Methodology
How we source, label, and verify
You are being asked to prepare on the basis of what we say a program requires. This page explains where each claim comes from, and what it means when we say we do not know.
Last updated
Every claim carries its own status
A program profile is not one document with one level of confidence. The application deadline may be published while the panel size is only ever mentioned by applicants. So each individual field carries a label, and the labels look like this:
- Official
- Taken from the program or employer’s own published material, with a source link and verification date.
- Candidate report
- Reported by a past applicant, moderated and dated. It is not an official requirement.
- Editorial
- Our own explanation or interpretation, written to help you act on the official material.
- Needs re-check
- Past its review date. Treat it as a starting point and confirm with the official source.
- Unverified
- We could not confirm this against an official source. It is not presented as current fact.
The labels differ by shape and by word as well as by color, so they stay distinguishable in grayscale, at high zoom, and to a screen reader.
How a profile is built
- A researcher drafts from the program’s and employer’s own published material.
- Every factual field gets a source, a verification date, and a status. Unknown stays unknown.
- An editor checks source support, non-affiliation language, and internal consistency.
- A trade reviewer checks terminology, practical realism, safety language, and whether the interview rubric actually applies to this pathway.
- A publisher releases an immutable version and records a date to review it by.
- Automated link checks and a freshness queue surface pages that need re-verification.
- Corrections open a ticket and a new draft. A published version is never edited quietly.
What we do not do
- We do not publish, solicit, buy, or reconstruct live protected exam items. If a submission contains answer-key material, it is quarantined rather than used.
- We do not merge candidate reports into official facts. They sit in a separate, labelled section.
- We do not let anyone pay for placement, ranking, favourable reviews, or edits to a factual profile.
- We do not imply endorsement by any program, union, utility, or testing vendor.
How the AI feedback works
When you record or type an answer, it is transcribed, you get a chance to correct the transcript, and then it is evaluated against a published rubric that has six weighted dimensions.
The evaluation is bound by rules we enforce in code, not just in a prompt:
- It may quote only your own transcript. If the evidence for a claim is not in what you said, the claim is rejected before you ever see it.
- It never predicts acceptance, a passing score, a rank, or employment, and every scorecard says so.
- It is not given your name, email, protected characteristics, payment data, or unrelated history. It receives the question, your transcript, timing, the rubric, and the safe published context for your target.
- An answer that is empty or too short gets a deterministic coaching response rather than a forced score.
- Model output that fails schema validation is repaired or rejected server-side. Nothing unvalidated is ever rendered.
Every scored response can be rated unhelpful and reported as unsafe, biased, irrelevant, or incorrect. Those reports drive the regression set we run before any prompt, rubric, or model change ships.
Independence
An independent preparation service. Not affiliated with, endorsed by, or acting for IBEW, NECA, the Electrical Training Alliance, EEI, any local union, JATC, utility, or testing vendor.
We name organisations only to identify what you are preparing for. Where we link to an official requirement, the link goes to the organisation’s own page so you can check us. If you find something wrong, report a correction — we publish a count of corrections on every profile.