Should a Planning Department Automate Review Routing or Cut Review Stages First?

The walk between desks is not the problem

Your department has a backlog, the council has a target, and the vendor demos all make the same promise. An application arrives, the system reads its type, location, and valuation, and it routes itself to every reviewer who needs to see it. No more folders on carts. No more emails asking who has the file. Everyone gets a dashboard.

That promise is real, and it answers the wrong question. Routing automation makes an application move faster between the same number of desks. If a project touches zoning, fire, engineering, public works, and health, and each of them holds its own clearance, the software has not removed a stage. It has removed the walk between stages. In most departments we look at, the walk was never where the months went.

The question worth answering before the purchase order is narrower and less comfortable: which of your reviews change an outcome often enough to justify being a gate, and which would do the same work as a comment?

What the software is actually selling

Read the permitting platforms' own descriptions and the pattern is consistent. OpenGov lists automated routing and workflow management among its core features. Accela emphasises configurable workflows that track jurisdiction-specific review steps from intake through approvals and rechecks. GeoCivix, Citizenserve, and the rest of the category say much the same thing. Capture the application in a structured form, then send it to the right reviewers based on rules.

Notice what every one of those descriptions preserves. The reviewers. The steps. The rechecks. None of these products is designed to tell you that a step should not exist, and it would be odd to expect one to. A configurable workflow engine is priced, implemented, and renewed on the complexity it holds. A department with fourteen review steps is a bigger implementation than a department with six.

That is not a criticism of the vendors. It is a description of their side of the table. Ours is the other side, and from here the order of operations looks different.

A dozen clocks, and any one of them can stop the others

Los Angeles is the honest illustration, because it publishes its own machinery. A single building permit there can require clearances from City Planning, the Bureau of Engineering, the Fire Department, the Department of Transportation, Urban Forestry at the Bureau of Street Services, the Department of Water and Power, Sanitation, and occasionally the City Attorney.

The detail that matters is in the Bureau of Engineering's own permit manual. For a driveway, curb ramp, or sidewalk clearance, engineering staff will review the item but will not approve it until Urban Forestry has approved theirs. One reviewer's gate sits behind another reviewer's gate. Practitioner guides to the process describe the result plainly: some departments will not review plans until another finishes, some divisions have only a few engineers serving the entire city, and a clearance phase can run anywhere from two to twelve weeks.

Now put routing software on top of that. The application reaches Urban Forestry the same afternoon instead of the following week. Good. It then sits in the same queue, behind the same citywide caseload, served by the same small team, and Engineering still waits for it. You have bought a few days at the front of a process measured in months.

This is the general shape. When a review stage is a hard gate, the total timeline is set by the slowest gate plus every dependency between gates. Routing shortens hand-offs. It does not shorten the slowest gate, and it does not remove a dependency. Only a decision about the stage itself does that.

The counterfactual is already written into law

Nobody has to speculate about what happens when you remove stages instead of speeding them up, because California legislated it.

SB 9 requires local agencies to approve qualifying duplexes and lot splits ministerially, without discretionary review or a hearing. Once a complete application is received, the agency has 60 days to approve or deny it, and if it does neither, the application is deemed approved. Compare that with the discretionary path the same project might otherwise take, with hearings, notice periods, and appeal windows measured in months or years.

Look at the mechanism rather than the politics. The legislature did not tell cities to route SB 9 applications faster. It took the discretionary stages out, replaced them with objective standards a planner can check, and put a clock on what remained. The speed came from subtraction.

San Francisco is running a local version of the same experiment. Under its PermitSF reforms, agencies review certain permits concurrently rather than sequentially, and the assigned planner consolidates every department's comments into a single plan check letter. The applicant stops answering multiple cycles of review from different groups of reviewers. KQED reported in March 2026 that average housing permit approvals in the city fell from 605 days to around 280. The reporting does not attribute that drop to any single reform, and neither will we. But the reforms themselves are structural: fewer review cycles, one letter, one coordinator. The platform is how that gets encoded, not what produced it.

Gate or comment: the distinction that decides the project

Every review in your process is doing one of two jobs, and most departments have never sorted them.

A gate is a review that can stop a project and regularly does. It exists because the consequences of getting it wrong are serious and hard to reverse. Fire access and life safety are the obvious examples. So is anything touching structural adequacy or a utility's capacity to serve.

A comment is a review whose findings are almost always resolved by a condition of approval or a minor plan revision. It adds value. It rarely changes whether or how the project proceeds. It can run in parallel, it can feed a consolidated letter, and in many cases it can be satisfied by an objective standard checked at intake.

The problem is that almost every review in a municipal process was constituted as a gate, because gates are how departments protect their authority and their liability. Over twenty or thirty years, reviews that began as a sensible control become a standing hold, and nobody asks how often the hold has actually changed an outcome. If you ask that question with data, you tend to find a small number of reviews that genuinely decide things and a larger number that mostly annotate. The data to answer it usually already sits in your permitting system, in the comment history nobody has ever pivoted.

Routing software encodes whatever you have. If you configure it with fourteen gates, it will enforce fourteen gates with perfect reliability, and every future reform will have to be made twice: once in policy and once in the workflow configuration. We made the same point about permit types in whether a city should automate intake or reduce permit types first, and it holds even more strongly for review stages, because a stage costs time on every application that passes through it.

Where routing does earn its money

We are not arguing against the software. Once the stages are right, routing automation is genuinely valuable, and the value is specific.

It makes concurrent review practical. Sending one application to five reviewers at once, tracking five sets of comments, and assembling one letter is miserable on paper and easy in a decent platform. It makes the queue visible, which matters because the slowest gate is often invisible from inside your own department. And it produces the timing data that lets you keep asking the gate-or-comment question after go-live, which is how a reform survives the next reorganisation.

All three of those benefits get larger after the stage decision, not smaller. Concurrent review of six gates beats concurrent review of fourteen. A dashboard showing where the real hold sits is more useful when there are fewer places to hide. If what you actually need is that visibility across departments, it is worth reading what orchestration means once you stop reading the brochure before you scope a platform.

It also helps to know what each review costs you in staff time, which is the number that makes the conversation with a department head concrete. We walked through where that number hides in what a permit application costs a county to process.

Why this is hard, and why it is still first

The analysis is not the difficult part. Pulling two years of review outcomes by stage and counting how often each one changed a project is a few weeks of work. The difficult part is that every gate has an owner, the owner has a director, and nobody wants to be the person who converted their department's hold into a comment the year before something goes wrong.

That is a governance problem, and software cannot solve it. A platform can make the conversation easier to have, because it shows everyone the same queue. It cannot have the conversation for you. Departments that buy the routing first usually discover this eighteen months later, when the dashboard is live, the backlog has barely moved, and the vendor's recommendation for the next phase is another module.

Where this article stops

We cannot tell you from here which of your reviews are gates and which are comments. That depends on your review outcome history, which dependencies between departments are written into code and which are habit, what your state law already requires to be ministerial, and where your slowest citywide reviewer actually sits. Those answers live in your own records and in the people who run the plan check queue.

If you are being quoted for permitting or plan review software and nobody has asked which of your review stages should still be a gate, talk to us before the configuration workshop. We will sort your reviews by what they actually decide, show you where the real holds sit, and hand your vendor a workflow worth encoding.