Questions to Ask During a Calibration Audit
For the auditor with calibration on the schedule: an internal or supplier audit of a plant's gauge program, a certification visit, or an assessment of a calibration lab, with whoever runs metrology or quality across the table. The list follows the trail an audit usually takes: what is in the program, how intervals and due dates are run, where traceability comes from, how the method and its uncertainty are controlled, what happens when an instrument is found out of tolerance, and what the records can show. Nearly every answer should be followed by a request to see something, because here the certificate, the label and the instrument on the bench settle more than the spoken answer does.
Want questions from the whole vault instead? Try the random question generator.
The questions
Each question, and why to ask it
The program
Can I see the master equipment list, and how do you know nothing is missing from it?
Why ask it
Work it in both directions. Take a few entries and go find the instruments, then take a few instruments you come across on the floor and look for them in the list. A caliper in a toolbox with no entry tells you more than a tidy database does.
How do you decide which instruments go into the calibration program and which stay out?
Why ask it
The rule you want to hear is about use: anything whose reading accepts product, sets a process or goes into a reported result is in. Ask to see where that rule is written, then find one gauge on the floor that sits outside the program and ask who decided that, and on what basis.
How is each instrument identified, and how do you know which certificate belongs to it?
Why ask it
Check that the ID or serial number is on the instrument itself and not only on its case, since cases get swapped. Sets are the hard part: ask how a single gauge block or pin from a boxed set is tied to its own line on the certificate.
What does a calibration label show, and what do you do when an instrument is too small or awkward to carry one?
Why ask it
Expect an ID, the date calibrated, the due date and who did it, plus any limitation such as a restricted range. For small items the status may be on the case or in a lookup list, so ask an operator to show you how they would check a pin gauge before picking it up.
Which instruments are labeled reference only or calibration not required, and what stops them from being used to accept product?
Why ask it
That label is fine on a ruler that sets up a rough cut and a problem on a gauge at final inspection. Note where two or three of them sit, then read the inspection records for that area to see what the measurement was really taken with.
Who is responsible for the calibration program, and who covers it when that person is out?
Why ask it
A program that lives in one person's head tends to slip during that person's vacation. Ask the named deputy, not the lead, to pull up what is due this month and see how far they get.
How do you match an instrument to the tolerance it will be checking?
Why ask it
A gauge can be in calibration and still be the wrong tool, for example one that reads in steps as coarse as the tolerance on the drawing. Ask whether there is a written rule for resolution and accuracy against the feature being measured, and whether any measurement system study backs it up.
How do you handle tools that belong to employees, customers or contractors?
Why ask it
Personal micrometers and a service contractor's test equipment are the usual gaps, because nobody bought them through the company. The answer should be either that they are in the program like anything else or that they are barred from acceptance work, and you can check either one at a workstation.
Which measurements are made by sensors, fixtures or software built into your machines, and how are those verified?
Why ask it
Programs often cover everything that can be carried to a bench and miss the thermocouple in the oven, the torque driver on the line or the test rig that prints pass or fail. A list of those built-in measuring points should exist, and at least one of them should have a record of its last check that you can read.
How are instruments stored, handled and moved so they stay good between calibrations?
Why ask it
You can see most of this answer for yourself: gauges loose in a drawer, a balance next to a door, reference weights picked up with bare fingers. The trip to and from an outside lab belongs in the answer as well, since a rough ride home can undo a good calibration before the instrument is ever picked up again.
How are adjustment screws and software settings protected so that nobody can change them between calibrations?
Why ask it
Look for an intact seal or paint mark over the adjuster on a few instruments, and for digital ones find out who holds the password to the calibration menu. Then put it to an operator: what would they do on finding a seal broken? Handing the instrument in is the answer the procedure should back up.
Intervals
How was the calibration interval for this instrument set, and where is the reasoning recorded?
Why ask it
Point at a specific instrument when you ask. 'Twelve months because it has always been twelve months' is not necessarily wrong, but a stronger answer cites the maker's recommendation, how hard the instrument is worked and its history of drift.
When did you last lengthen or shorten an interval, and what data led to it?
Why ask it
Intervals that have never moved suggest nobody is reading the as-found results. If one was lengthened, ask to see the run of in-tolerance calibrations that justified it and who signed off on the change.
How does the system flag what is coming due, and who acts on the flag?
Why ask it
Have them run the due list in front of you. Then ask what happens when the flag reaches a supervisor who would rather keep the instrument on the line for one more week.
What is past due today, and where is each of those instruments right now?
Why ask it
The length of the list matters less than where the instruments are: tagged and pulled from use, or still sitting at a workstation. Go and look at one.
Can a due date be extended, who approves it, and how many extensions were granted in the past year?
Why ask it
An extension with a written reason and a named approver is a control. A grace period applied to everything is a longer interval that nobody justified. Count the entries in the log yourself.
Are any intervals based on use instead of the calendar, and how is the use counted?
Why ask it
Counting cycles or hours makes sense for something like a torque wrench or a plug gauge that wears a little with each use. The weak point is the count, so ask who records it and what happens when the tally sheet goes missing.
What checks are done between calibrations to catch drift early?
Why ask it
A daily check weight on a balance, a zero check on a caliper or a run against a master part all count. Each needs a limit that triggers action, and the log should hold real readings, not a column of identical ticks.
What happens to a new instrument between the day it arrives and its first use?
Why ask it
Look for a step that gives it an ID, an interval and a first calibration or a reviewed certificate before it is issued. The most recent purchase is the one to follow through the paperwork, because a gauge that went straight from the box to the line skipped the system.
What happens to an instrument that is taken out of service, and what has to be done before it comes back?
Why ask it
Inactive status is a fair way to avoid paying to calibrate something nobody needs this year. Check that such items are tagged and stored apart, and that the procedure calls for calibration before reuse instead of simply switching the status back.
Traceability
Can you take me from this instrument's certificate back to a national or international standard?
Why ask it
Choose the instrument yourself. Each step should be a certificate that names the standard it was compared against, until you reach a national metrology institute or an accredited lab. A broken link, such as a standard whose own certificate had lapsed on the day it did the work, is worth writing up.
Which outside calibration labs do you use, and how did each get onto your approved list?
Why ask it
You are checking that the lab was evaluated as a supplier and not chosen on price alone. Ask what the approval rested on: accreditation, an on-site audit, or a questionnaire that nobody followed up.
Is the lab accredited to ISO/IEC 17025, and does its published scope cover this measurement at this range?
Why ask it
Accreditation is granted measurement by measurement, so a lab can be accredited for pressure and not for the torque wrench you sent it. Accreditation bodies generally publish each lab's scope, so pull it up and compare the parameter, range and stated capability against one certificate.
What do you tell the lab when you send an instrument out?
Why ask it
A purchase order that only says 'calibrate' leaves the lab to choose the points, the tolerance and whether to adjust. Look for an order that asks for as-found and as-left data, names the specification to test against, and calls for stated uncertainty and an accredited certificate where the program requires one.
Who reviews a certificate when it comes back, and what do they look for before releasing the instrument?
Why ask it
Filing is not reviewing. Ask the reviewer to talk you through the last one: right serial number, the points and range that were ordered, as-found results, uncertainty, and the accreditation mark if one was expected. Some trace of the review, such as initials and a date, should exist.
When a certificate lists a correction or a limit on use, how does the person using the instrument find out?
Why ask it
A balance passed only up to half its capacity, or a thermometer that reads a known amount high, is fine as long as the user knows. Find one certificate with a remark like that and see whether the label, the register or the work instruction carries it.
For calibrations done in house, which reference standards are used and when were they themselves last calibrated?
Why ask it
Note each standard's ID and check its certificate the way you would any other. Then open one in-house calibration record to confirm it names the standard that did the work, because without that the chain cannot be rebuilt later.
How are reference standards kept apart from everyday measuring equipment?
Why ask it
Masters that get borrowed for production work wear and drift like anything else. Look at where they are stored, who holds the key or the sign-out sheet, and whether any show scuffs a reference should not have.
Where no traceable standard exists for a measurement, what is the calibration based on?
Why ask it
This comes up with in-house fixtures, color or appearance masters and some chemical or software measurements. ISO 9001 asks for the basis to be kept as a record in that case, so ask to see it and who approved it, and check the wording of the standard you are auditing against.
When a new instrument arrives with a manufacturer's certificate, do you accept that as its calibration?
Why ask it
A statement that the unit met specification when it shipped is not the same document as a calibration certificate with measured values and traceability. Either can be acceptable under a written rule. What you are checking is that someone made the choice knowingly.
Method
What procedure does the technician follow for this calibration, and where does the method come from?
Why ask it
Ask for the document, then watch a calibration being done and compare the two. A published standard method or the maker's manual is a sound source. A method that exists only as habit will vary with whoever is on shift.
How are the acceptance limits for each instrument decided?
Why ask it
Manufacturer's specification is the default, but some programs widen the limits to what the process needs. That is defensible when it is written down and approved, and a problem when the lab's certificate says pass against one tolerance while the register lists another.
What is the measurement uncertainty for this calibration, and who built the budget behind it?
Why ask it
For a calibration lab this is central, so ask to see the budget and what feeds it: the reference standard, resolution, repeatability, environment. In a company program the fair question is narrower: does anyone read the uncertainty on incoming certificates and compare it with the tolerance?
How does the accuracy of your standard compare with the tolerance of the instrument it calibrates?
Why ask it
A standard barely better than the unit under test cannot tell a good instrument from a bad one. Four to one is a familiar rule of thumb, but ask which ratio their own procedure names and what they do when it cannot be met.
When a reading lands close to the limit, what rule decides whether it passes?
Why ask it
This is the decision rule: whether uncertainty is taken into account before a result is called in or out of tolerance. Ask where it is written, in the procedure or in the agreement with the outside lab, then find a certificate with a borderline point and see whether the rule was followed.
What environmental conditions does this calibration need, and how are they monitored?
Why ask it
Temperature matters most for dimensional work, and humidity, vibration or drafts for balances and some electrical measurements. Look at the log for the day of one specific calibration, and check that the thermometer or logger doing the monitoring is itself in the program.
Who is authorized to perform calibrations, and how was their competence shown for each type of instrument?
Why ask it
A training record signed after a classroom session is weaker than a record of someone being observed doing the work. Compare the authorization list with the technician names on a handful of recent records, and note anyone who calibrated something they were not signed off for.
How are spreadsheets or software that calculate calibration results validated and protected from changes?
Why ask it
A formula overwritten by accident can pass or fail instruments for months before anyone notices. The evidence is a validation record, locked cells or settings, and a test of each new version before it goes live.
Do you take part in proficiency tests or comparisons with other labs, and what did the last one show?
Why ask it
Mainly one for calibration labs, where comparing results with other labs is the outside check on everything else. If a result came back questionable, ask to see the investigation. For an in-house program, a simpler version is whether two technicians measuring the same item agree.
Out of tolerance
Can you show me the last instrument that was found out of tolerance and what happened next?
Why ask it
Ask for the actual case file. You want to see a date, the size of the error, who was told, and a conclusion about the work done with that instrument. If you hear that nothing has been out of tolerance in years, read some as-found data yourself before accepting it.
Do your calibration records give the readings as found, before any adjustment, and again as left?
Why ask it
Without the as-found values nobody can tell whether the instrument was reading wrong while it was in service, so the out-of-tolerance process has nothing to act on. Check certificates where an adjustment was made, since that is where the before data most often goes missing.
How do you find out which product or results were measured with that instrument since its last good calibration?
Why ask it
This only works if inspection or test records name the instrument that took the reading. Pick a recent inspection record and see whether a gauge ID is on it. If it is not, an impact assessment here can only be a guess.
Who judges whether the error was large enough to matter, and what is that judgment based on?
Why ask it
The usual test sets how far the instrument was off against the tolerance of what it measured and how close past results sat to the limit. Look for that comparison written down with numbers. 'No impact' with no working shown is an opinion.
Who decides whether a customer has to be told, and has it ever happened?
Why ask it
Some contracts and sector rules set out when a customer must be notified, so ask what applies to this site and where it is written. A real example with the letter or email attached shows the process has been run and not just described.
What happens when an instrument is dropped, damaged or starts giving readings someone doubts?
Why ask it
Put this one to an operator on the floor as well as the lead. The honest answer out there is sometimes 'I check it against another one and carry on', which tells you whether people know they are meant to hand it in.
Where does a failed or suspect instrument physically go, and what keeps it from being picked up again?
Why ask it
A tag helps. A locked shelf or a quarantine bin is better. Go to the spot and check whether anything in it has been waiting so long that people have started borrowing from it.
After an adjustment or repair, how is the instrument checked before it goes back to work?
Why ask it
Find out what the procedure requires after a repair: a new part or a large adjustment generally calls for a full calibration, and one check point is thin evidence across a whole range. The as-left readings on the record should have been taken after the final adjustment, not before it.
How quickly does an outside lab tell you about an out-of-tolerance result, and is that written into your agreement with it?
Why ask it
If the news arrives only with the certificate, weeks may pass while affected product ships. Look for a notification clause in the purchase order or contract, and compare the date on one failed certificate with the date the investigation was opened.
How many out-of-tolerance results did you have in the past year, and what did you do about the pattern?
Why ask it
A count shows the data is collected. What you are after is the next step: an interval shortened, a model of gauge replaced, a workstation retrained. The same type of instrument failing again and again with no change means problems are being logged and not learned from.
Records
How long are calibration records kept, and can we pull one from several years ago?
Why ask it
The retention period should come from a written rule that reflects what customers and regulators in that industry require, so ask where the figure comes from. Then test it by naming an instrument and a year. How long the search takes is part of the answer.
Who can edit a record or change a due date in the calibration database, and is every change logged?
Why ask it
Look at the list of users with edit rights and ask why each one needs them. Have someone bring up the change history on a single record. A due date that moved with no reason given is a thread worth pulling.
What did the last internal or external audit of calibration find, and how was each finding closed?
Why ask it
Read the earlier report before you ask, then go and check one closed finding at the place it was raised. A finding closed on paper that you can still see on the floor is a bigger issue than the original was.
What do you report to management about the calibration program, and how often?
Why ask it
On-time rate, number overdue and out-of-tolerance rate are the usual figures. Ask what management did the last time a number moved the wrong way, because a report nobody reacts to changes nothing.
If I pick three instruments from the floor right now, can we pull the complete file on each?
Why ask it
Save this for late in the audit and choose from different areas: one from production, one from the lab, one from a maintenance cart. For each you want the register entry, the label, the current certificate, the standard it was calibrated against and the previous as-found result.
What would you change about the program if you had the time or the budget?
Why ask it
A good way to close, because the person running the program usually knows its weak spots better than any auditor will after a day. Whatever they name, treat it as a lead to verify and not as a finding in its own right.
How to run the calibration part of an audit
Practical guidance for the conversation itself
Preparing for the audit
Read the standard and the site's own procedure first
A calibration audit is judged against written requirements: the clause on measuring resources in ISO 9001, ISO/IEC 17025 for a laboratory, a customer's supplier manual, and above all the site's own calibration procedure. Read the edition that applies before you arrive and mark the places where the procedure makes a promise you can test, such as a review of every certificate or a tag on every overdue instrument. Where a rule depends on the industry, the customer or the regulator, ask the site which one applies to it instead of assuming.
Ask for the register and the lab list ahead of time
Request the equipment list, the approved outside labs and the last audit report a week or so beforehand. With the list in hand you can choose your sample before anyone knows which instruments you will ask about, and you can look up each lab's scope of accreditation at your desk instead of spending audit time on it.
Know whether you are auditing a user or a lab
A manufacturer's program is mostly about control: is everything in the system, calibrated on time, and acted on when it fails. A calibration lab is also judged on the technical work itself, so the questions under Method carry far more weight there. Decide which you are facing and move your time accordingly.
Plan the sample as well as the questions
Choose instruments across types and areas: something dimensional, something electrical or thermal, a reference standard, an item calibrated in house and one sent out. Add at least one you expect to be awkward, such as a gauge on the night shift cart or a sensor built into a machine.
On the floor and at the bench
Turn each question into a request to see something
An answer in a meeting room describes the procedure. The certificate, the label and the instrument describe what happened. After nearly every question on this page the next words are 'can you show me one', and the one should be your choice.
Trace forward and backward
Go from the list to the instrument to confirm the records match what is on the floor, and from an instrument you find to the list to confirm nothing is missing. Do the same with product: from an inspection record to the gauge that made the measurement, then to that gauge's certificate for that day.
Talk to the people who hold the instruments
The metrology lead knows the system as designed. Operators and inspectors know what they do when a gauge is overdue, dropped or missing. Ask an operator how they would know that the micrometer in their hand is good to measure with today, and compare that with what the lead told you.
Watch one calibration from start to finish
If any work is done in house, ask to observe a routine one with the procedure open beside you. Note the standard's ID, the warm-up or soak time, the number of points and how the readings are written down. Small departures from the written method are easier to see than to hear about.
Keep your own notes traceable
Write down instrument IDs, certificate numbers and dates as you go. A finding that says 'some gauges were overdue' can be argued with. One that names the gauge, where it was and the date on its label cannot.
Weighing what you find
Separate a slip from a system gap
One overdue caliper in a sample of thirty, already tagged and on its way to the lab, is a slip. Three overdue instruments still at workstations, or one with no entry in the register at all, says the control does not work. Widen the sample before deciding which you are looking at.
Follow the consequence past the instrument
The instrument is only the start of an out-of-tolerance finding. What makes it serious is whether anyone asked what had been measured with it. A program can calibrate everything on time and still fail at that step.
Read certificates as data
A certificate is more than evidence that calibration happened. Compare the points tested with the range the instrument works over, and read the as-found values, the stated uncertainty and any remark about limited use. A certificate that covers half the working range is a gap even though the label says calibrated.
Give credit for what the site found itself
A program that catches its own overdue items, logs its own out-of-tolerance results and shortens intervals in response is working, even if its numbers look worse than those of a site that records nothing. Say so in the report.
Traps for the auditor
Auditing the labels and stopping there
Walking the floor checking stickers is quick and finds overdue gauges. It says nothing about whether the calibrations behind the stickers were sound, traceable or reviewed. Spend at least as long on certificates and out-of-tolerance cases as on labels.
Taking accreditation as the whole answer
An accredited lab is a strong start, but only for the measurements on its scope and only when the certificate was issued under the accreditation. Check both before you tick the box.
Insisting on your own preferred interval or ratio
Quality standards mostly ask a site to set its intervals and limits and to be able to justify them. They rarely name a number. If the site has a reasoned rule and follows it, a different rule you met elsewhere is not a finding.
Letting the sample be chosen for you
If the host offers to bring a few records to the meeting room, thank them and pick your own. The instruments nobody volunteers are where the gaps tend to be.