Twenty questions to ask before you buy.
None of them need an instrument to answer. Most of them get asked too late, or by us, after the money is already spent.
Metrology is three things you are already doing: collecting data, processing it, and reporting it. These questions run in that order, with a set on requirements ahead of all three, because the requirement is where the expensive mistakes start.
Read them straight through. The ones you cannot answer are the useful ones. If you can answer all twenty, you are in better shape than most plants we walk into, and you do not need to call us.
Before you spend a dollar
Is the problem you have the problem you think you have?
Is the requirement even correct?
Somebody set that tolerance. Can they tell you why? A number copied off the last drawing is not a requirement.
What decision does this measurement feed?
If nobody can name the decision, you are paying to generate a number nobody uses.
Is this actually a measurement problem?
A plate that comes out twisted is a shrinkage, restraint, and sequence problem. Measuring it more precisely will not straighten it.
Can your process hold the tolerance you have been given?
No instrument closes a capability gap. It only tells you how far short you are, more often.
If the equipment arrived Monday, who runs it?
Capital is the easy part. A tracker with nobody qualified to plan a shot is an expensive tripod.
Collecting the data
The only part most shops think about, and still the part most often planned last.
How do you plan the shot?
Where you set up, how many stations, in what order, and what you do when line of sight is blocked at station three.
Are you measuring from the drawing’s datums, or from something convenient?
A convenient alignment gives you a clean report and the wrong answer.
How is the part held while you measure it?
Free state, fixtured, on its side, still on the machine. Restraint changes the result, and the drawing may not say which condition it means.
How many points, and where?
Your sampling strategy decides whether you find the deviation or average it away.
Would two of your people get the same answer?
If not, you are measuring your operators, not your parts.
Do you know your uncertainty, or your instrument’s spec sheet?
Catalog accuracy is a laboratory claim. Your uncertainty includes the setup, the fixture, the environment, and the person holding the probe.
Processing the data
Where two people can take one honest scan and reach opposite verdicts.
What happens to the data after you collect it?
For most shops the honest answer is that it sits on a laptop until the laptop gets replaced.
Who does the alignment and the fitting, and does everyone do it the same way?
Two analysts, one scan, two verdicts. That is not measurement, that is opinion.
Could you reproduce a result from six months ago?
That means the raw data and the recipe that was run on it, not the PDF that came out the end.
How would you know if you did it wrong?
A measurement that cannot fail a check has not passed one.
Are you looking at trends, or only at pass and fail?
The drift you throw away is the warning you will wish you had kept.
Reporting the result
The end everyone treats as paperwork, read later by people who were not there.
Who reads the report, and can they act on it without calling you?
A report that needs its author in the room is not a deliverable, it is a meeting.
Does the report answer the question your customer actually asked?
A colored deviation map is not an answer to whether it will fit.
How long from the measurement on the floor to a decision made?
A result that arrives after the part shipped is a record, not a control.
If you are a supplier, do you know what your customer does with your data?
They may be using it to judge your capability, plan their fit-up, or decide the next award. You are being measured by your measurements.
If a few of these stopped you, that is the conversation.
No charge to find out whether we are the right people to help. Plenty of first calls end with us pointing you at something you can fix yourself.