First-hand and original evidence
Use direct evidence only for the work it actually records.
Original evidence can come from a documented observation, test, process or source record produced through direct work. A new example can also teach a method, but a fictional demonstration is not an observation. Keep those roles distinct and preserve access limits.
Free learning from Semantec SEO.
Start with the question and the available access.
State what the evidence needs to answer and what you can inspect. A person with access to a draft can review its wording; that does not give them access to real visitor behavior, a client’s private account or a platform’s internal ranking system.
Choose a method suited to the question: inspect a defined record set, reproduce a documented procedure, compare before and after text, or run a controlled test with stated settings. Record who did the work, when it happened and what was included.
If the needed access is unavailable, name the missing input and its consequence. A polished explanation cannot replace an unperformed test. Do not write “we tested” because a model can imagine a plausible result.
Separate observations, interpretations and examples.
An observation records what was inspected or measured under stated conditions. An interpretation explains what that record may mean. A hypothetical example illustrates a method with invented values or facts. The three can be useful together when their boundaries remain visible.
Keep raw notes, method, version, date, selection rules and any contradictory result. Record negative and inconclusive outcomes as well as successes. A screenshot should have a task, a relevant detail and a text explanation; it does not prove behavior outside its recorded view.
Do not turn one observation into a universal promise. A successful local test does not establish every version, user environment or future outcome. Record which assessment it supports and what further check would be needed.
Worked demonstration: inspect six fictional source cards.
The following demonstration is invented for this lesson. No real records were inspected and no client or visitor result is claimed. Its purpose is to show a transparent method for recording an evidence check.
Define the teaching question: does each source card contain a document identifier and a version or date sufficient to select the intended technical record? Inspect exactly six fictional cards A–F against that stated rule. Do not count a card as complete simply because it has a title.
The invented cards A, B, D and F include both fields. C has an identifier but no date or version; E has a date but no document identifier. The demonstration count is four of six cards meeting this narrow field check. It does not mean that four claims are true, that the documents are current or that a reader completed a task.
A real application would preserve the observed card records, collection method, access date and reviewer. The next step for C and E is an evidence request, followed by checking the identified documents themselves. The fictional demonstration remains labeled even if its layout resembles a real test log.
| Fictional card | Identifier | Version or date | Field-check outcome |
|---|---|---|---|
| A | Present | Present | Meets stated field rule |
| B | Present | Present | Meets stated field rule |
| C | Present | Missing | Hold; request date or version |
| D | Present | Present | Meets stated field rule |
| E | Missing | Present | Hold; request document identifier |
| F | Present | Present | Meets stated field rule |
Document a real test without broadening its claim.
For an observed test, record the question, protocol, sample selection, equipment or system version, settings, start and end dates, observations, exclusions and failure conditions. Preserve the output or a suitable evidence reference and identify the reviewer.
Write the result separately from its explanation. State what was observed before proposing a cause. If the comparison lacks a control or changes different conditions at once, keep those limitations visible. A before and after observation is not automatically proof of causation.
An original example can add clarity without real measurements. Label it as a teaching example at the point of use, not only in a distant disclaimer. Keep generated illustrations separate from documentary photographs or observed interface captures.
Preserve permission, privacy and source ownership.
Check the permitted use of supplied records, screenshots, names and quotations before publishing them. Keep only information needed for the lesson, remove unnecessary personal details and record any restrictions from the source owner. Access to internal material does not itself authorize public reuse.
If confidential details prevent a real example being published, explain the access limit and use an explicitly fictional demonstration if it still teaches the task. Do not invent a customer quotation, brand relationship or test outcome to fill the space.
When permission is unclear, request an approved extract, an anonymized record prepared by the responsible owner or another suitable source. Do not describe the missing material as verified evidence.
Choose a useful contribution and review it.
An observation can improve a definition, expose a condition, support a comparison, explain a failed step or identify a useful next route. Place it near the claim it qualifies and explain what the reader should learn. A personal anecdote without a relevant task does not automatically strengthen an answer.
MIRENA’s supplied first hand input method separates the raw record from its published use, checks source identity, freshness, permission, support and relevance, and assigns an owner and placement. This is an attributed editorial process, not a confirmed search ranking formula.
Google’s helpful content questions discuss experience and explanation of how content was created. They do not turn a fictional demonstration or an unperformed test into evidence.
Keep missing evidence visible.
Use states such as available for review, missing source, permission needed or unsupported claim. Preserve the reason and next responsible action. If a source changes, revisit the dependent explanation rather than merely updating its date.
Continue with quantitative claims and useful contribution. Return to the evidence and useful contribution overview or learning overview.