How do I take a new engineering delivery model from one team to the whole firm?
Your pilot worked because of the system, not the people or the tool. Here is how to take that model from one team to the whole firm without losing it.

You take it firm wide by treating the first team as a model, not a story. The pilot worked because of the system it ran on, so you write that system down once and deploy it team by team. The Monday after the pilot, the job is not to celebrate. The job is to package what made it work before anyone copies it their own way and you end up with forty experiments instead of one model.
What actually made the pilot work?
It was not the people. You may have good people, but you have good people in the other teams too, and those teams are still slow.
It was not the tool. Every firm buys the same tools now. If the tool were the answer, every firm that bought it would be fast.
What made the first team work was the system around the tool:
- A library of past work the team could reach without asking anyone.
- Templates for the documents engineers produce every week.
- A review filter that caught bad work before it reached a client.
- A daily rhythm that made the work and the problems visible the same day.
Those four things are repeatable. A new tool is repeatable too, but on its own it changes nothing. The system is the asset.
Why does it break when each team copies it its own way?
Because each team will make a reasonable local choice, and the reasonable local choices will not add up.
One team will use the library differently. One will run reviews at a different point. One will set its own daily rhythm. Each choice sounds fine in the room where it is made. Six months later you have six delivery models, six sets of templates, and no way to tell which one is working.
That is the real cost. Not the wasted effort, but the lost comparison. When every team does it its own way, nothing you learn in one team can be trusted in another. The pilot stops teaching you anything.
What does the handoff Monday actually look like?
It looks like a written model, owned by a small central team, with the first team's name kept out of it.
The central team writes down the library structure, the templates, the review steps and the daily rhythm. It writes them at the level of detail a new team can follow without asking the original team a question. Anything the first team did that cannot be written down is either not part of the model or not yet understood.
Then it deploys the model one team at a time. One team, then the next, with the central team in the room for the first weeks of each. The central team fixes the model as it learns, and the fix reaches every team after it.
This is slower at the start than letting everyone copy at once. It is the only version that holds at ten teams.
Why does the central team stay small?
Because its job is the model, not the work. A large central team starts making delivery decisions for the teams, and then you have rebuilt the old firm with a new name.
Keep it to a handful of people who own four things: the library, the templates, the review filter and the rhythm. They do not own the projects. They own the thing every project runs on.
The measure is simple. If a new team can start on the model and be productive without the central team living inside it, the model is working. If the central team has to carry each new team, the model is not written down yet.
What does this look like on lump sum work?
On lump sum work, the pilot showed what a small team on one model can carry. A single engineer and an EIT pair working inside the system ran at a multiple of what the same pair carries in the rest of the firm. Run that through the lump sum pipeline and one such pair, at the prices we see, sits somewhere between half a million dollars and about two million a year of work.
One pair does not prove anything. The point of the handoff is the second pair, then the tenth. If the model is real, each new pair lands near the same number. If each team copies it its own way, the numbers scatter and you learn nothing from any of them.
How do you know a new team is actually on the model?
Do not ask whether it feels like the first team. Check four things:
- Is the team using the shared library, with new work added back to it?
- Are the documents built from the shared templates, not from last year's files?
- Is the review filter catching work before the client sees it, and is it catching less over time?
- Is the daily rhythm running, with problems surfacing the same day?
If all four are yes, the team is on the model and its numbers can be compared to the others. If any are no, the team is running its own experiment, and its numbers tell you nothing about the model.
What to do next week
Write down the four parts of the system your pilot team actually runs: the library, the templates, the review filter, the daily rhythm. Write them for a team that has never seen the pilot. Then pick one more team to start on it, with your small central team in the room for the first two weeks.
Do not announce it firm wide. Do not let every team copy it at once. Prove the written model on the second team the way you proved the idea on the first. The Monday after the pilot is not the finish. It is the day the model starts travelling.
From the book Engineer Your Freedom
Michael Groselle, P.E., is the founder of Modern Engineering Solutions, a multi-state water and wastewater engineering firm. He left salaried engineering in 2017 and built the firm from one resignation letter and zero clients to more than $2M in annual revenue, with a team of 37 and 525+ completed projects.
Sold on Amazon. Prefer the Kindle edition? Buy it on Amazon ($9.99).