yr |
| 2023 | 6,158 | 4,674 | 1,484 | 24.1% | 3.2 yr |
Counts, not estimates — no censoring correction is needed for "has this permit completed yet", only for "how long did it take". The share rises with recency because recent cohorts have had less time, which is why only the 2020 and 2021 rows should be read as an outcome rather than a snapshot.
By year of issue, with the censoring visible
| Issued in | Permits | Completions observed | KM median | KM upper quartile | Naive median |
|---|---|---|---|---|---|
| 2020 | 3,373 | 2,856 | 486 | 1,010 | 395 |
| 2021 | 5,071 | 4,174 | 508 | 1,076 | 427 |
| 2022 | 7,120 | 5,625 | 536 | 1,170 | 429 |
| 2023 | 6,158 | 4,674 | 512 | 1,105 | 394 |
| 2024 | 6,394 | 4,384 | 476 | not reached | 342 |
| 2025 | 6,933 | 3,045 | 462 | not reached | 258 |
| 2026 | 4,229 | 214 | not reached | not reached | 126 |
Read the last two columns against each other. In 2020, with six years of observation, the naive median is 395 against a corrected 486.
In 2025 the naive median collapses to 258 days, which is not evidence that ADUs got faster — it is evidence that only the fast ones have had time to finish. This table is a censoring diagram, not a trend line.
Where in the city the permit sits
| Area planning commission | Permits | Completions observed | Median days | IQR | Within 2 years |
|---|---|---|---|---|---|
| South Valley | 11,379 | 7,328 | 464 | 277–1,014 | 67% |
| North Valley | 9,764 | 6,444 | 467 | 273–975 | 68% |
| South Los Angeles | 6,297 | 3,883 | 551 | 321–1,258 | 61% |
| West Los Angeles | 4,196 | 2,539 | 504 | 301–1,001 | 66% |
| East Los Angeles | 3,346 | 2,138 | 527 | 305–1,070 | 65% |
| Central | 3,274 | 1,986 | 585 | 334–1,370 | 59% |
| Harbor | 1,022 | 654 | 529 | 314–977 | 65% |
LADBS area planning commission field. Not a ranking: permit-type mix differs across the seven areas and we have not held it constant. The spread from South Valley (464 days) to Central (585) is 121 days at the median.
The geographic spread is real and we are not going to over-explain it
The two San Fernando Valley commissions carry 21,143 of the 39,278 permits in the frame — more than half the city’s ADU activity by permit count — and they complete fastest, at medians of 464 and 467 days.
Central Los Angeles is slowest at 585 days with an upper quartile of 1,370, and South Los Angeles sits at 551.
HYRE analysis: lot size, the prevalence of detached new construction over conversion, contractor availability, financing, and the density of existing structures on a parcel all plausibly differ between a Valley R1 lot and a Central multifamily parcel, and any of them could produce this ordering.
We have not decomposed it, so we are not attributing it. The honest statement is that where the permit is predicts something about how long it takes, and this dataset does not say what.
Method, in the order it was run
Every classifier choice we could have made, and what it does to the headline
| Specification | Permits | Completions | Median days | IQR |
|---|---|---|---|---|
| Headline: permit adds a net ADU or JADU (adu_changed > 0 or junior_adu = 1), supplementals excluded | 39,278 | 24,972 | 499 | 292–1,066 |
| Department use classification only (use_desc contains “Accessory Dwelling”), supplementals excluded — the ADU-04 classifier | 25,979 | 16,660 | 506 | 301–1,071 |
| Both classifiers must agree | 25,360 | 16,267 | 508 | 304–1,071 |
| Headline classifier, supplemental permits left in | 41,359 | 26,618 | 485 | 280–1,025 |
The estimate sits between 485 and 508 days across every reasonable definition of an ADU permit in this dataset — a range of 23 days around a headline of 499. The finding is not an artefact of the filter.
What a homeowner should do differently
- Budget the interest reserve for the upper quartile, not the median
If construction financing is involved, the exposure that hurts is the tail. A quarter of Los Angeles ADU permits are still open 1,066 days — nearly three years — after issuance.
Our financing study covers how the loan products treat a project that runs long; the short version is that most of them assume it will not.
- Ask a builder for their permit numbers, not their timeline
A contractor quoting "four to six months" is describing crew time on site, which is a real number and a different number. Permit records are public.
Ask which Los Angeles permits they were the contractor of record on, then look up the issue and certificate-of-occupancy dates yourself — our permit-contractor index is built from exactly those records.
- Treat a conversion and a detached build as different projects on the calendar
The 145-day median gap between them is larger than most people’s entire mental model of the schedule. If the timeline is the binding constraint rather than the rent, that gap is a real input to the type decision. The type selector walks the trade-offs; this is one of them.
- Do not add our two medians together
The approval median (162 days, from a different cohort and classifier) and this build median (499 days) are not summable into a single "661-day ADU". They are different frames. If you want a planning envelope, use each half’s own IQR and understand that they overlap in ways no dataset here records.
Questions
How long does it take to build an ADU?
Why is your number so much longer than the 4 to 6 months builders quote?
What is Kaplan-Meier and why did you use it?
What share of ADU permits never get finished?
Is a garage conversion faster than a detached ADU?
Does this apply outside Los Angeles?
Are ADUs getting faster to build?
Can I use this data?
Is this legal or construction advice?
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 LADBS resource pi9x-tg5x retrieved 2026-09-05; feed rowsUpdatedAt 2026-08-31; observation window closes 2026-08-31. 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.
The data behind this page
Every figure on this page is computed from one file, and that file is published here so the analysis can be checked, disagreed with, or reused.
Sources & retrieval dates
- City of Los Angeles Department of Building and Safety — Building and Safety: Building Permits Issued from 2020 to Present (N) , Socrata resource pi9x-tg5x on data.lacity.org. Fields used: permit_nbr, permit_type, use_desc, adu_changed, junior_adu, issue_date, cofo_date, status_desc, status_date, apc, business_unit. 409,619 rows downloaded in full; rowsUpdatedAt at retrieval 2026-08-31; maximum refresh_time on the retrieved rows 2026-08-30. Retrieved 2026-09-05.
- California Government Code § 66317 , The ministerial 60-day ADU decision, which governs the review half of the clock and not the half measured here. No California statute sets a deadline for completing construction once a permit has issued. Recodified out of former § 65852.2 by SB 477 (Stats. 2024, ch. 7); amended by SB 543 (Stats. 2025, ch. 520). Retrieved 2026-09-05.
- California Government Code § 66335 , The junior accessory dwelling unit provisions, including the 500-square-foot ceiling and the existing-dwelling requirement, cited in the JADU comparison above. Retrieved 2026-09-05.
- E. L. Kaplan and Paul Meier, “Nonparametric Estimation from Incomplete Observations,” Journal of the American Statistical Association 53:282 (1958), 457–481 , The product-limit estimator used for every median and quartile on this page. Cited as method, not as an ADU source. Retrieved 2026-09-05.
- HyreADU Research Desk — ADU construction duration, per-cut results , The computed table behind every figure on this page, as CSV. Kaplan-Meier medians and quartiles plus completion shares at one, two and three years, by permit type, unit type, area planning commission and year of issue. Retrieved 2026-09-05.
- HyreADU municipal permit reconnaissance (2026-08-29 to 2026-09-05) , DataSF i98e-djp9, Sacramento BldgPermitIssued FeatureServer, San José CKAN resources 761b7ae8 and df4b8461, Marin mkbn-caye, San Diego seshat.datasd.org approvals-issued. Opened and checked for a completion date paired to an ADU classification. San José publishes FINALDATE but no ADU-flagged issuance pairing we could use on this pass; the others publish no completion date at all. All excluded, and the exclusion is stated rather than papered over. Retrieved 2026-09-05.
Plan against the quartiles, not the median
The timeline planner lets you set your own review, construction and finishing durations. Its defaults are editorial placeholders, deliberately, because a single number for "construction" is the thing this study says is unsafe. Put 292 days in the optimistic column and 1,066 in the pessimistic one and see what it does to your carrying cost.
HyreADU does not design, permit or build accessory dwelling units, does not file applications and does not schedule construction. This page is informational and is not legal, construction, financial or scheduling advice.
Every figure is a HYRE calculation on City of Los Angeles Department of Building and Safety permit records (Socrata pi9x-tg5x, retrieved 2026-09-05), not an LADBS finding and not a statement by the City of Los Angeles.
Durations are calendar days from permit issuance to certificate of occupancy or permit final, estimated with a Kaplan-Meier product-limit estimator that treats the 14,306 permits incomplete at 2026-08-31 as right-censored.
The frame is the City of Los Angeles only; nothing here is a California or a national ADU build time. Permit-type and area cuts are the department’s classifications and are not rankings.
No causal claim is made about why any cut differs from any other.