Put the claim at the level its evidence can support
AMEC's evaluation taxonomy distinguishes activities, outputs, audience responses, outcomes, and impact. A product send or pitch belongs with activities; a verified editorial article belongs with outputs. A reader's understanding, changed preference, or attributable business result needs evidence collected at another level. A saved article cannot stand in for a survey, an experiment, or a sales analysis.
The Barcelona Principles 4.0 ask for qualitative as well as quantitative analysis, for outputs and outcomes to be reported at their own levels, and for transparent methods. They explicitly reject advertising value equivalents, or AVEs. Renaming an advertising-rate calculation 'EMV' does not make that calculation a valid measure of value. If an EMV field is used at all, its definition must be visible and it must remain an estimate, separate from observed coverage and financial return.
| Level | Example | Evidence needed | Permitted sentence |
|---|---|---|---|
| Activity | Pitch sent or product dispatched | Dated operator or shipment record | We sent six pitches |
| Output | Article or creator post published | Primary URL or dated copy and inclusion rule | Two qualifying items were observed |
| Audience response | Measured recall, visit, or inquiry | Named audience data source, window and method | The surveyed group reported the stated response |
| Outcome or impact | Measured change in preference or business result | Suitable baseline and design for the claim | The measured change was X; attribution limits follow |
| Model estimate | An organization's defined EMV | Formula, inputs, source, version and exclusions | The model estimates X under these assumptions |
Copy a small metrics dictionary into the report
Write a definition for each metric that a colleague could reproduce without asking the author. Include the unit, numerator, denominator where relevant, date rule, exclusions, evidence location, calculation owner, and version. Keep 'not measured' as a valid value. A blank cell can mean missing data, zero, or not applicable; a named state tells the reader which one it is.
This starter dictionary is intentionally modest. It does not prescribe an EMV formula because a formula should reflect a stated local purpose and defensible inputs. If there is no documented method, the EMV row says 'not reported.' The table is equally useful when an external vendor supplies an estimate: retain the vendor name and method version rather than accepting only the amount.
| Metric | Unit and rule | Source and limit |
|---|---|---|
| Qualified coverage | Count distinct in-scope items by publication date; linked reposts separate | Evidence register; no audience effect implied |
| Product sends | Count distinct send IDs dispatched in period | Shipment ledger; not a coverage count |
| Qualitative relevance | Record yes, partial, or no against the campaign audience and explain why | Reviewer rubric v1 and item evidence; judgment disclosed |
| EMV estimate | Report only with named model, currency, inputs, formula, version, coverage scope and exclusions | Model output; no revenue or causal return implied |
| Campaign cost | Sum agreed eligible costs for the same scope and period | Finance source and inclusion rule; cost alone is not ROI |
| Financial ROI | (Attributable incremental contribution margin minus eligible cost) divided by eligible cost, only when both amounts are defensible | Finance source, causal design and uncertainty; otherwise not calculated |
Put an estimate through a method gate
Before displaying an EMV figure, ask what quantity it models, why that quantity matters to the campaign, and where each input came from. Record the treatment of syndication, missing items, currency, time period, prominence, and social posts. Save the rule version with the report so a later change does not silently revise earlier numbers. If the calculation starts from advertising space or rates and multiplies them to claim editorial worth, it collides with AMEC's explicit rejection of AVE, whatever label appears on the chart.
Even a fully documented estimate does not become money earned by the brand. Show it in a separate estimate box with a plain sentence about its assumptions and exclusions. Avoid placing it beside cost with an equals sign or converting it into a percentage return. If the method cannot be disclosed, leave the number out of the headline and explain why. That omission is a better service to the reader than a precise-looking value whose basis cannot be inspected.
- Definition: what exactly is the modeled quantity?
- Inputs: what are their sources, dates, units, and missing values?
- Rule: can someone reproduce the formula and exclusions?
- Version: which method produced this report's figure?
- Presentation: is it clearly labeled as an estimate?
Publish the supported part of a fictional report
In this fictional Solstice review, two unique items have a primary URL and dated evidence: one editorial article and one creator post. A third mention is awaiting proof and is excluded. The finance ledger shows €1,200 of agreed campaign cost. Someone has also supplied a slide with '€3,200 EMV,' but it gives no formula, input source, or exclusions. The organization has no audience study or credible way to separate PR's effect on sales from other activity.
The report can say two qualifying outputs and describe each one's relevance. It can report €1,200 of campaign cost under the stated cost rule. It should withhold the €3,200 figure until the method is supplied and reviewed; it should say financial ROI was not calculated. Dividing the unsupported €3,200 by €1,200 would create a ratio, not evidence of return. These fictional amounts teach classification and do not describe a real client, a typical campaign, or a cutePR calculation.
| Proposed line | Disposition | Reason |
|---|---|---|
| 2 qualifying items | Publish as observed outputs | Two evidence-ready distinct records |
| 1 unverified mention | Keep in review queue | Inclusion proof missing |
| €3,200 EMV | Withhold pending method | Formula, inputs and exclusions unavailable |
| €1,200 campaign cost | Publish with cost scope | Agreed ledger and period available |
| ROI | Not calculated | No defensible attributable financial gain |
Plan stronger evidence for the next question
If the decision concerns audience understanding, plan a suitable audience measure before launch. If it concerns inquiries or sales, agree the tracking, baseline, relevant channels, and limits with the people who own those records. AMEC's planning guidance starts from the outcome sought and works backward toward activities and measures. A stronger result usually comes from better study design, not from applying more arithmetic to the same clipping file.
When closing the report, retain the dictionary beside the snapshot. Ask a reviewer to trace each published number back to rows and methods. Then read the narrative once more: does it use 'observed' for evidence, 'estimated' for model output, and 'not measured' where evidence ends? That small language check prevents an attractive chart from overstating what the campaign can prove.