Timesheets tell you exactly how busy your engineers are. What they do not tell you is whether all that busyness produces anything. A department can be fully booked for weeks while half the time goes into exporting, retyping bills of materials and repairing drawings that came back from the shop floor. On the timesheet, that looks identical to designing.
If you want to steer, you need something other than hours. In this article we lay out four engineering KPI metrics that actually mean something: lead time from order to release, first-time-right of drawings, share of rework, and the ratio of routine work to design work. For each indicator you will read how to measure it with minimal effort, without a registration circus. What you do with the outcomes differs per lever: routine work, for example, can be automated away with a toolbox like Thundercad, but first it has to be visible where the time leaks away.
Why hours say so little
Hours measure input, not results. Forty hours booked on a project says nothing about whether the design is done, whether the drawings were buildable in one go, or whether the customer gets the machine sooner or later. Two departments with identical timesheets can perform completely differently.
Steering on hours also has a perverse edge. The engineer who automates his export work and finishes sooner books fewer hours and looks less productive on paper. The one plodding through the evening on an error that could have been prevented looks like a hero. The measurement rewards exactly the wrong behavior.
Hours remain fine for job costing and quotations. But as a performance indicator for the department, you need something that talks about outcomes: how fast, how good, and what the time is actually spent on.
Four KPIs that do mean something
Together, the four indicators below answer the questions a manager really has. They are deliberately chosen so you can measure them with data that largely already exists:
| KPI | Which question does it answer? | Where does the measurement come from? |
|---|---|---|
| Lead time from order to release | How long does the rest of the company wait for engineering? | Order date from ERP, release date from Vault or release folder |
| First-time-right of drawings | How often does work come back from the shop floor? | Tally sheet at work preparation |
| Share of rework | How much capacity goes into changes after release? | One extra booking code |
| Routine work versus design work | How much time is left for real engineering? | Sample week |
Lead time from order to release
Both dates already exist: the day the order came in sits in your ERP, the day of release sits in Vault or follows from the date of the released drawing package. Count the working days in between and you have your first KPI, without anyone recording anything extra. Use the median instead of the average: one monster project would otherwise distort the whole picture. And split by order type if needed, because a customer-specific machine and a repeat order do not belong in one number.
First-time-right of drawings
Every drawing that comes back from the shop floor with a question or an error cost time twice: in engineering and in production. Measure this with a tally sheet at work preparation or the shop supervisor: every time a drawing has to go back to engineering, one mark plus the reason in three words. Divide the number of returns by the number of released drawings in the same period and you have your first-time-right.
The reason in three words matters more than the number itself. "Dimension missing", "old revision", "hole collides": after a month of tallying you see which cause tops the list and where the first countermeasure should land. What those returns cost and how to trace them back to the source structurally is something we covered earlier in Failure costs from data errors: follow them from the shop floor back to the source.
Share of rework
Rework is all the time spent on documents that were already released: fixing errors, adding forgotten details, processing changes that could have been known during design. You measure it without a circus by introducing one extra code next to the normal project code: "change after release". Engineers do not have to do anything new, only pick the right one of two codes when booking.
The signal is in the trend. A stable small share comes with the business; customers change their minds. A growing share means errors surface later and later, and late-found errors are the most expensive ones.
Routine work versus design work
The sneakiest time sink is work that requires no thinking: exporting, printing, transferring BOMs to Excel, updating iProperties, sorting out folders. Nobody books it separately, so nobody sees how much it is. Measure it with a sample: for one week, every engineer notes at the end of the day, at quarter-hour level, how much time went to three categories: design, routine, other. One measuring week per quarter is enough for a reliable picture; permanent registration mostly adds irritation.
Does the sample show that a sizeable part of the week goes into exporting, BOMs and iProperties? That part is the easiest to push back: Thundercad automates exactly this kind of routine work, from Batch Publish to Export BOM.
Try 30 days freeMeasuring without a registration circus
The death of every KPI is a measuring burden bigger than the insight. Four principles keep it light:
- Use data that already exists. Order dates, release dates, booking codes: better a slightly coarse measurement from existing systems than a perfect one nobody keeps up.
- Sampling over permanent registration. One measuring week per quarter says enough about the ratio of routine to design.
- Trend over snapshot. Whether the median is twelve or fourteen days matters less than whether it is rising or falling.
- Measure the department, not the person. The moment KPIs are plotted per individual engineer, they become a verdict and everyone starts optimizing the measurement instead of the work.
One sub-measurement we deliberately leave aside here: the quality of the CAD data itself, such as filled iProperties and correct BOMs. How to measure that health is covered in How healthy is your CAD data? How to measure your metadata quality.
From measuring to improving
A KPI is conversation material, not a grade. Discuss the four numbers with the team periodically and pick one each time to dig into: where does the deviation come from, and which single countermeasure do we try until the next review? Long lead times often turn out to be waiting for missing customer information at the front. Low first-time-right points toward drawing checks or a release checklist. A high routine share is fixed fastest by automating export and BOM work.
Finally, watch out for the familiar effect that a measurement which becomes a goal in itself loses its meaning. Demand that first-time-right must go up, and you sometimes mainly get fewer reports instead of better drawings. So keep it small and grown-up: two KPIs to start with, tallied in the open, and the question "what do we learn from this" above the question "who did that".
Frequently asked questions
Which KPI is the best one to start with?
Lead time from order to release, because nobody has to record anything new for it: both dates are already in your systems. After that, put the first-time-right tally sheet at work preparation. With those two you are having a conversation about facts instead of feelings within a month.
Does this also work for a department of two or three engineers?
Especially there. Small departments feel every wasted afternoon directly in the planning, and the measurements above cost almost no overhead. The numbers do get more erratic with small quantities, so look at the trend across several periods rather than at single outliers.
What can Thundercad do with these KPIs?
The measuring is up to you; the toolbox mainly helps on the improvement side. Routine work such as exporting drawing packages in bulk or pushing BOMs into your own Excel format largely disappears from the weekly workload, and you see that back in the ratio of routine to design work. Experiencing what that saves is free for a month via the trial download.