Set the question and unit before reading
Begin with the communication objective, intended audience, campaign message, qualifying coverage rule, and evidence cutoff. Then decide the unit of review. Usually it is one unique article, broadcast segment, newsletter item, or creator post. A syndicated copy can stay linked to its original unless distribution itself is the question.
Only assess what the stored evidence permits. A clipped paragraph may support a message observation but not a claim about headline prominence. A screenshot may prove a product was visible but not show the full article's tone. Mark the dimension unknown when the evidence is incomplete. Complete the evidence review before judging quality.
| Prepare | Write down | Why it matters |
|---|---|---|
| Objective | The communication change sought | Quality depends on the question |
| Audience | The people the item needed to reach | Outlet fit is contextual |
| Core message | One testable sentence | Reviewers need the same reference |
| Unit | Unique item or declared distribution unit | Counts and judgments use the same base |
| Evidence | Full item, excerpt, screenshot, or other artifact | Unknowns stay visible |
Use four observable dimensions
For each item, choose one category per dimension and add a short rationale. Relevance asks how directly the editorial subject matches the campaign and intended audience. Message delivery asks whether the item accurately carries the agreed core point. Prominence asks where and how substantially the subject appears. Product presence asks what a reader can identify, without treating visibility as sales or audience response.
Do not add the categories into a single number. Direct relevance and headline prominence describe different qualities, and their importance changes with the campaign objective. Keep the words and evidence beside the category so another reviewer can challenge the interpretation.
| Dimension | Categories | Observable cue |
|---|---|---|
| Relevance | Direct, adjacent, weak, unknown | Subject and audience fit visible in the item |
| Message delivery | Core accurate, partial, absent or conflicting, unknown | Specific sentence, caption, or spoken passage |
| Prominence | Headline or opening, substantial body, passing, unknown | Placement and amount of treatment |
| Product presence | Clear in context, named only, absent, unknown | Visible or written identification in the evidence |
Record judgment so it can be reviewed
Give each row the coverage ID, reviewer, review date, evidence reference, one category per dimension, rationale, and confidence note. The rationale should point to an observable cue, such as headline wording, the second paragraph, a caption, or a timestamp. Avoid labels such as strong or excellent unless their meaning is defined for this campaign.
For higher-stakes items, ask a second person to review without seeing the first set of categories. Compare the rows, discuss differences, and preserve the original readings. Consensus can be recorded as a later field. If disagreement remains, report it. Reviewer disagreement is information about ambiguity, not an error to average away.
- Read the same evidence version.
- Choose categories independently before discussion.
- Point every rationale to a passage, visual, or timestamp.
- Use unknown when the artifact cannot support a decision.
- Keep both original reviews and any later consensus.
- Change the grid only between review rounds, with a version note.
Apply the grid to four fictional Solstice items
The fictional review covers four unique, in-scope items. COV-201 is J-014's Northline Journal feature. COV-202 is a creator post with a dated screenshot. COV-203 is a short trade digest for which only a clipping is available. COV-204 is a local roundup that names the Mistral jacket once. Three records are evidence-ready; COV-203 still needs fuller proof.
The two reviewers agree on every category except COV-203's message delivery. Reviewer A reads the phrase as a partial version of the core rainwear message. Reviewer B judges it absent because the clipping does not connect the claim to the product. The report keeps that disagreement and the incomplete-evidence flag rather than converting both opinions into a midpoint score.
| Item | Relevance | Message | Prominence | Product | Evidence |
|---|---|---|---|---|---|
| COV-201 | Direct | Core accurate | Substantial body | Clear in context | Ready |
| COV-202 | Direct | Partial | Opening caption and primary visual | Clear in context | Ready |
| COV-203 | Adjacent | Reviewer disagreement | Short paragraph | Named only | Needs proof |
| COV-204 | Weak | Absent | Passing | Named only | Ready |
Describe the pattern without manufacturing a score
Report the denominator first: four unique items in scope, three evidence-ready, one with incomplete evidence. Then describe each dimension. Relevance is direct for two items, adjacent for one, and weak for one. Message delivery has one agreed core result, one agreed partial result, one agreed absence, and one unresolved review. Those statements can be traced to rows.
The grid describes outputs and their content. It does not show what audiences noticed, understood, felt, or did. AMEC's Barcelona Principles 4.0 call for both qualitative and quantitative analysis, while the AMEC taxonomy separates outputs from out-takes, outcomes, and impact. Use a separate survey, analytics source, business record, or agreed research method when the question moves beyond the coverage itself.
| Report statement | Supported here | Still unknown |
|---|---|---|
| Two of four items are directly relevant | Four reviewed rows and stated categories | Whether target audiences noticed them |
| One item carries the core message by consensus | Rationales from both reviewers | Whether readers understood or accepted it |
| Three items have complete evidence | Evidence checklist status | Whether missing evidence can be recovered |
| No combined quality score | Dimensions remain separate | How a different objective would weight them |
Turn the reading into a bounded next action
End with a decision connected to the evidence. For the fictional set, the operator can preserve the one-sentence core message, investigate why the local roundup lost that message, and retrieve the full COV-203 item before drawing any conclusion from it. Each action names an owner, an input, and a review date. None claims that the campaign changed audience behavior.
A spreadsheet can hold the grid beside the evidence register. cutePR currently lets an operator record manual coverage, quality context, evidence, and frozen report snapshots, but it does not provide automatic monitoring, client workspaces, a validated qualitative model, team calibration, or causal outcome measurement. Current report exports use English labels. Keep the proposed categories and reviewer notes visible when carrying the analysis into the campaign retrospective.