Your Permit Clock Starts Later Than You Think
Mid-construction commercial tower, the stage that follows permit issuance
Five American cities publish how long it takes to approve a project. All five measure from a starting line that developers never actually cross.
In March, San Francisco's Budget and Legislative Analyst delivered a report to the Board of Supervisors containing two numbers that should be read side by side.
The first: 280 days. That was the median time to process a building permit for a housing project, counted from the day the application was filed, for permits issued between January 2024 and August 2025.
The second: 114 days. That was the median for applications submitted from January 2024 onward, once the entitlement phase was stripped out of the count.
Same city. Same department. Same era. A difference of five and a half months, produced entirely by where you decide to start the stopwatch.
This is not a San Francisco problem. It is the standard practice of American permitting, and once you see it you cannot unsee it. Nearly every jurisdiction in the country that reports how fast it reviews plans is reporting a number that begins after the application has been accepted, screened for completeness, and formally logged into the queue. The developer's clock started earlier. Sometimes much earlier. The gap between those two moments is where projects quietly die, and almost nobody measures it.
Call it the acceptance gap. It is the single most useful thing a developer can understand about permitting, and it is almost never on the dashboard.
The two clocks
Every American permit application passes through the same sequence, whatever the local vocabulary. You submit. Somebody at intake screens the package for completeness. If something is missing, a seal, a survey, a stormwater calculation, a fee, the package goes back to you. When the jurisdiction is satisfied that the submission is whole, it accepts the application, or in the language of several state statutes, deems it complete.
Only then does the official review clock start.
Everything that happens before acceptance is real time in the developer's life and invisible in the city's statistics. So is the time an application spends sitting on the applicant's desk between review cycles, waiting for a consultant to answer a correction notice. Cities generally do not count that either, and they have a defensible reason: they are measuring their own performance, not yours.
Seattle is the rare jurisdiction that says this plainly. Its Department of Construction and Inspections publishes a permit performance dashboard with two separate measures. One is total calendar days, covering everything from intake through issuance. The other is calendar days in city control, which counts only the days the city is actually working the file and excludes the stretches when the application is with the applicant awaiting a resubmittal or a payment. Seattle states that the total time an applicant experiences runs roughly double the days in city control.
Roughly double. That is the acceptance gap, quantified by a city willing to publish it.
Most cities publish only the smaller number, or a number whose definition is not documented well enough to tell which one it is.
What five cities actually publish
We looked at the five largest American permitting jurisdictions that maintain open construction permit datasets: Chicago, Austin, San Francisco, New York, and Los Angeles. The question was simple. Can a developer, using public data, learn how long it will actually take?
The answer varies from partially to not really.
City | Primary open dataset | What the published clock measures | Best available time figure |
Chicago | Building Permits (ydr8-5enu) | Application start to issuance, via a processing-time field whose exact definition is not documented publicly | No current city-published median. The city's dedicated permit performance datasets stopped updating in 2014 |
Austin | Issued Construction Permits (3syk-w9eu) | Applied to issued, but only for permits that were actually issued | 91-day median for building permits; 465-day average for site plans across six review cycles |
San Francisco | Building Permits (i98e-djp9) | Filed date to issued date | 280-day median from filing; 114 days once entitlement is excluded |
New York City | DOB Permit Issuance (ipu4-2q9a) plus DOB NOW datasets | Filing to issuance, split across two systems that must be merged | Plan review wait of roughly 4.9 days for new buildings, which measures queue time only |
Los Angeles | LADBS permit datasets | Issuance, with limited usable application-date fields | Department estimate of 15 to 30 business days for regular plan check |
Take these one at a time, because the failure modes differ.
Chicago publishes a genuinely useful dataset covering 2006 to the present and refreshed close to daily. It carries an application start date, an issue date, and a pre-computed processing-time field, which sounds like exactly what a developer needs. The trouble is that the city does not document what that field nets out. Whether it excludes applicant hold time is unclear from the published dictionary. Meanwhile Chicago's dedicated permit performance datasets, the ones designed to answer this question directly, have not been updated since 2014. A developer who wants a current Chicago median has to compute it, and then has to caveat it.
Austin publishes the most useful commercial-relevant metrics of any city in the study. Its Development Services Department tracks average days for the initial review cycle, average days in completeness check, average review cycles per permit issued, and total application-to-issuance days, broken out for site plans specifically. That completeness-check figure is precisely the measurement most cities omit. Austin's issued-permits dataset carries the standard limitation: it contains only permits that made it out the other side, so every application that was abandoned, withdrawn, or is still languishing is missing, and any median computed from it runs optimistic.
San Francisco now has the best documented picture in the country, because someone asked. The March report from the Budget and Legislative Analyst, prepared at the request of Supervisor Bilal Mahmood, is the most detailed public accounting of a major city's permitting performance we found. It reports the 280-day median, the 114-day post-reform median, a 15-day median to first review, and, crucially, a median of three submissions per permit. Two rounds of staff comments before approval. For context on how far the city had drifted, California's housing agency data showed San Francisco averaging 605 days to permit a project that was already entitled.
New York presents a structural problem rather than a measurement problem. Permit records live in two systems. The legacy Buildings Information System holds older and paper filings. DOB NOW holds filings from 2018 forward, and the department is migrating filing types across as it goes. Any citywide analysis has to union both datasets and deduplicate by job number, because a single job can spawn multiple permit rows for different work types. The department does publish a service-level tracker showing average plan review wait times, and the numbers look wonderful, under five days for a new building. But that measures how long a submission waits in queue before a plan examiner opens it. It is not the time from filing to permit, and it should never be quoted as if it were.
Los Angeles is the thinnest. LADBS distinguishes between express permits requiring no plan check and permits that go to plan review, and it tells applicants to expect 15 to 30 business days for regular plan check, 6 to 10 for expedited. Those are service estimates published by the department, not audited outcomes. The open datasets are oriented toward issuance and do not carry clean application-date fields, which makes a reliable intake-to-issue interval difficult to derive. We could find no recent controller audit measuring LADBS plan check turnaround. The most recent controller review of the department we located dates to 2006 and concerns expired permits, not review speed.
So: one city that publishes the honest number and is not in the study (Seattle), one that had it extracted by its own legislative analyst (San Francisco), one that tracks the right metrics but only for issued permits (Austin), one that has the data but not the definitions (Chicago), one split across two systems (New York), and one that mostly offers estimates (Los Angeles).
The number that actually predicts your timeline
Here is what changed our thinking while assembling this.
The variable that best predicts how long a project takes is not the city. It is the number of review cycles.
San Francisco's median permit required three submissions. Austin's site plans averaged six review cycles, and in the years before its reform program, a city review found that 81 percent of applicants revised three or more times. Seattle's city auditor found the median second-round review taking 27 days against a 14-day target, roughly double.
Do the arithmetic. If each additional cycle costs somewhere between two and four weeks of city review, plus whatever time your consultants take to respond, then the difference between a two-cycle project and a six-cycle project is on the order of six months. That is a larger swing than the difference between the fastest and slowest city in this study.
And the cycle count is the one variable in the entire system that the applicant substantially controls.
This is an uncomfortable finding for anyone who prefers to blame the bureaucracy. Cities are genuinely slow, and some of them are slow in ways that no amount of applicant diligence can fix. But a plan set that clears completeness screening on the first pass, and that answers the reviewer's standard objections before they are raised, moves through a system that is otherwise identical at a fraction of the elapsed time. The correction notices that generate additional cycles are strikingly repetitive: incomplete submittal packages missing checklist items or professional seals, setback dimensions that cannot be verified from the drawing, parking counts and accessible stall geometry that do not comply, stormwater documentation that does not tie to the site calculations, utility easements the layout encroaches on, fire apparatus access and turning radii that fail, survey discrepancies between the plat and the plan.
None of that is exotic. All of it is preventable at the drafting stage. It is also, not coincidentally, the reason the studio publishing this article exists.
Why the shot clocks do not save you
Several states have tried to legislate the problem away by imposing statutory deadlines. It is worth understanding exactly how these are drafted, because the drafting is where the acceptance gap becomes law.
California enacted AB 2234, which requires a local agency to determine whether a post-entitlement permit application is complete within 15 business days, then to approve it or return a full set of comments within 30 business days for projects of 25 units or fewer, and 60 business days for larger ones. Missing the deadline is a violation of the Housing Accountability Act. The law reaches only developments where at least two thirds of the square footage is residential, so a pure commercial project gets nothing from it.
Florida requires a local government to approve, approve with conditions, or deny an application within 30 business days for structures under 7,500 square feet and 60 business days at or above that threshold, counted from a complete and sufficient application. Florida also did something clever: the local government must give written notice of completeness within five business days, and if it fails to, the application is automatically deemed properly completed and accepted. That is one of the only statutory provisions in the country that puts a meter on the pre-acceptance phase itself. A separate provision covering private plan review providers goes further, deeming an application approved as a matter of law if the building official does not issue a deficiency notice within 20 days.
Texas gives municipal authorities 30 days to approve, conditionally approve, or disapprove a plat, failing which it is approved by operation of law, and sets a 45-day shot clock on city building permits. Legislation in 2023 clarified filing and completeness procedures and allowed a qualified third party to complete the review when a city misses its deadline.
Washington requires a final decision within 65, 100, or 170 days depending on whether the application needs public notice and a hearing, with a general 120-day ceiling. The statute says explicitly that the count runs from the day completeness is determined. For applications filed after January 1, 2025, a jurisdiction that blows the deadline may owe the applicant a refund of 10 to 20 percent of the application fee, and jurisdictions above 20,000 population began filing annual performance reports in March 2025.
Read those again and notice the common structure. With the partial exception of Florida, every one of these clocks starts at completeness. The legislature did not overlook the pre-acceptance phase. It wrote the statute to exclude it, because a city cannot reasonably be held to a deadline on a submission that is missing half its sheets.
Which is fair. It is also why a developer who reads the statute and plans to a 60-day deadline is planning to the wrong number.
What the delay costs
Convert this to money, because that is the language a construction lender speaks.
Commercial construction debt in 2026 has been running roughly 7.5 percent to above 10 percent depending on the lender. Bank construction facilities have priced around SOFR plus 275 to 400 basis points, putting all-in coupons in the 7.0 to 8.75 percent range, while debt funds have priced wider, roughly SOFR plus 400 to 550. One market tracker put the weighted average construction loan rate at 8.4 percent as of April 2026, built on a 4.8 percent SOFR base plus spreads of 2.5 to 4.0 points.
Interest accrues only on drawn funds, so the effective carry during construction sits below the headline coupon. Still, the arithmetic is unforgiving. Average outstanding balance multiplied by the rate, divided by twelve:
$5 million average drawn at 8.4 percent: roughly $35,000 per month.
$10 million average drawn at 8.4 percent: roughly $70,000 per month.
Every additional review cycle, at two to four weeks each, therefore costs a mid-size commercial project somewhere between $17,500 and $70,000 in pure interest carry. Before escalation. Before extended general conditions. Before the rate lock expires and has to be repriced.
The National Association of Home Builders and the National Multifamily Housing Council, in their joint study of regulatory cost, put a number on the piece attributable to delay alone. They found that regulation imposed by all levels of government accounts for an average of 40.6 percent of multifamily development cost, that changes to building codes over the preceding decade were the largest single component at 11.1 percent, and that the additional interest generated purely by regulatory delay averaged 0.7 percent of total development cost. They describe that 0.7 percent as the cost regulation would impose even if it created no other expense at all. The same study found that organized neighborhood opposition, where it appeared, added an average of 5.6 percent to total development cost and pushed completion out by an average of 7.4 months.
On a $30 million project, that 0.7 percent is $210,000 spent on nothing but waiting.
The blind spot: almost none of this research is about commercial
Here is a caveat that most coverage of permitting omits, and that anyone building a retail center, an industrial park, a hotel, a self storage facility, or a senior living campus should hold on to.
The overwhelming majority of rigorous American research on permit timing measures housing. The NYU Furman Center and Pew work on California's streamlined ministerial approval track measured affordable housing developments, finding that in Los Angeles the median approval fell from about seven months before the law to about three months after, and that in San Francisco qualifying projects averaged around four months against roughly a year previously. Berkeley's Terner Center studied 2,474 San Francisco developments and found permitting timelines ranging from under a year to as long as fourteen years with little consistency by project type. The NAHB and NMHC cost study surveyed multifamily developers. San Francisco's Budget and Legislative Analyst report examined housing permits.
All of it is excellent. None of it necessarily describes what happens to a car wash.
The commercial evidence that does exist is thinner and more scattered. Austin's site plan metrics are the best commercial-relevant series any city publishes, and they are sobering: an average of 465 days across six review cycles. San Francisco data from 2017 showed commercial alteration, addition, and repair permits averaging 172 days to issue. Los Angeles publishes its 15 to 30 business day plan check estimate, which covers commercial work. Santa Monica's new self-certification pilot, launched in January 2026, is aimed squarely at commercial tenant improvements and promises permits in five business days or less for qualifying project types, which appears to be the only program in the region that removes plan check entirely for identified commercial work.
That is close to the full inventory. If you are underwriting a commercial project and someone quotes you a permitting benchmark, ask what asset class produced it. The odds are good the answer is apartments.
What is actually changing
Two of the five study cities have measurably improved, and the mechanism in both cases is worth noting.
San Francisco launched PermitSF in February 2025 under a mayoral executive directive. The subsequent analysis found that applications submitted from January 2024 forward were clearing at a 114-day median against the 280-day historical figure, that 98 percent of applications were being screened for completeness within the 21-day target, that 72 percent of first-round reviews met the 30-day target, and that 88 percent of second-round reviews met the 14-day target. A proposal surfaced in January 2026 to merge the planning and building inspection departments outright.
Austin's reform program cut initial site plan review by 56 percent, from an average of 87 to 99 days down to 32, and pulled follow-up cycles from around 50 days to just under 15, close to the city's own 14-day target. In October 2025 the city reduced the change-of-use site plan review from eleven disciplines to three, reportedly shortening that pathway from about two months to about two weeks, and opened an expedited site plan review pilot.
Both cities also adopted automated plan review. Austin brought in an AI checking tool in October 2024 after a three-month pilot; Los Angeles city and county launched a similar tool at the end of April 2025, initially in service of wildfire rebuilding. Los Angeles separately created a self-certification pilot for single-family homes up to three stories destroyed in the January 2025 fires, letting qualified architects certify residential code compliance and bypass plan check, though not inspection.
It is too early to know whether automated review changes outcomes or simply relocates the bottleneck. What the San Francisco and Austin numbers do establish is that a large fraction of permitting delay is administrative rather than substantive. Nobody rewrote the building code. They changed how the queue works, and the timeline moved by months.
The clock you can control
Return to the two San Francisco numbers.
The 114-day figure reflects a city that reorganized its process. The 280-day figure reflects the same city before it did. Both are real, and neither one is the number a developer should plan around, because the number a developer should plan around depends on something the city does not control and does not report: how many times the plan set comes back.
Three submissions is the San Francisco median. Six cycles is the Austin site plan average. Those are not measures of municipal competence. They are measures of the distance between what was submitted and what the code required, and that distance is set at the drafting table months before anyone stands in line at intake.
The permitting system in this country is slower than it should be, and reforming it is a legitimate civic project. But a developer waiting on reform is waiting on a legislative session. A developer who submits a complete, code-checked, reviewer-anticipating plan set is operating on the one lever available today, and on the evidence, it is the larger lever.
The clock starts when they accept your application. How long you spend getting there, and how many times you go back, is mostly up to you.
Sources:
Municipal open dataChicago Building Permits
Austin Issued Construction Permits and DSD Site Plan Performance Metrics
San Francisco DBI Building Permits
NYC DOB Permit Issuance and DOB NOW
Los Angeles LADBS permit datasets and plan check service estimates
Official reports and auditsSan Francisco Board of Supervisors Budget and Legislative Analyst
Post-Entitlement Permitting in San Francisco
Seattle Office of City Auditor construction permitting performance audit
Comments