Calibration intervals & drift: fixed schedules vs predictive

“How often should we calibrate this?” is one of the most expensive questions a lab answers — too often wastes money, too rarely risks shipping bad measurements. The best answer comes from the instrument’s own history.

Why intervals matter

An instrument drifts between calibrations. Set the interval too long and it may spend months out of tolerance, quietly invalidating everything it measured. Set it too short and you pay for calibrations — and downtime — you didn’t need. The interval is a risk/cost trade-off, and it deserves more than a default.

Fixed intervals: the starting point

Most labs begin with fixed intervals — often 12 months — from the manufacturer’s recommendation or accreditation guidance. It is simple and defensible to start, but it treats a stable reference and a hard-used field instrument exactly the same, which is rarely right.

Risk-based intervals

A better approach weights the interval by consequence and exposure: how critical the measurement is, how heavily the instrument is used, how harsh its environment, and its history of passing or failing. A critical gauge used daily earns a shorter interval than a reference that lives in a controlled cabinet.

Predictive intervals from drift data

The most powerful input is something you already collect: as-found drift at each calibration. Plot an instrument’s error over successive calibrations and a trend appears — you can see how fast it approaches its tolerance limit and project when it will cross. That projection, not a calendar default, is the evidence-based interval.

Fixed schedules ask “how long since last time?” Predictive intervals ask “how long until it drifts out?” Only the second is about risk.

Adjusting with confidence

Accreditation bodies expect intervals to be reviewed and justified — and they generally welcome lengthening an interval when drift data supports it, or shortening it when an instrument misbehaves. The key is the evidence: a documented history that backs the decision.

Make the data work for you

This only works if drift history is captured consistently and analysed automatically. When the software that records each calibration also models the drift and recommends the next interval, optimisation stops being a spreadsheet project and becomes a standing feature — safer for the lab and cheaper over time.

More from the Cali blog: browse all calibration & metrology guides →