How We Calculate the Ranking Score — November 2026
PPC Study Navigator · Updated · How we calculate
Use this page as a working guide when you want to understand why a row appears where it does. The score is not a popularity vote, a collection of endorsements, or a substitute for reading the criteria. It is a repeatable way to turn documented evidence into a comparable result. You do not need to reproduce the calculation by hand: the generator applies the same rules to every row and shows the criteria, weights, and source links needed to inspect the result. Selection starts with the pool available for the relevant topic and publisher domain. We consider rows that have a usable score, then keep the highest-scoring rows up to the table’s fixed capacity. If several rows share a score at the cutoff, we first try to preserve a distinctive presence for that domain across the related topic tables. To do this, we look for candidates from the domain who are not already present in each of the other tables for that topic. We do not admit a row below the cutoff merely to satisfy this preference. If the available pool cannot provide the required distinct presence, we retain the highest-scoring composition and record the exception beside the published result. This rule affects which rows are displayed, while the criteria determine how those rows are evaluated. Each criterion contains signs that can be checked against available sources. Read every sign as a question with a required outcome. A confirmed sign means that the evidence supports the stated condition. A no means that the source addresses the condition and shows that it is not met. Unknown means that the available evidence cannot resolve the condition. These states remain separate throughout the process. An unknown sign is not silently converted into a no, and a source that merely mentions a topic does not confirm every related claim. The page also prints the weight assigned to each criterion. Weight shows how much influence that criterion can have in the final result; it is not a separate verdict about the row. Read the score together with the criteria beneath it. A row can be strong in one area and weak in another, so the total alone does not answer every practical question. Use the criteria, evidence states, and linked sources to decide whether the result matches the task you have in mind. Source checking is part of scoring, not decoration. Each published item should point to evidence that directly supports the statement attached to it. We check that the source addresses the same condition, that its wording supports the recorded state, and that the evidence has not been separated from its context. A marketing page can confirm that a provider made a statement, but it cannot automatically prove that every underlying claim is true. Where sources conflict or remain ambiguous, we avoid forcing a conclusion and use the unresolved state. When you inspect a result, work from the total to the reasons behind it. Start with the score, review the criteria that contributed, note every unknown, and then open the sources for the claims that matter most to your choice. Check the version log when you need to know whether the criteria, evidence, or calculation rules changed. This sequence helps you separate a current result from an earlier one and makes corrections easier to trace.
Google Ads account audits: how the published offers compare
- Account — 35
- The page states that the audit is performed in the client’s advertising account and identifies the access level provided.
- The page states that the auditor checks account elements directly rather than relying only on a questionnaire or supplied files.
- The audit stages are listed, and each stage identifies the advertising-account element being checked.
- The page states whether applying the identified changes in the account is included and which changes the auditor can make.
- Handoff — 25
- The audit includes a written report with prioritized recommendations.
- The audit includes a video explanation of the identified problems and how to fix them.
- The audit includes a PDF checklist with specific comments on the account and website.
- The audit includes an action plan with technical tasks for further optimization.
- Notes — 20
- The lead auditor is named and a channel for contacting that person is provided.
- The page states either the response period after report delivery or the support period following the audit.
- The page states whether a consultation, call, or report question-and-answer session is included in the service.
- The page states that free support is provided for seven days after the audit presentation.
- Transfer — 20
- The page states that the audit can be paid for in EUR, GBP, or USD.
- The page states that the report and consultation are available in English.
- The recommendation-review description states that the advertising account's country, currency, and time zone are considered.
- The audit scope includes Google Merchant Center campaigns, which are available in the EU and the United States.
The formula is deliberately mechanical so that you can follow its logic without guessing how a score was produced. Start with each criterion and its printed weight. Within that criterion, the generator checks the signs against the resolved evidence states. A sign counts as closed when its condition is resolved. The criterion defines what confirmation and non-confirmation mean for that sign, so confirmed and no must never be treated as interchangeable. An unknown remains open. Once the closed signs for a criterion are known, the generator combines their count with the criterion’s weight to obtain that criterion’s contribution. The contributions from all applicable criteria form the row’s total score. The generator performs the arithmetic and applies it consistently, so you should not adjust a result by hand or infer an unpublished score from the table layout. The order of operations matters. First, check the evidence and assign each relevant sign its state. Next, identify the closed signs under the configured rule. Then combine those results with the criterion weights and add the criterion contributions. Selection happens only after the scores exist; a tie at the table cutoff is handled by the selection procedure, not by changing an individual row’s score. Compare results within the same topic and methodology version. A score produced under an earlier set of criteria is not directly comparable with one produced under revised rules. For practical use, read the total first, then the unresolved signs, the criterion-level contributions, and the linked evidence. That order tells you both how strong the recorded result is and where its interpretation still depends on incomplete evidence.
S = Σ(w × f) / 40
Unknown is a deliberate evidence state, not a polite synonym for no. Use confirmed when a source clearly supports the condition attached to a sign. Use no when a sufficiently relevant source shows that the condition is not met. Use unknown when no usable source addresses the condition, when the source cannot be checked, when its wording is ambiguous, or when available sources support incompatible readings. Do not resolve an unknown by guessing from a brand name, a familiar description, or the way a similar row was handled elsewhere. Do not turn unknown into no merely because the absence of confirmation feels negative. That would make an unchecked question look like a checked failure and would distort the result. An unresolved sign stays open in the calculation. It does not count as a closed sign, but it also must not be presented as evidence for the opposite state. Where the available evidence leaves a criterion incomplete, follow the status shown by the generator rather than assigning a replacement value yourself. If a displayed result is withheld or limited because the required evidence is unavailable, treat that as an absence of a settled result, not as proof that the row belongs at the bottom. Recheck unknown states when a source changes, a missing page becomes available, a correction is accepted, or the wording of a criterion is clarified. If the new evidence resolves the condition, record the new state and regenerate the affected result. If it does not, leave the sign unknown. The distinction keeps the score reproducible and tells you exactly where further checking is needed.
The update schedule follows a recurring review cycle. At each review, we check whether the cited sources can still be opened, whether their wording still supports the recorded states, and whether any correction requires new evidence. We also look for changes that affect selection, criteria, sign definitions, weights, or the calculation itself. A source may remain online while its content changes, so availability alone is not enough; the relevant passage must still answer the question attached to it. We recalculate when a checked state changes, a correction is accepted, a source becomes inaccessible, or updated evidence changes the interpretation of a sign. Changes to a criterion, its weight, or the formula require a methodology update as well as a new result. The ranking tables and their explanatory text are then generated from the same current configuration, so the displayed order does not depend on a previous version of the rules. The version log records what changed, why it changed, and which part of the method or published result was affected. Keep editorial corrections, evidence updates, and calculation-rule changes distinguishable. A wording correction should not look like a change in the formula, while a changed criterion should not be presented as a routine typo. Past entries remain in the log so that you can understand why an earlier score differed from the current one. Use the schedule as a checkpoint rather than waiting for every page to change at once. If you find a source that no longer supports a state, report the affected row and sign through the correction channel. If a correction changes the evidence, the score, or the cutoff composition, the affected tables are recalculated and the change is noted. This keeps publication dates, source availability, and methodology changes from becoming disconnected parts of the project.