Sensor Coverage Is Not Data Quality

Connectivity is only the beginning

A dashboard showing thousands of connected points can create confidence that the building is understood. It may instead show that many values can be transported. The difficult questions are whether each value represents the intended asset, uses the correct unit, has a reliable timestamp and remains meaningful in the operating mode being analysed. Coverage and trust are separate properties.

Hong and Li identify dataset documentation as foundational to reproducible building-AI work, including sources, building characteristics and measurement specifications [HONG]. JLL also describes fragmented real-estate data as an implementation concern [JLL]. These are reasons to establish data quality before using scale as a selling point.

Give every critical point a usable identity

For each point that supports an operational decision, record the asset, physical location, measured quantity, unit, interval, source system and accountable owner. Add whether it is a measured value, a command, a calculated quantity or a manual entry. A command to open a damper does not establish that the damper actually moved.

Maintain the relationships among points. Supply temperature without the relevant operating schedule, equipment state or outdoor condition may be difficult to interpret. Asset hierarchy also matters: duplicate labels across floors or buildings can cause an apparently coherent model to combine unrelated equipment.

Figure 1. Four checks before accepting a performance claim Original Osmos Global framework informed by Hong and Li (2026), §§3.1–3.5, CC BY 4.0. No source diagram reproduced. Prepared 1 September 2026.

What this means: A strong algorithm cannot compensate for an invalid comparison.

Treat missingness as information

A gap in a time series can reflect communication loss, maintenance, controller replacement or a system being intentionally disabled. Filling all gaps automatically may conceal the very event engineers need to investigate.

Record the missing-data rule and preserve an indicator showing which values were observed and which were estimated.

Osmos recommends quality flags that are intelligible to users: current and validated, stale, missing, substituted or under investigation. These flags should travel into downstream charts and recommendations.

A confidence badge that disappears when data is exported cannot protect the next decision-maker.

Test the physical and digital chain together

A commissioning check should follow selected points from the physical sensor to the controller, integration layer, analytics calculation and user display. Confirm timestamps and units at each stage. Where the system will influence consequential action, testing should include known changes and a documented response under safe, authorised conditions.

Quality also requires maintenance. Sensor replacement, firmware changes, tenant alterations and revised schedules can invalidate an earlier mapping. The data owner should receive relevant change notifications and periodically review whether the dictionary still reflects the building. Otherwise, confidence quietly degrades while the connection count remains stable.

Illustrative decision rehearsal

An illustrative example is a temperature point that appears unusually stable. The flat trend might indicate good control, a failed sensor, a frozen communication value or a rounded display. An analytics model can confidently interpret the wrong explanation if the data pipeline does not expose status and freshness. The visual smoothness of the chart is not evidence of sensor reliability.

The operating team should compare the value with an appropriate independent reference under approved conditions, inspect timestamps and confirm whether the source provides a quality or communication flag. If the record is suspect, restrict its use while the cause is investigated. Do not silently replace it with a neighbouring sensor and continue presenting it as the original measurement.

A practical first-cycle audit can select the points that drive the highest-consequence recommendations rather than attempting to inspect every sensor at once. Trace those points end to end, record the known uncertainty and agree what happens when they become stale. This prioritisation makes data stewardship manageable while protecting the most important decisions.

Report the resulting findings in ordinary language: which decisions have trusted inputs, which rely on substitutions and which should remain manual. The audit should leave an owner and a correction route for each issue. It is complete when users know the limits of the information, not merely when a spreadsheet of point names has been filled in.

Use a decision-specific quality threshold

Not every point requires identical scrutiny. A broad portfolio trend can tolerate different uncertainty from a control decision affecting a critical environment. Define what quality is sufficient for the particular purpose, and restrict uses that exceed that threshold. This is more efficient than either demanding perfection everywhere or accepting all connected data as equivalent.

The sources do not supply a universal completeness or sensor-accuracy threshold for every building.

Engineers must select thresholds appropriate to the system and consequence. The management lesson is to report usable, trusted coverage separately from raw connected coverage. A smaller dependable dataset can support a stronger decision than a larger unexamined one.

Source notes

[HONG] Tianzhen Hong and Han Li. Good practices for documenting AI-based studies on energy and buildings. Energy & Buildings / Elsevier; author copy hosted by Lawrence Berkeley National Laboratory, 2026-01-20. Sections 2, 3.1–3.6 and 4; pp. 1–4. DOI: 10.1016/j.enbuild.2026.117043. Accessed 1 September 2026. https://eta-publications.lbl.gov/sites/default/files/2026-06/1-s2.0-s0378778826001039-main.pdf [JLL] Yuehan Wang. Reality check: The true pace and payoffs of AI adoption in corporate real estate. JLL, 2025-10-27. Key highlights; AI pilot selection; Lessons learned. Accessed 1 September 2026. https://www.jll.com/en-hk/insights/global-real-estate-cre-technology-survey

Editorial and visual note

This is original Osmos Global analysis informed by the cited publications. Reported findings are distinguished from Osmos recommendations and illustrative scenarios. Source findings and trademarks remain attributable to their owners. Original visual designs do not imply endorsement by source organisations. The content is general research and does not replace site-specific professional advice.