The short answer
Save the exact proposed sentence beside the private source identity, data definition, included and excluded records, reproducible operation and limitations. Review permission separately. After publication, add the real URL and date. A file digest helps identify the source bytes; it cannot establish that a claim is true or that publishing personal data is permitted.
What you’ll take away
- A claim needs its own unit, clock and selection rule.
- A matching digest identifies bytes rather than certifying truth.
- Record corrections without overwriting the earlier claim.
Freeze the sentence you intend to support
“My audience grew” and “two selected post observations totalled 250 impressions at 24 hours” require different evidence. Start with the exact sentence. Define its subject, metric, unit, observation clock and scope before choosing a persuasive screenshot.
Keep the draft status explicit. A prepared sentence is not a published result. The blank templates in English and German leave space for the actual publication URL, which you fill only after the page or post is live. Do not turn a scheduled item or preview into a completed-publication receipt.
Identify the source version privately
Record where the authorized source came from, when it was observed and which exact file version was used. A byte size and SHA-256 digest can help distinguish versions without copying private rows into the note. Keep the sensitive source in suitable private storage rather than attaching it publicly.
Hash the actual original bytes, not text copied through an editor. Line-ending or encoding changes can change the digest without changing the visible table. Matching a recorded digest supports byte identity; it does not prove provenance by itself, prevent a fabricated source or grant rights to publish it.
Make selection and arithmetic reproducible
Write the inclusion rule and show missing, excluded and duplicate records separately. Preserve the dictionary version that explains the metric. Specify whether an operation is a sum, median, ratio or set comparison. A screenshot of the final number rarely explains these choices.
Use the smallest adequate worked trace. For a ratio, retain the numerator and denominator; for a comparison, retain both scopes and clocks. If the source cannot answer the intended question, narrow the sentence instead of filling the gap with a guessed value.
Trace the complete fictional packet
The original reporting example has three rows. Its dictionary defines observations at age 24 hours, blank impressions as unavailable, and distribution as a declared category. Selecting rows with declared organic distribution and an available impressions value keeps p-01 and p-03. The values are 0 and 250; the sum is 250.
The claim manifest records the exact byte sizes and SHA-256 values of that CSV and dictionary, the selection and the limited fictional statement. The unknown-distribution row with a blank measurement is not silently converted to zero. This exercise demonstrates a review trail; it is no creator benchmark, observed customer outcome or new research study.
| Evidence step | Constructed result | Limit |
|---|---|---|
| Input | 3 fictional post observations | No actual account sampled |
| Selection | 2 declared-organic rows with available values | Category declaration is not platform verification |
| Operation | 0 + 250 = 250 impressions | Not unique people or customers |
| Clock | 24 hours after each fictional publication | Not all displays during a calendar month |
Review permission independently of arithmetic
A correct calculation can still be inappropriate to share. Review whether names, quotes, links or private interactions belong in the public sentence and artifact. Remove irrelevant personal detail and obtain any needed permission before using a person’s contribution. A digest does not anonymize the underlying source or replace that decision.
Record the permission status without exposing private correspondence. If the intended evidence cannot be shared, describe the method and limitation honestly. Do not invent an independent reviewer, attribute a fictional exercise to a real user or imply that an internal check measured reader preference.
- Copy the exact draft claim into the template.
- Record source identity and dictionary version privately.
- Write inclusion/exclusion rules and reproduce the operation.
- Review factual scope, permissions and remaining limitations.
- After actual publication, add its URL; link later corrections to the earlier record.
Keep the record useful after publication
Add the actual public URL and date once verified. If the source changes later, retain the version used for the published sentence. A current dashboard total cannot silently replace a historical observation while the public wording still claims the earlier clock.
For a correction, save the earlier sentence, the corrected sentence, the reason and affected publication. Link the records instead of erasing the previous trail. The evidence packet supports checking what you did; real reader benefit and business outcomes still need separate observation.
Sources & product references
Product details reflect the publication date. External features and interfaces may change.
