How do you calculate OT data completeness?

Multiply tags per cycle by planned cycles to get gross samples, then multiply by the delivery rate and the in-spec quality rate. For 500 tags x 60 cycles = 30,000 gross, x 0.995 x 0.98 = 29,253 complete samples.

What is a good OT data completeness rate?

Analytics-grade historian feeds typically target 98% or better usable samples. In the example, 29,253 of 30,000 is about 97.5% complete, just below a strict analytics threshold, driven mostly by the 98% quality rate.

What is the difference between delivery loss and quality loss?

Delivery loss is samples that never arrived: dropped polls, timeouts, link failures. Quality loss is samples that arrived but are bad-quality or out of range. Here delivery costs 150 samples and quality costs 597, so quality is the bigger problem.

Why does a 99.5% delivery rate still lose samples?

Across volume, small percentages add up. At 30,000 gross samples, even 0.5% delivery loss is 150 missing tags, and the 2% quality loss removes another 597, leaving 29,253.

Should I worry more about delivery or quality losses?

Look at where the bigger count lands. In this case quality loss (597) is four times the delivery loss (150), so effort is better spent on sensor calibration and range checks than on network reliability.

How does completeness affect analytics and ML?

Models and KPIs assume the data they see is representative. Missing or bad-quality samples bias results, especially when losses cluster during the very upsets you most want to analyze. Always qualify a dataset's completeness before trusting its conclusions.

IIoT, SCADA & Edge Connectivity calculator

OT Data Completeness Calculator

OT data completeness estimates how many tag samples actually land in your historian or IIoT platform in-spec, after accounting for delivery losses (dropped polls, timeouts) and quality losses (bad-quality or out-of-range values). Data engineers, historian administrators, and analytics teams use it to know how much of their planned data they can trust before feeding it to dashboards, KPIs, or machine-learning models. It matters because analytics built on incomplete OT data quietly produce wrong answers; a 99.5% delivery rate sounds excellent until you see how many samples it costs across thousands of tags and cycles. The calculator converts polling design and two reliability rates into a concrete count of usable samples.

What this calculator does

Estimate the count of OT tags landing complete in the historian per period from tags published per polling cycle, planned cycles in the period, the share of polls that arrive (delivery rate), and the share of arriving polls that are within quality bounds.
Use it when an analytics lead, MES owner, or historian admin needs a clean count of usable tag samples in a period before signing off on a model rebuild or KPI dashboard.
It computes the number of complete, in-spec tag samples delivered in a period from gross polling volume, after applying delivery and quality rates.

Formula used

Gross tag samples = tags per cycle × planned cycles
Complete tag samples = gross samples × delivery rate × in-spec quality rate

Inputs explained

Tags published per polling cycle:
Planned polling cycles in the period:
Tag delivery rate:
In-spec quality rate:

How to use the result

Use it when validating historian coverage, sizing data pipelines, or qualifying a dataset for analytics or ML.
It treats delivery and quality as independent flat rates; in reality, losses often cluster during network outages or sensor faults, so the same total loss can hit far worse if it is bursty rather than spread evenly.

Current U.S. benchmarks

Global copper trades at $13,484 per tonne (IMF via FRED, May 2026), up 41.5% in a year, and U.S. industrial electricity averages 8.66 cents per kWh. Both feed electrified-hardware unit economics.

Common questions

How do you calculate OT data completeness? Multiply tags per cycle by planned cycles to get gross samples, then multiply by the delivery rate and the in-spec quality rate. For 500 tags x 60 cycles = 30,000 gross, x 0.995 x 0.98 = 29,253 complete samples.
What is a good OT data completeness rate? Analytics-grade historian feeds typically target 98% or better usable samples. In the example, 29,253 of 30,000 is about 97.5% complete, just below a strict analytics threshold, driven mostly by the 98% quality rate.
What is the difference between delivery loss and quality loss? Delivery loss is samples that never arrived: dropped polls, timeouts, link failures. Quality loss is samples that arrived but are bad-quality or out of range. Here delivery costs 150 samples and quality costs 597, so quality is the bigger problem.
Why does a 99.5% delivery rate still lose samples? Across volume, small percentages add up. At 30,000 gross samples, even 0.5% delivery loss is 150 missing tags, and the 2% quality loss removes another 597, leaving 29,253.
Should I worry more about delivery or quality losses? Look at where the bigger count lands. In this case quality loss (597) is four times the delivery loss (150), so effort is better spent on sensor calibration and range checks than on network reliability.

Last reviewed 2026-05-12.