Attendance data and privacy • 4 min read

How to analyze badge-swipe office attendance data

Build a privacy-conscious badge-data analysis with a documented purpose, denominator, correction route, access controls, retention period, and aggregate outputs.

Quick answer

Begin with the decision, not the access log

State the operational question, confirm that badge events are suitable and necessary, define one qualifying-day rule, remove known data-quality failures, provide corrections, aggregate where possible, and delete identifiable detail on a stated schedule.

An entry event is not a full workdayTailgating, missing exits, shared entrances, off-site work, visitors, forgotten badges, and system outages can all change what the log means.

A practical framework

Work through the decision in a visible order.

01

Write the measurement specification

Name the workforce group, sites, time zone, policy period, eligible workdays, qualifying event, partial-day rule, leave and exception treatment, late corrections, and output audience. Version the document whenever a rule changes.

02

Audit data quality before rates

Profile missing days, duplicate taps, impossible sequences, unmapped doors, system outages, and differences between access control and workplace definitions. Report unknowns instead of forcing every record into present or absent.

03

Calculate aggregates with context

For space planning, weekday and location distributions may be more useful than individual compliance tables. Show coverage, corrections, exclusions, and confidence limits beside each output.

04

Limit access and retention

Give identifiable records only to roles that need them for the documented purpose, log access, separate confidential exception data, and delete or de-identify raw events when the approved period ends.

05

Publish a data-quality statement with every report

State source coverage, outage dates, unknown records, corrections pending, excluded populations, aggregation level, suppression rules, and the decision the report may support. A percentage without this context invites false precision.

06

Worked example: three entry events can represent one observed day

A fictional badge file contains three entry events for the same authorized pseudonymous identifier on the same local date. Under an agreed rule counting distinct observed person-days, those events contribute one observed day. A fourth event after midnight needs the site's actual time-zone and shift rule before classification. Counting raw events as office days would overstate attendance.

A missing event is not automatically evidence of absence, and an entry does not establish hours worked or productivity. Record outages, forgotten badges, shared entrances and other known limitations. Keep a category for records that cannot be classified under the agreed rule instead of quietly assigning them a zero or an assumed full day.

07

Write a measurement specification before importing identifiable data

State the operational question, the population, the period, the local date rule and the permitted result. For space planning, a site-level aggregate may answer the question without a named employee report. Confirm that the purpose and data use are authorized under applicable rules before moving an access-control export into another workflow.

The ICO's UK guidance emphasizes matters such as necessity, proportionality and transparency, and currently says it is under review. Use it as jurisdiction-specific guidance rather than a universal compliance checklist. Get appropriate privacy and employment review for the actual setting. A website workbook cannot establish a lawful basis or authorize a new monitoring purpose.

08

Publish the limitations with the result

A useful report states its numerator and denominator, event-to-day rule, excluded records, known outages and the period covered. If eight observed days are divided by a requirement of nine, the arithmetic gives about 88.9%; it does not resolve whether the requirement, exceptions or observations are correct. Show those inputs before interpreting the percentage.

Limit access to the working data, separate any re-identification key and use the approved retention process. Provide a route to correct errors before an individual result is used. Never equate the report with productivity or make an automatic adverse decision from incomplete badge events. Keep a data-quality note beside the workbook and name the person accountable for its review.

Action checklist

Before you consider the task complete

  • Convert authorized events to the defined unit before calculating a rate.
  • Publish missing-data limitations and a correction route with the result.
  • Document purpose, lawful basis or authority, necessity, and proportionality.
  • Define population, period, qualifying event, denominator, and exclusions.
  • Validate duplicates, outages, missing events, and location mapping.
  • Publish correction, access, retention, and review controls.

Common questions

Questions this guide should answer

Does a missing badge swipe prove someone was absent?

No. Missing events can result from data or access-process limitations. State what the source can observe, keep unknown records visible and resolve corrections before using an individual result. Badge events do not establish productivity.

Can badge data prove a person worked a full day?

Usually not by themselves. They record access-system events, which may not show working time, off-site duties, breaks, exits, or data failures.

Should managers receive individual records?

Only when that access is necessary, authorized, transparent, secured, and proportionate to the stated purpose. Aggregate planning outputs are often sufficient.

How should missing badge events be treated?

Keep them unknown until a documented source or correction resolves them. Do not silently classify every missing event as absence or presence.

Can badge data be used for workplace capacity?

Potentially, when the data is suitable, necessary, quality-checked and aggregated for that purpose. Combine it with reservations, occupancy observations, unusable inventory and user reports.

How long should badge-derived records be kept?

Only for the approved period necessary for the stated purpose and applicable obligations. Record deletion or de-identification and review the schedule after purpose or system changes.

Evidence and review notes

Primary references

Sources support the factual and safety context. The guide keeps interpretation and limitations visible rather than turning a reference into a universal personal rule.

Related tools and templates

Make the next decision concrete without an account.

An optional next step

When office-day planning needs one shared workflow

HybridTrack is the iOS companion for individuals who want a clearer office-day record and better hybrid schedule planning. Workplace teams can also discuss a coordinated employer workflow.

Download HybridTrack Discuss an employer pilot Available for iPhone. The website guides and tools remain free.