What is the difference between as-found and as-left calibration data?

As-found is the reading taken before any adjustment, capturing the instrument in the condition it was used on the floor. As-left is the reading taken after adjustment or repair (or after confirming none was needed). As-found tells you whether past measurements were valid; as-left tells you the instrument is fit to use going forward.

What is reverse traceability in calibration?

Reverse traceability is looking backward from a failed as-found reading to assess the impact on product already measured. You define the suspect window from the last known-good calibration to the failure, list what the instrument measured, evaluate the error against each characteristic's tolerance, and disposition the affected product by quarantine, re-measurement, recall, or documented acceptance.

Do both readings have to be on the calibration certificate?

In a serious program, yes. Under ISO/IEC 17025:2017, accredited calibration certificates report results before and after any adjustment or repair when that data is available. Recording only as-left hides whether past measurements were valid; recording only as-found cannot prove the instrument is safe to keep using.

What if the instrument was in tolerance when found?

Then as-found and as-left are effectively the same reading and no adjustment was needed, which is the good outcome. A history of in-tolerance as-found readings is also the evidence that lets you safely extend the calibration interval, turning a fixed-schedule cost into a risk-based program.

What is guard-banding in calibration?

Guard-banding accepts an instrument only if its as-found reading sits inside the tolerance by a margin that covers the calibration uncertainty. Because every measurement carries uncertainty, the guard band reduces the chance of passing a marginal gauge that is actually out of tolerance. It is the same measurement-uncertainty thinking behind gage R&R and MSA.

What do auditors look for in calibration records?

Both as-found and as-left readings on the certificate, a defined reaction when as-found is out of tolerance (nonconformance, impact assessment, disposition), and metrological traceability, an unbroken calibration chain back to national standards such as NIST and ultimately to SI units. Accredited ISO/IEC 17025 certificates also state measurement uncertainty.

As-Found vs. As-Left in Calibration Records

Q: Why does an out-of-tolerance as-found reading matter?

Because it is the only evidence of how the instrument behaved during the interval just ended. If it drifted out of tolerance, every part measured with it since the last good calibration is suspect. The reading should be treated as a nonconformance that triggers an impact assessment and reverse traceability, not filed as routine.

As-left is the comfortable number. As-found is the only witness to what your gauge did all quarter.

As-found is the reading a calibration takes before any adjustment; as-left is the reading after. As-found tells you whether every measurement made with that instrument since its last calibration can be trusted; as-left confirms the instrument was returned to tolerance. Both belong on the certificate, because they answer different questions: what happened, and what was fixed.

The two readings look like a bookkeeping detail and are anything but. As-left is the comfortable number, the instrument back inside its limits, ready to work. As-found is the uncomfortable one, the only evidence you will ever have of how the instrument actually behaved during the interval that just ended. When it comes back out of tolerance, it does not just fail a gauge; it puts every part that gauge measured over the last months under suspicion.

What is the difference between as-found and as-left?

As-found captures the instrument's condition on arrival, in the state it was used on the floor. As-left captures its condition on departure, after any adjustment, repair, or none. If the instrument arrives in tolerance and needs no adjustment, as-found and as-left are effectively the same reading, which is the outcome you want and the one that lets you lengthen calibration intervals with confidence.

	As-found	As-left
When taken	Before any adjustment, on arrival	After adjustment or repair (or confirming none was needed)
Question it answers	Were my past measurements valid?	Is the instrument fit to use going forward?
What it protects	Product already made and shipped	Product about to be made
If out of tolerance	Triggers impact assessment and reverse traceability	Instrument stays out of service until brought back in or replaced
Direction it looks	Backward, over the interval just ended	Forward, into the next interval

The reason both are mandatory in a serious calibration program is that dropping either one hides a risk. Record only as-left and you can never reconstruct whether last quarter's shipments were measured by a drifted gauge. Record only as-found and you cannot prove the instrument is safe to keep using. Under ISO/IEC 17025, accredited calibration certificates report results before and after adjustment when that data is available, precisely so the customer can make both judgments.

The as-found reading is the only witness to the interval. If it fails, everything measured since the last good cal is in scope.

Why does an out-of-tolerance as-found reading matter so much?

Because a drifted gauge does not announce itself. If a caliper reads 0.03 mm high and only the as-left is recorded, every part it passed during the interval was potentially measured wrong, and nobody knows. The as-found reading is the trigger that turns that invisible risk into an action. Treat an out-of-tolerance as-found not as a filing task but as a potential nonconformance: it opens an impact assessment on the product measured since the last known-good calibration.

That backward look is reverse traceability. You identify what the instrument measured over the interval, decide how the observed error affects conformance, and disposition accordingly, which may mean quarantine, re-measurement, recall, or a documented rationale that the error is too small to matter for those characteristics. The direction of the error matters: a gauge reading high may have passed parts that were actually oversize, while reading low may have rejected good parts. Both are problems, and both need the paper trail that reverse traceability builds.

How far back does reverse traceability reach?

To the last calibration where the instrument was confirmed in tolerance, because that is the last moment you have evidence it was reading true. Everything between that point and the failed as-found reading is inside the suspect window. This is why calibration records are not archival clutter: the as-found history is what defines the boundaries of any recall, and a missing prior certificate can widen the suspect window from one interval to "we are not sure."

Record the failed as-found reading as it arrived, with the magnitude and direction of the error, before any adjustment touches the instrument.
Open a nonconformance and freeze the instrument's use until it is confirmed back in tolerance or replaced.
Define the suspect window: from the last known-good calibration to now, and list what the instrument measured in that span.
Assess impact per characteristic comparing the error against the tolerances of the features it checked; a large gauge error on a tight tolerance is serious, a small error on a loose one may not be.
Disposition the affected product: quarantine, re-measure, recall, or accept with a documented technical rationale, and record the decision.
Feed the result back into the interval: a gauge that drifts out repeatedly needs a shorter calibration interval or replacement, tracked over time.

Where do as-found and as-left fit in the wider quality system?

A calibration program is one node in a chain of trust that runs from your reference standards down to the accept/reject decision on a part. Every gauge on the floor inherits its credibility from calibration, and every downstream tool inherits it from the gauge. When a control chart flags a shift, or a statistical process control study reports a capability number, both are only as honest as the measurement system feeding them. An undetected gauge drift shows up as phantom process variation, sending engineers chasing a machine problem that is really a metrology problem.

The chain reaches outside your walls too. When you require calibration certificates from a supplier you are extending the same as-found/as-left logic to their gauges: their out-of-tolerance events can invalidate the inspection data they ship with a lot. This is why a supplier's calibration recall notice is not junk mail; it can define a suspect window in your own incoming material. Treating calibration as a system, not a sticker, is what keeps a single drifted instrument from quietly poisoning decisions three steps downstream.

What do auditors expect to see?

Auditors, whether for ISO 9001, IATF 16949 or aerospace, look for the same three things: both readings on the certificate, a defined reaction when as-found is out of tolerance, and metrological traceability of the calibration itself. The traceability requirement means the standard used to calibrate your instrument was itself calibrated against a higher reference, in an unbroken chain back to national standards such as those held by NIST and ultimately to SI units. An accredited ISO/IEC 17025 certificate carries that chain plus stated measurement uncertainty, which is why it is worth more than an unaccredited "calibration sticker."

As-found sits outside the band; adjustment returns the as-left inside it. The certificate records both, not just the happy ending.

Auditors also probe the reaction time. When an external calibration lab finds one of your instruments out of tolerance, they issue an out-of-tolerance notification, and the clock starts. A finding they will write up is a stack of those notifications with no matching impact assessments, evidence that the as-found data was received and ignored. What they want to see is a closed loop: notification in, nonconformance opened, suspect window defined, product dispositioned, record filed. The presence of the as-found reading is worth nothing if nothing happens when it is bad.

One nuance auditors reward: guardbanding. Because every measurement carries uncertainty, some programs apply a guard band, accepting an instrument only if its as-found reading is inside the tolerance by a margin that covers the calibration uncertainty. It reduces the odds of calling a marginal gauge "good" when it is actually out. This is the same measurement-uncertainty thinking that underpins measurement systems analysis and gage R&R: a gauge you cannot trust silently corrupts every downstream decision, from first article inspection to the control charts running on the floor.

How do as-found records improve the program over time?

The as-found history is a data set, not just a compliance artifact. Instruments that keep coming back in tolerance can safely have their calibration intervals extended, cutting cost; instruments that drift out repeatedly need shorter intervals, environmental fixes, or replacement. Tracking as-found trends turns calibration from a fixed-schedule expense into a risk-based program, which is exactly what standards like IATF 16949 clause 7.1.5 and its equivalents are pushing toward. The standards facts worth pinning down:

Under ISO/IEC 17025:2017, calibration certificates report results before and after any adjustment or repair when that data is available, i.e., both as-found and as-left (ISO/IEC 17025:2017).
Metrological traceability requires an unbroken, documented chain of calibrations linking a result to a reference such as a national metrology institute and ultimately to SI units (NIST, Metrology).
Measurement decisions should account for measurement uncertainty, the basis for guard-banding conformance decisions on calibration and inspection (NIST, Measurement Uncertainty).

The hard part of an as-found/as-left program is never the concept; it is keeping certificates, out-of-tolerance notifications, and impact assessments linked so that when a gauge fails, you can reconstruct the suspect window in minutes instead of days. That reconstruction is scattered-paperwork work, exactly what Harmony's paperwork digitization and AI search pulls into one searchable layer alongside your QMS. See it running in a plant like yours in the CLS case study. However the records are stored, the rule holds: record both readings, react to the bad one, and never let the comfortable as-left number bury the as-found story.