A trend can show that a pressure spiked at 3:14 in the afternoon, but it cannot tell you the spike was a scheduled bypass test rather than a real upset. That context lives in people's heads and is usually lost by the time anyone looks back at the data. A historian annotation is the mechanism for capturing it: a text note attached to a specific value or moment on a tag, so the explanation travels with the data forever. This page explains what an annotation is, why it never alters the value it describes and is protected from being compressed away, and how these small notes make later analysis far more reliable.
Historian annotation in one line: A historian annotation is a text note attached to a specific tag value or a specific point in time on a trend, adding context such as an explanation of an event, without changing the underlying value. The annotation travels with the data, is typically recorded with who wrote it and when, and is protected so that compression never removes it. Its purpose is to preserve the human explanation of what the data means, so that someone analyzing the trend later understands why a value looks the way it does.
The essential property of an annotation is that it is separate from the value it describes. When an operator annotates a pressure reading with spike due to bypass test, the pressure value itself is untouched; the archive still holds the real measurement, and the note sits alongside it as metadata anchored to that timestamp and tag. This separation is deliberate and important. The data must remain a faithful record of what the instruments measured, so that trends, calculations, and audits reflect reality, while the annotation adds the interpretation a human brings. Mixing the two, by editing a value to look normal, would destroy the record; annotating it preserves both the fact and the explanation.
Annotations are typically anchored to a precise point, a particular sample or timestamp on a particular tag, so that when the trend is drawn, a marker appears exactly where the note applies. Some systems also allow annotations over a span of time or attached to an event rather than a single sample, which suits explanations that cover a period, such as unit in startup or instrument under calibration from this time to that. Either way, the anchoring is what makes the note useful, because it appears in context on the trend rather than in a disconnected log that no one thinks to check.
Because they carry human judgment, annotations usually record their own provenance: who wrote the note and when it was added, and often they cannot be silently altered later. That makes an annotation a small but accountable statement about the data. When several people work the same asset over years, that provenance is what lets a later reader weigh the note, knowing who observed the event and when, rather than treating an anonymous comment as unquestionable fact.
Historians compress data to save space, and much of that compression works by discarding samples that do not add information, keeping only the points needed to reconstruct the signal within a tolerance. Annotations must be exempt from that logic. A note is not a redundant sample that can be inferred from its neighbors; it is unique information that exists nowhere else, so if compression removed the sample an annotation was attached to, the explanation would be orphaned or lost. For that reason historians protect annotated points and the annotations themselves from being compressed away, ensuring the note survives as long as the data does.
Auditing annotations matters for the same reason it matters for manual data. An annotation can influence how a later investigation interprets an event, so knowing who wrote it, when, and whether it was ever edited keeps the note trustworthy. If annotations could be quietly changed or deleted, someone could rewrite the story of an incident after the fact, which is exactly what a good record should prevent. Treating annotations as auditable, append-oriented entries, where corrections are recorded rather than overwriting the original, keeps the explanatory layer as reliable as the data layer.
This protection also affects retention. Because annotations are sparse and small compared to the raw stream, keeping them indefinitely costs little, and they gain value over time as the memory of an event fades from the people who witnessed it. A historian that ages out old raw detail but retains annotations, or that ensures annotated periods are preserved, keeps the most human and least recoverable part of the record intact. The value that produced the annotation might be summarized, but the note that says what it meant is the part no aggregate can reconstruct.
For analysis, annotations are what turn a wall of numbers back into a story. Months later, an engineer investigating a recurring excursion can see at a glance which past spikes were tests, which were genuine upsets, and which coincided with a known maintenance activity, without having to track down whoever was on shift. That context dramatically changes the conclusions drawn from a trend, because an unexplained pattern and a fully explained one call for very different responses. Annotations let the data carry its own footnotes, so the analysis starts from what actually happened rather than from guesses.
In remote and cloud monitoring, annotations are especially valuable because the people looking at the data are often not the people who were on site. A central team watching many assets sees trends but lacks the local knowledge that a plant operator has. When operators annotate events as they happen, that local knowledge is captured and travels to the remote viewers through the shared history, closing the gap between who saw the event and who later analyzes it. An annotation added at the site becomes visible on the same trend a distant engineer opens, so context is not lost to distance.
The practical habit that makes annotations pay off is annotating in the moment, when the explanation is known, rather than intending to add it later. A note written while the event is fresh is accurate and specific; one reconstructed weeks afterward is vague or never gets written at all. Encouraging operators to attach a short, honest note to notable events, and protecting those notes from being edited away or compressed out, builds up an explanatory layer over the years that becomes one of the most useful assets in the historian, precisely because it holds the information no sensor could record.
No. An annotation is a separate note anchored to a value or timestamp, and the underlying value is left untouched. The archive still holds the real measurement, while the annotation adds human context beside it. This separation is deliberate, because the data must remain a faithful record of what the instruments measured, and the annotation supplies the interpretation without altering the fact.
Compression works by discarding samples that can be inferred from their neighbors, but an annotation is unique information that exists nowhere else, so it cannot be reconstructed if removed. If the sample an annotation is attached to were compressed away, the explanation would be orphaned or lost. Historians therefore exempt annotated points and the annotations themselves from compression so the note survives as long as the data.
Because the annotation is stored with the data in the shared historian, anyone who can view that trend sees the note in context, including a central or cloud-based monitoring team that was not on site. This is one of the main benefits of annotations in remote operations, since the local knowledge captured by an on-site operator travels to distant viewers through the same history, closing the gap between who witnessed the event and who later analyzes it.
Merobix reads your field devices into a cloud SCADA - the real thing behind these terms, live in days from any browser.