Measuring Automation Gains: What Does "90% Faster" Actually Mean?
CAD automation tools promise 70%, 85%, 90% speed-ups. Here is what those numbers measure, what they leave out, and how to build an honest measurement for your own process.
In This Article
There is always a percentage on an automation tool's landing page: 70%, 85%, sometimes 90%. Those numbers are not invented — but what they measured is almost never written down.
This post is about how to read that percentage, and how to build your own.
A percentage of what?
When a vendor says "drawing time reduced by 90%", what was usually measured is: time to first output, for a single operation, under ideal conditions.
What was not measured:
- Time to prepare the input (gathering parameters, cleaning the model)
- Time to verify the output
- Time to fix what verification catches
- Time to maintain the automation itself
Add those four and a 90% gain typically lands somewhere between 30% and 50%. That is still a good result — but it is the realistic one, and expectations should be set there.
An automation's return is capped by its slowest link. Speed the generation step up tenfold and leave verification untouched, and your total gain is whatever verification allows.
Measure all four steps
The measurement frame I use on my own projects has four lines:
| Step | Before | After | Why measure it |
|---|---|---|---|
| Preparation | ? min | ? min | Automation demands input; input is not free |
| Generation | ? min | ? min | This is usually the only line the vendor measured |
| Verification | ? min | ? min | Output nobody trusts gets checked every time |
| Correction | ? min | ? min | If the error rate did not fall, the gain is misleading |
Filling in all four rows, before and after, dissolves most inflated claims on its own.
Three traps to watch
1. Measuring the best case. Try the automation on the most convenient part and you will get a beautiful number. Try it on the average job — that is the number you will actually live with.
2. Ignoring the learning period. A new tool slows work down for the first two weeks. Measure in week one and you will miss the gain; never measure it and you hide the real cost.
3. Forgetting maintenance. Automation is living software. When the product family changes, the rule layer changes with it. A few days of maintenance a year come out of the gain.
When measuring is not worth it
Honestly, not every automation needs measuring.
For a ten-minute task done twice a month, the cost of setting up a measurement exceeds the automation itself. In those cases the decision is already clear: it is either an obvious small convenience or it should not be built at all.
Measurement earns its keep when the decision is genuinely uncertain — deciding whether to widen scope, asking for budget, or arguing about whether a tool is actually working.
An example we measured
On the light steel CAD/CAM platform, internal measurements reported a marked speed-up in the preparation stage against the previous CAD system. What makes that number meaningful is that it measured the whole preparation stage rather than a single operation — input gathering and output checking included.
Details are in the case study.
Summary
Look at the four steps, not the percentage. The only question worth asking about an automation tool is: which step does it shorten, and which step does it lengthen?
The final post in this series covers automating the move from 2D drawings to 3D models.