What a Permit Application Actually Costs a County to Process, End to End

The number you found is the wrong number

Somebody on the board, or a commissioner, or a builder at a public hearing, asked what it costs the county to process a permit. You went looking. Every result on the first page answered a different question: what a homeowner pays for a deck permit, what a contractor budgets per square foot, a fee schedule from a jurisdiction two states away. All of that is the applicant's cost. None of it is yours.

The county-side number does exist. It is in the fee study your finance office commissioned the last time fees were updated, and it has been sitting there since, read once by the people who set the fee and never again by the people who run the department. That gap is the subject of this article. The fee study already contains the operating metric. Almost nobody uses it as one.

What the fee study actually computed

A fee study, done properly, produces two things.

The first is a fully burdened hourly rate: what an hour of staff time in the permitting function costs once salary, benefits, supervision, facilities, IT, and the department's share of county overhead are loaded on. Los Angeles City Planning's consultant, NBS, developed a composite rate of $199 per hour across all of the department's routine case processing programs, and the memo is explicit that the rate is total annual cost divided by the hours available to do the work. Napa County's study arrived at $153.62. Madera County's Planning Division uses $282. The rate varies with the jurisdiction, and every one of them is a real, defensible number that somebody was paid to compute.

The second is a time estimate per activity. That same Los Angeles memo prices a historic resource assessment review at 5.25 hours and a first entitlement through the planning commission at 30 hours, and multiplies each by the rate to get a cost of service. Ouray County, Colorado, hired MGT Consulting to run the same arithmetic across every department that touches a land use application, and the worked example shows Land Use, Administration, Legal, and Road and Bridge each carrying their own burdened rate and their own estimated hours. Four departments, four rates, one application.

Multiply the two and you have what the permit costs the county. Not what the applicant pays. What it costs.

Why the number then disappears

The fee study is commissioned to answer one question: are we allowed to charge more, and how much. In California the answer is bounded by the constitution, which caps fees at the cost of service, so the study establishes a ceiling and the council decides whether to charge it. That is the whole use case. Once the fee ordinance passes, the study has done its job.

Nobody asked the study the operating question, which is the same data read the other direction. If a first entitlement is estimated at 30 hours, how many hours did the last fifty actually take? Which permit types consume more staff time than the estimate, and which consume less? The cost side is already measured to the quarter hour for fee purposes. The variance is never tracked, because tracking variance is a management activity and the study was a revenue activity.

Los Angeles is a useful case precisely because it is well run. The last comprehensive study was completed in November 2020, fees took effect in December 2021, and a new study landed in October 2025. Even on that cadence, the department is operating for four years at a time on time estimates nobody re-checks in between, with an automatic inflation adjustment that moves the fee and tells you nothing about the hours. That is the norm, not the exception. Thurston County, Washington, adopted new fees for 2026 after finding its previous comprehensive review dated to 2009. Sixteen years is a long time to run a department on a number.

What the number tells you when you finally read it

Placer County published the version every county should want. Its 2025 fee study across building, engineering, planning, and short-term rentals found the fee-related work costs about $31.3 million a year and current fees bring in about $15.2 million, a recovery rate around 48 percent and a general fund gap of roughly $16.1 million. Of the 1,804 fees reviewed, 1,637 were set below cost.

That headline is not the interesting part. The breakdown is. Building services recovered about 49 percent. Engineering and surveying, about 47 percent. Planning, about 35 percent. The short-term rental program recovered 104 percent and was the only one that paid for itself.

Read that as an operating report rather than a fee proposal and three categories fall out, and they fall out in every jurisdiction we have looked at.

Permits the county loses money on. Usually the discretionary ones, where a planner writes a staff report, a hearing gets noticed, and legal reviews the findings. The fee was set years ago, the hours have grown with every new environmental and design standard, and the subsidy is invisible because it never appears as a line. Thurston's own analysis said exactly this: current fees fell short of covering review and inspection under modern construction and environmental standards.

Permits the county subsidises on purpose. Some of that 35 percent in planning is policy. A council may decide that a fence permit for a homeowner should be cheap. That is a legitimate choice. It is only a problem when nobody can say which permits are subsidised by choice and which by neglect, and in our experience the honest answer is that the second list is longer.

Permits that are mostly a tollbooth. The ones that cost the county forty minutes and generate a fee, a routing rule, a review checklist, and a queue position. A permit that cheap to process is not paying for itself so much as paying for its own existence. It is the toll that keeps on tolling.

A county that can sort its permit types into those three buckets can have an adult conversation about fees, about staffing, and about which permit types should exist at all. A county that cannot is having the staffing argument without the arithmetic, and the staffing argument is hard enough already, as we described in our work on the government talent gap.

The measurement you can actually take

The rate you already have. The hours are the missing half, and there is a shortcut most departments overlook.

Mountain View and Ventura County both run planning cost recovery on actual staff time, billed at a fully burdened rate against an initial deposit. That model only works if staff log hours per application, which means those jurisdictions are already tracking the exact data a per-permit cost needs. They could produce a cost per permit type, by actuals rather than estimates, tomorrow. Most of them have never printed the report.

If your county bills on deposit for any permit class, start there. Pull twelve months of logged hours by permit type, multiply by the burdened rate the fee study gave you, and compare it to the estimate the fee was built on. The variance is your finding. Where actual hours run well over estimate, you have either a fee that is stale or a process that has grown, and those need different fixes. Where actuals run under, you have a permit type that may be a consolidation candidate.

If you bill flat fees everywhere and log nothing, the measurement is the same one we describe for what a district actually spends on enrollment paperwork: throughput, not allocation. One intake season, one tally, applications fully processed per reviewer per week by type. It takes a tally sheet and a supervisor who will protect the exercise for six weeks.

What this means for the software decision

The reason this question usually gets asked is that a permitting vendor has quoted a price and someone wants a baseline to hold it against. The per-permit cost is that baseline, and it also tells you where the software will and will not help.

Automation removes handling time. Intake, routing, fee calculation, status notices, the front counter's job of explaining which of sixty permit types an applicant needs. It does not remove a planner's 30 hours of discretionary review, and a vendor case study that claims a 50 percent reduction in processing time is almost always measuring elapsed days on the simplest permit class. Before you accept that figure, run it through the discipline in the automation ROI math vendors show you, recalculated. The honest version of that math starts with which permit types are handling-heavy and which are judgment-heavy, and the only way to know is the per-type cost you just built.

There is a harder implication. If a permit type costs the county forty minutes and is issued nine times a year, digitising it preserves it. The right move may be to fold it into a neighbour, and no vendor is going to suggest that, because there is no license revenue in a permit that stops existing.

Where this gets specific is where an article has to stop. Which of your permit types are under-recovering by policy and which by drift, what your fee study's time estimates assumed, whether your deposit-billed hours are clean enough to trust, and which discretionary reviews have quietly doubled since the estimate was made: those answers live in your fee study appendix, your permitting system's time logs, and about three weeks of sitting with the people who do the work.

If you have been asked what a permit costs your county and the only number you can find is the fee, talk to us. We will turn the fee study you already paid for into the operating metric it should have been, sort your permit types into the three buckets, and tell you which ones to reprice, which to consolidate, and which are actually worth automating.