A project engineer at a mid-sized civil firm kept a stopwatch running for a week. Every task got logged. At the end of the week, he had a spreadsheet that showed something he had always suspected but never measured. The billable hours were fine. The unbillable hours were the problem.
The task-level breakdown
For a single parcel, full pre-acquisition screening:
FEMA flood panel lookup and zone confirmation: 18 minutes on a clean case, 35 minutes when the parcel straddles two panels or has a LOMA on record.
USDA Web Soil Survey: 25 minutes to pull and interpret, longer if the soil map unit has a septic suitability qualifier or a wetland inclusion note.
Topographic review, slope analysis, and drainage basin delineation from USGS data: 30 minutes, sometimes 60, depending on tools.
Municode or AmLegal ordinance walk for the zoning classification, dimensional standards, and use table: 45 to 90 minutes. This is the longest single task and the most variable. A city with an 800-page unified development ordinance, no in-text hyperlinks, and a broken search returns can eat two hours alone.
Zoning overlay identification, historic district check, and special use permit review: 20 to 40 minutes.
County GIS for ownership, plat history, and easements of record: 30 minutes on an efficient county portal, up to three hours on a county that still requires a records request.
State DOT for access class confirmation on state-route frontage: 15 minutes if the class is already mapped, a phone call of unknown length if not.
Summary memorandum: 45 to 90 minutes of writing.
On a clean case, the total is about five hours. On a complicated case, it is nine or ten.
The invisible cost
The visible cost, the one that shows up on the timesheet, is the research time. The invisible cost is the context-switching cost. A project engineer doing parcel research opens six to nine browser tabs in a single session. Each tab serves a single agency, a single portal, a single look-up. The cognitive tax of switching between ordinance text, flood panels, soils maps, and GIS layers is real.
Research shows context switches cost around ten to fifteen minutes of recovery time each. A five-hour screening with thirty context switches is, by one measurement, closer to a seven-hour screening once the recovery time is accounted for.
The firm's billing rate is $145 an hour. The five-hour screening bills at $725. The true cost, loaded with recovery time and the opportunity cost of not doing engineering work, is closer to $1,100.
Where the time goes, grouped
Three large buckets absorb most of the hours.
The first is ordinance reading. Municipal codes are dense. Internal cross-references are common. The section on front-yard setbacks in a commercial district might reference a definition three chapters earlier and an overlay two chapters later. Walking those references by hand is slow and error-prone.
The second is reconciliation. The county GIS says one thing about zoning. The city zoning map says something slightly different. The Municode-hosted text of the zoning ordinance has not been amended to match the city's most recent rezoning actions. Sorting out which source is authoritative is a professional skill, and it takes time every time.
The third is writing. Converting raw findings into a client-facing memo. This is the task most resistant to automation and, correctly, the task firms are least eager to automate.
What changes
A research agent that handles the first two buckets, ordinance reading and reconciliation, leaves the professional with the writing and the judgment. The five-hour screening becomes a one-hour screening plus a one-hour memo. The hourly cost drops from $1,100 loaded to $300 loaded.
The more interesting effect is downstream. A developer who can run twenty screenings in a week instead of five asks different questions of their market. The deal pipeline widens. The firms that deploy the tools are doing more deals per capita, not fewer. Employment is not the risk. Throughput is the opportunity.
What firms should measure
Track your unbillable hours. Separate research time from drafting time. Count context switches per screening. The numbers will surprise you. Once they do, the case for compression makes itself.