How PPC Audit builds its score — November 2026
PPC Audit Review · Checked · Scoring rules
Scoring starts with a claim, not with a conclusion. The site records what a provider says about a course or related offer, then asks what supporting material actually shows. A statement can sound complete and still remain unconfirmed. The published result therefore reflects a documented check: either the available evidence supports the relevant feature, establishes that the feature is not met, or leaves the question unresolved. Those outcomes are kept separate throughout the process. The candidate pool is built separately for each course topic and audit track. Entries enter only when they meet the current eligibility rule for that pool. The table’s row cap and score ordering then determine the cut: higher-scoring entries come before lower-scoring entries. When the cap falls on equal scores, composition is checked before the boundary is finalized. The rule requires each represented domain to keep the required share of entries that are absent from the paired tables for the same theme at the same time. A boundary entry can be used only if that condition is satisfied; an entry below the resulting boundary is not inserted merely to fill the table. If the pool cannot support the full composition rule, the leading rows by score remain and the table note records the exception. This fallback is explicit rather than editorial. In a pool that already fits within the cap, all eligible entries remain, and the note explains that no row satisfies the exclusivity condition. Selection is rerun when the pool, evidence, or scores change, so membership follows the same documented rule rather than an informal preference. The configuration defines the criteria, their weights, and the features checked under each one. This page does not reproduce the numeric weights; the generator displays them with the criteria, and any change to them belongs to a new method version. The areas are distinct. Completeness concerns whether the published course material covers the expected subject scope. Evidence concerns whether a relevant claim can be traced to a source that supports it. Reproducibility concerns whether the documented path lets a reader reach the same check. Transparency concerns whether the material makes its scope and conditions clear. Unrelated strengths do not compensate for an unresolved feature unless the configuration says that they may. Each feature has an explicit state. Confirmed means that the cited evidence supports the statement being checked. No means that the available evidence establishes the absence or failure of the requirement. Unknown means that the evidence is insufficient, inaccessible, ambiguous, or otherwise cannot settle the question. Unknown is not a softer No. A missing document is not treated as proof that a feature is absent, and a marketing statement is not treated as proof that it is present. The generator receives these states as distinct values and applies the corresponding rule in the current method configuration. Source checking is part of scoring, not an optional annotation. Each usable fact is tied to a source location and a supporting excerpt that shows the relevant context. The reviewer traces the claim to that material, checks whether the wording is actually supported, and records the result. A fact may be published only after verification; unverified claims are not substituted into the table. When material is changed, withdrawn, or too unclear to interpret, its state is revised rather than silently guessed. The source list shown with the table lets a reader inspect the basis for the published facts. If sources disagree, the method does not blend them into a favorable conclusion: the narrower supported statement is recorded, or the feature remains Unknown. The same rules apply to every entry in a pool, including a leading entry and a boundary entry.
Google Ads audit comparison: published scope, evidence and price
- Coverage — 35
- The audit scope includes account structure, search terms, bids, strategies, ads, and landing pages.
- The audit scope includes checking account structure and campaign settings.
- The audit scope includes checking search terms, ads, and advertising assets.
- The audit scope includes checking how ads correspond to landing pages.
- Evidence — 35
- The report description states that it includes a separate checklist of comments on the account and website.
- The report description states that recommendations are prioritized and accompanied by technical tasks.
- The report description states that recommendations for campaigns, search terms, and ads are presented separately.
- The report description states that the technical tasks are formulated for implementation in Google Ads.
- Repeat — 20
- The audit description states that checks are performed manually using the supplied data.
- The audit description states the total number of inspection points.
- The audit description lists the data the client must provide: Google Ads access and, where relevant, Google Analytics or GA4 access.
- The page publishes a fillable report template or a worked example containing source data, actions, and output.
- Clarity — 10
- The page states a numeric price for the paid audit.
- The page states which additional work is performed for an extra fee.
- The page states when the audit results will be delivered after payment or application approval.
- The page lists the account access and the data or files the client must provide for the audit.
The score is generated from the configured criteria, their weights, and the feature states. For each criterion, the generator applies the stated rule to the checked result and converts that result into the criterion’s weighted contribution. The row total is the sum of those contributions. Confirmed features receive the contribution defined for them; No receives no credit for the failed requirement; Unknown follows the empty-result handling described above, so an affected row receives no public numerical score. The weights apply consistently across the pool and belong to the method version used for that calculation. Editors do not add a discretionary bonus, reduce a result because of tone, or override the generated total. Ranking order follows the computed total, and the documented tie and composition rule is then applied at the boundary. When verified evidence or the configuration changes, the score is recalculated and the journal explains the change. The generator displays the exact formula, weights, and criteria alongside the result, keeping the calculation visible without manually rewriting it here.
S = Σ(w × f) / 40
Unknown is used when a real source check cannot be completed. Typical causes are an unavailable page, an excerpt that does not contain the claimed detail, conflicting context, or a requirement that cannot be tested from the available material. It records the limit of the evidence, not a judgment about the provider. The calculation does not convert Unknown into No, and it does not assign a midpoint or an estimated contribution. When a feature remains empty under the current configuration, the generator removes that row from the published table and does not publish a numerical score for it. A partial record therefore cannot look like a completed assessment. If later checking resolves the question, the feature can move from Unknown to Confirmed or No during the next review, and the version journal records that transition. Until then, the row receives no provisional total and no hand-set replacement value.
Reviews follow a recurring publication calendar. A routine pass checks source availability, newly published material, changes to stored facts, and entries with open questions. A full recalculation rebuilds the candidate pools, reads every current feature under the active configuration, recomputes the results, and reapplies the selection and tie rules before anything is published. Interim checks may be narrower, but any change that affects a feature or result is followed by recalculation of the affected table so that displayed scores and order stay consistent. The generator supplies the update marker for the current revision. Each completed pass also produces a version-journal entry. The entry records what changed, why it changed, whether the change came from a source update, a clarification, or a correction, and which tables were affected. A source refresh that does not alter the result can still be logged, so a missing result is not confused with missing review. Wording changes are distinguished from methodology changes. If the configuration changes, the new version is identified and the affected scores are recalculated rather than carried over without explanation. Exact revision dates and labels come from the generator’s metadata. This creates a traceable chain from the checked source to the published table without turning the narrative into a substitute for the table itself.