How we helped a civil engineering firm earn 34% more operating profit per dollar of net revenue.

More projects weren't producing the margin they wanted.

The civil engineering firm wanted to raise operating margin to 20%, but its land-development teams kept absorbing repeat work on fixed-fee projects. Every additional hour spent correcting a plan set used more of the fee they'd already agreed to.

County review was part of the delay, and hiring more senior engineers looked like a way to get more work through. But before recommending more people or another tool, we traced what was coming back from the counties and how much time the firm's reviewers spent on it.

The same requirements kept coming back as new corrections.

We compared county comment letters with the firm's project records and studied how engineers spent their time. What we found was that recurring checklist omissions, inconsistencies between sheets, and requirements raised on earlier projects were still reaching the county.

The senior engineers knew what each jurisdiction expected, but that knowledge wasn't available to the people preparing the drawings in a form they could use before review. So the people with the least time to spare kept finding and explaining the same issues.

Sets waited a median of seven business days for senior review before the county's review even began. Adding reviewers could help with the queue, but it would leave the team dependent on them to catch requirements the firm already knew.

We needed to make that knowledge usable earlier, while the engineers preparing the set could still act on it.

We gave the team the checks they needed before the drawings reached a senior engineer.

We worked with the senior engineers to capture their county requirements and connect them with the firm's past comment letters and drawing markups. Those sources became the basis for a pre-submission check that flagged likely issues and showed the requirement or earlier comment behind each flag.

The preparer could inspect and resolve each issue before senior review, while anything the check couldn't support with a source stayed with a human.

We also gave engineers a way to ask questions against the same requirements while preparing the drawings. The knowledge could now help prevent an issue before it became another correction.

Licensed engineers retained the final review and seal. The check helped the team prepare a better set for them, while they remained responsible for the engineering decisions.

The check had to work on their examples before it went into daily use.

The senior engineers helped define examples of the issues the check needed to find. We used those examples to test whether it caught relevant defects without filling the review with unnecessary flags, and checked it on held-out examples before rollout.

The review process then put a first pass with the readiness report ahead of the senior engineer's final review. That moved known corrections to people who could resolve them earlier, leaving fewer of those issues for the final reviewer to work through.

Our first report made the team leave the work to deal with it.

We initially delivered the flags in a separate report. But during rollout, one team was dismissing flags in bulk under a deadline, and we found that the report sat outside the drawing-review tool where they were already working.

We'd added a step that made it harder to act on the findings. We worked with the client's team to put the flags into their existing review tool, so engineers could deal with an issue alongside the drawing it concerned. The dismissal rate fell the following week.

Senior engineers regained a full workday per set, and more plans cleared the first submission.

At the six-week operating follow-up across the pilot offices, senior review time had fallen from 14.2 to 6.1 hours per set. That freed 8.1 hours per set, just over an entire eight-hour workday of senior engineering capacity.

First-submission approval more than doubled, from 29% to 61%, and reached 66% at the later quarterly check. Review cycles fell from 2.4 to 1.3 per set, while the median wait for senior review fell from seven to two business days. The team had removed a business week of waiting before submission as well as reducing the work that came back afterward.

The division also accepted three pursuits it had previously deferred, putting the freed capacity behind work it wanted to take on.

We completed the engagement in April 2025. At the twelve-month follow-up in April 2026, the client confirmed 20.4% operating margin, surpassing its 20% goal, and 54% fewer unpaid rework hours in land development. The client credited about 2.5 to 3 percentage points of the margin improvement to the program, with business mix and market conditions contributing the rest.

Their team could keep the requirements current without waiting for us.

County requirements would change, so the handover needed to cover how to update the check as well as how to use it. We provided an editor, operating instructions, and training in testing changes before release.

The client's team then added another jurisdiction, ran the tests, and published the update without us operating the system. They had a way to keep the requirements current and extend the work themselves.

The technical companion explains the checks, professional review, measurement, and handover.

Less repeat work gave the firm more room to grow.

The firm needed more margin from the work it was already winning. We found that repeat corrections were using up both fixed fees and senior review capacity, then helped the team catch those issues earlier and maintain the checks themselves.

That meant more of a senior engineer's time could go toward work that needed their judgment, while fewer hours disappeared into another round of corrections.

If you're trying to grow your own firm, start with your goal, look at what it'll take to achieve it, and then identify what's keeping you from taking action. Just remember that every firm is different and that repeat corrections and senior review were just their bottleneck and it might not be yours.