← People analytics

People analytics · technique

Trend analysis: distinguish change from noise

Trend analysis examines how a measure changes over time.

Statistical and evaluation practice; ADM HR worked application · 2026 teaching guide · Original editorial explanation
Trend analysis: distinguish change from noise — simplified model sketchObserve over time · Check seasonality · Investigate change. Editorial interpretation, after Statistical and evaluation practice; ADM HR worked application (2026 teaching guide).Time → · schematic only, not observed dataADM HR · ORIGINAL INTERPRETATIONStatistical and evaluation practice; ADM HR worked application · 2026 teaching guide
Original simplified sketch after Statistical and evaluation practice; ADM HR worked application (2026 teaching guide). Read the explanation for assumptions and limitations. Download SVG ↗

The core idea

Trend analysis examines how a measure changes over time. A time-ordered plot helps reveal shifts, unusual observations and patterns that a single period hides. In people analytics, the definition and denominator must remain comparable. A rising count may reflect workforce growth rather than a higher rate, and a seasonal pattern should not be mistaken for a lasting deterioration.

Source and attribution [1]

Using it in practice

Choose an appropriate interval and show enough history to understand variation. Mark known process changes and breaks in data collection. Plot counts and rates when both matter. Distinguish a descriptive chart from a forecast, and avoid drawing a confident conclusion from one unusual month. Ask whether the apparent pattern survives a change in the display scale or aggregation period.

An example, not a reported case

Worked example · illustrative

A fictional employer records 20 leavers from an average workforce of 200 in one year and 24 from 300 in the next. The count rises, but the comparable annual turnover rate falls from 10% to 8%. A monthly chart also shows recurring departures at the end of seasonal contracts. The review separates those events from voluntary departures before proposing a retention intervention.

What to watch for

Rolling averages can smooth noise but also delay or obscure a real change. Different periods may have different exposure and composition. A chart cannot establish why a shift happened. Do not compare a partial month with a completed month without an explicit adjustment and explanation.

Keep the series reproducible

Record the metric definition, extraction date and any revisions. If a system change alters what is recorded, mark a break rather than pretending the whole series is uniform. Retain the underlying counts so a rate can be checked.

Choose a review trigger

Agree what pattern merits investigation before repeatedly scanning for an alarming result. Use appropriate statistical advice for formal control limits or time-series models. A useful chart supports a proportionate question, not a reflex response to every fluctuation.

Make the series comparable over time

Check whether definitions, systems, organisational boundaries or reporting practices changed. Mark those changes on the timeline rather than hiding them in a note nobody reads. A sudden rise in recorded absence might follow better recording, while a falling count might reflect a smaller workforce. Choose a rate where appropriate and show its denominator. Do not join incompatible periods with a smooth line that suggests an uninterrupted measurement process.

Distinguish a count from a rate

In an illustrative example, ten exits from an average workforce of 100 gives a 10% period turnover rate; twelve exits from an average workforce of 200 gives 6%. The count rose while the rate fell. Both facts may matter. Use a consistent definition of average workforce and period, and avoid casually annualising a short, volatile interval. Explain what decision the trend is intended to inform.

Investigate a change before calling it a trend

Look for seasonality, one-off events, changes in workforce mix and delayed data entry. A few points do not establish a durable direction. Smoothing can help show a pattern but may conceal a sudden shift, so keep the underlying data available. If using formal process-control or time-series methods, use their assumptions and appropriate expertise rather than adding arbitrary warning bands to make a chart appear statistical.

Connect the chart to a review question

Ask what evidence would distinguish the leading explanations and what action is proportionate. A rising absence rate might prompt a workload review, while an isolated administrative spike might need a data correction. Record the interpretation as provisional where necessary. After an intervention, continue observing long enough to judge whether movement persists. A line turning down after a launch is not by itself proof that the launch caused the change.

Check whether the measurement changed before explaining the trend

A rise in recorded incidents may reflect a new reporting system, a wider definition or improved willingness to report. Mark such changes on the time series and keep the observation period consistent. Use rates where exposure changes, such as incidents per relevant hours, and retain the underlying counts. A percentage without a denominator can conceal a large change in the population or in the amount of work performed.

Distinguish signal from normal variation

Look for sustained patterns, seasonality and unusual events rather than reacting to every monthly movement. Compare like periods where seasonal demand is important and avoid choosing a starting point that exaggerates the story. Explain what the trend can and cannot show. If an intervention begins during an unusually bad month, subsequent improvement may partly reflect a return towards normal conditions. Use a suitable comparison and operational context before attributing the movement to the intervention.

Take it into your next conversation

Three useful questions.

  1. What question can this method answer, and what can it not establish?
  2. Are the comparison, observation period and assumptions defensible?
  3. What decision follows, and how will its consequences be reviewed?

Go to the evidence

Sources & attribution

Further reading: Nadeem Khan and Dave Millner (2023), Introduction to People Analytics: A practical guide to data-driven HR, 2nd edition, PDF pp. 227, 244 and 294–295. Edition & reading notes ↗

[1] NIST: Run-sequence plots ↗

The core idea is an original summary of the cited work. Application notes, examples and sketches are our interpretations, not quotations or reproductions of the authors’ figures. Publisher records may require access to read the full original work.

Explore People Analytics →

Application guidance expanded on 26 September 2026. These practical suggestions and invented examples are editorial explanations; further-reading sources are not the original authors of the historical models unless expressly identified. Original overview review dates are retained.

Published 2026-09-20 · Reviewed 2026-09-22. Editorial approach

Keep exploring

A little more context.

People analytics · practical playbook01

From a business question to a measured outcome

A complete worked route through casual-worker retention, from definitions to a fair pilot and an honest conclusion.

ADM HR editorial workflow6 min
People analytics · models & decisions02

Logic models for HR interventions

Separate resources, activities and deliverables from the changes an initiative is meant to produce.

Programme-evaluation tradition6 min
People analytics · technique03

Cohort analysis: compare like starting points

A cohort is a group defined by a shared starting event or period, such as joining in a particular quarter.

Statistical and evaluation practice5 min