Checklist

Predictive Maintenance Data Readiness Checklist

A practical, interactive checklist to evaluate asset inventories, sensors, data quality, labeling, event synchronization, governance, and pilot sizing before launching a predictive maintenance or condition-based monitoring pilot.

Interactive Tool

Predictive Maintenance Data Readiness Checklist

This interactive checklist helps maintenance and operations teams quickly assess whether they have the right assets, sensors, labeled events, governance, and pilot parameters to run a credible predictive maintenance (PM) or condition-based monitoring (CBM) pilot. Use it to capture a simple readiness score, record evidence, and create next steps before you invest in model development.

Each item includes helpful examples and pragmatic guidance — adapt thresholds and targets to your site, asset criticality, and risk tolerance.

Includes canonical asset IDs, location, system/hierarchy, equipment type, BOM link, maintenance history, and a documented criticality ranking (safety/production/cost impact).
List where the inventory is stored, examples of asset IDs, and links to CMMS or asset master. (Optional)
Confirm which sensor types exist (vibration, temperature, current, pressure, acoustic, oil analysis, encoder), signal paths, tag names, and whether raw time-series are stored. Note sample rates per sensor type.
Examples: vibration 6–12 kHz for bearing/frequency analysis, 1 Hz+ for slow-moving temps, 10–100 Hz for rotating equipment signatures. Note any sensors with insufficient sampling or intermittent capture.
Timestamps (UTC or documented timezone), column schemas, raw vs processed data storage, file formats (CSV/Parquet), retention periods, and where data engineers can access it (data lake, historian, SCADA).
Provide dataset paths, owners, sample file names, and any authentication/roles needed. (Optional)
Can sensor timestamps be aligned with CMMS/SCADA events and operator logs? Check clock sync, timezone handling, and whether events include clear start/stop timestamps and reasons.
Supervised models need labeled failure events or annotated degradation windows. If labels don't exist, have you defined label definitions, annotators, and a quality-review process?
Describe available labeled events (type, date range) and approximate counts. Examples: 120 bearing failures 2018–2024; 25 unplanned gearbox replacements. If none exist, estimate how many events you can realistically label in the pilot.
Document the common failure modes you care about and what constitutes a true positive vs a nuisance alert. This helps test and evaluate models and reduces false positives.
Confirm ownership, who can see model outputs, encryption at rest/in transit, data retention policy, and any regulatory or contractual constraints on data use.
Choose a clear metric you will measure during the pilot (pick one primary).
Example: 'Reduce downtime by 20% across pilot assets' or 'precision >= 0.6 and false positive rate <= 0.2'.
Guidance: choose a small, representative group (5–20 homogeneous assets) rather than a single asset. For heterogeneous fleets, run parallel smaller pilots by asset class.
Typical pilots run 8–16 weeks to gather useful run-to-failure or degradation windows; longer if time-between-failures is long.
Rule-of-thumb: 50–200 events per failure mode improves chances for basic supervised models; fewer events may require unsupervised or physics-based approaches.
Rate 1 (poor) to 5 (excellent) across completeness, consistency, timestamp accuracy, and signal-to-noise ratio.
1.0 10.0
Rate 0 (not ready) to 10 (ready). Use this as a quick summary after completing the checklist.
1.0 10.0
E.g., 'Collect additional labeled events', 'Increase sample rate on vibration sensors', 'Implement clock sync between PLC and CMMS', or 'Start a 10-asset pilot for 12 weeks'.
You can explore this tool now. Sign in or create an account to save your responses and return to them later.
Sign In to Save Create an Account
Make this tool part of your work

Save a personal copy, bring it to your team, or tailor the questions and workflow to fit what you are hungry to improve.

Member customization and team collaboration are coming soon.
Discuss Ask the Librarian

Discussion

Comments and conversation will live here.