The useful takeaway
If two people mean different things by the same metric, settle the definition before comparing the result.
Use definitions as a reporting agreement.
This glossary explains terms for product and reporting conversations. It does not replace a source’s own metric documentation or the terms of an engagement. When a platform uses a specific definition, keep that definition attached to the data.
Before sharing a report, identify the words most likely to be interpreted differently. Write the intended meaning beside the measure, including the source, observation window and denominator where relevant. That small step can prevent a confident but misleading comparison.
A–F / Context & identity.
Start with what is being compared and which records belong together.
Attribution
The stated connection between an outcome and a preceding activity. A tagged visit can support journey reporting; it does not by itself establish that the activity caused a purchase.
Baseline
The reference used for comparison. Specify the account, format, time window and sample so another person can understand what changed.
Coverage
The sources, entities, formats, fields and time periods included in a report. A complete record for one post does not prove complete coverage of an account.
Entity resolution
Connecting records that refer to the same athlete, team, event or other subject. A shared name is not sufficient evidence on its own; ambiguous matches need review.
Freshness
How recent the underlying information is for the intended decision. Publication time, observation time and refresh time answer different questions.
I–O / Interpretation & observation.
Distinguish the activity a source records from the behavior an analyst infers.
Impressions
Source-defined instances of content being displayed. Repeated displays can be counted; impressions should not be relabeled as unique people.
Intent
An interpretation of likely future behavior. It requires a defined method and supporting evidence; a reaction or like does not establish a purchase decision.
Median
The middle value in an ordered sample; for an even sample, the average of the two middle values. It can reduce the influence of a single very large result.
Observation window
The period during which measurements are collected. Comparable windows help separate a content difference from a timing difference.
P–S / Evidence & reporting.
Keep the original source and the boundaries of the sample close to the result.
Provenance
The source and context needed to trace a piece of information. For a report, this may include the original link, relevant times and how a claim was derived.
Reach
Distinct accounts or people under a source’s definition. Do not estimate unique reach by adding views, impressions or separate platform audiences.
Recorded interactions
The supplied action counts included in a report, such as likes or comments. State the categories and distinguish a partial total from a complete one.
Sample share
A record’s proportion of a defined sample total. It is useful within that sample and should not be presented as league-wide or market-wide share.
Sentiment
An interpretation of expressed tone or attitude. Sarcasm, context and sample selection affect the result; sentiment is not automatically loyalty or intent.
Source rights
The conditions governing an intended use of source material. Access, retention, display and redistribution can have different requirements.
V / Change over time.
Repeated comparable observations let you discuss movement. They do not remove the need to account for source delays, corrections or changing definitions.
Velocity
Change over a specified interval. State what changed and when; a lifetime average is not a current rate, and a single interval does not establish acceleration.
Put the language into practice.
Try reading a report without its headline. Can you identify the sample, the source of each measure and the decision being proposed? If a term could mean two different things, ask the author to state the definition before interpreting the result.
Use the Measurement Guide for worked calculations and Sample Briefings to see how definitions support a practical recommendation. For an evaluation, agree on this vocabulary with both the business owner and the person preparing the data.
References & further reading
Examples are illustrative unless explicitly identified as source observations. Availability depends on approved sources and customer configuration.