Trace ranking keywords, page concentration, and a directed gap

Keep three evidence sources separate

Ranking rows belong to the ranked-keywords module, Top Pages belongs to its own module, and Keyword Gap belongs to the explicit paid run that requested the domain pair. In a Domain profile, ranked keywords and Top Pages can come from different latest-successful source runs. The pair-specific gap is deliberately not composed from another profile source and remains attached to the selected base run.

Read the source notice, provider date, module state, persisted count, total count when available, and collection limit before joining the three panels into one story. Matching market settings help comparison, but they do not make separate module dates or row universes identical.

Ranking rows connect a query to its observed landing page

The overview ledger exposes keyword, provider-observed position, ranking URL, provider intent, monthly search volume, estimated traffic, CPC in USD, paid-search competition, and keyword difficulty. The default presentation favours strong observed positions, but every sortable column changes only the loaded working view. It does not recollect or rerank the provider evidence.

Position and URL describe the observation in this country and language. Search volume and ETV are provider estimates, not Search Console impressions or first-party visits. CPC, top-of-page bids, and competition describe paid search. Keyword difficulty is a separate organic competition estimate. Intent is a provider classification that still needs a human reading of ambiguous queries and the current SERP.

The overview and Rankings tab use several honest bases

The headline organic keyword count and position-bucket counts come from rank-overview aggregates. The overview table is paginated and reports loaded versus provider total coverage when the manifest carries both. In the detailed Rankings tab, server aggregates cover every persisted ranked-keyword row, up to the run's bounded storage scope, while the table itself can continue loading cursor pages.

The detailed position chart therefore has two bases. Keyword mode uses the headline position buckets. Search-volume and estimated-traffic modes sum persisted detail rows per bucket. Difficulty classes, volume classes, and SERP-feature counts also use every stored row. A demand-history chart, by contrast, combines monthly searches from the currently loaded visible keyword rows and can change after loading or sorting the working set.

• Older stored rows can lack monthly searches, bid ranges, SERP features, or average referring-domain fields; the detail table hides a rich column until at least one loaded row carries it.
• A dash means the field is absent. Formatted zero means a stored numeric zero.
• The detailed KPI delta compares the current aggregate with a previous comparable stored analysis when one exists. Provider new, up, down, and lost counters are a different signal.

Intent mix covers every persisted ranking row

The current Intent Distribution is a bounded server aggregation over all persisted ranked-keyword rows for the source snapshot. It is not limited to the first cursor page. For each stored intent value it shows keyword count, share of stored rows, and summed estimated traffic.

A nonzero strip segment receives a 1.5 percent visual floor so a small category remains reachable. Exact count, calculated share, and ETV stay in the rows and tooltip. Unknown or missing intent reflects unavailable classification, not an instruction to exclude the keyword.

• Intent mix describes the retained provider set, not every query for which the domain might appear.
• Commercial value still depends on product fit, audience, funnel, and business evidence.
• Mixed intent can prompt a landing-page review, but it does not prove cannibalization or a wrong page by itself.

Top Pages is a bounded provider list ordered by estimated traffic

The Top Pages source stores provider relevant-page rows with URL, full organic segment, and paid segment. The overview starts by estimated organic traffic and shows each page's organic keyword count. Its small comparison bar is relative to the strongest currently visible page, so it is not the page's share of the whole domain.

The current plan defaults to 50 Top Page rows and caps the request at 200 before any effective policy restriction. Coverage text distinguishes loaded, returned, total, and limit when those values are available. A page outside the retained set is not evidence of zero organic value.

• Estimated page traffic is provider ETV, not pageview or conversion data.
• Page keyword count is the provider's page-level aggregate and must not be substituted for the separate domain keyword headline.
• Search and sorting apply to loaded rows. Load More expands the stored view without buying a new provider response.

The Pages tab measures concentration inside the stored page set

Server aggregates sum page-level keyword counts and ETV across every stored Top Page row. The Top 1, Top 3, and Top 10 concentration values divide the strongest pages' ETV by that stored-page ETV sum. They do not use first-party sessions or the full domain ETV headline as denominator. The visible Top Pages summary is limited to ten server-sorted URLs.

The portfolio map shows at most the first fifteen loaded pages. Horizontal position is page keyword count, vertical position is estimated traffic, and bubble size is that page's Top 10 keyword count. Its coordinates use logarithmic scaling, so read exact values rather than treating chart distance as a linear multiple.

• URL depth groups homepage, one path segment, two segments, and three or more segments, then charts stored ETV per group. It describes URL structure, not crawl depth or information-architecture quality.
• The detailed table separates organic and paid keyword counts, organic and paid ETV, stored-set traffic share, Top 3, Top 10, rank composition, and provider up/down counters.
• A high-ETV page can still be commercially weak, outdated, or dependent on unrelated queries; a low-ETV page can still be strategically important.

Page movement and Site Audit health need their own dates

The page movers summary totals provider new, up, down, and lost counters across stored page segments. A page's net value is up minus down. Crawl Foundry shows at most five positive and five negative net movers. This is not a change calculated from two Crawl Foundry page snapshots and does not prove which edit caused the movement.

When a Site Audit inventory is available, Crawl Foundry cross-checks up to twenty stored Top Pages, ordered by ETV, against the workspace audit inventory. It reports matched coverage, HTTP status, and open issue counts, and lists pages needing attention. Domain Overview source time and Site Audit update time remain separate. An unmatched URL means no current inventory match, not a healthy page.

Keyword Gap answers one directed pair question

The request sends the analyzed source domain as target 1 and the comparison or own domain as target 2, with intersections disabled. The stored result therefore means: the source domain ranks for this keyword and the missing-from domain does not appear in the provider intersection result. Reversing the domains reverses the question and requires its own explicit run.

A gap is not a reusable module assembled from the newest evidence for each domain. It belongs to the exact source domain, missing-from domain, country, language, request limit, provider response, cost settlement, and run date. The current default requests up to 250 gap rows and the code cap is 1,000 before an effective policy can lower the scope.

• The source-domain position is shown; absence for the other domain is not a measured position of zero.
• Brand, navigational, regulatory, local, irrelevant, or infeasible queries can be valid provider gaps and poor strategic targets.
• The result does not prove missing content, weak content, a required new page, or expected traffic gain.

Gap rows carry more evidence than the compact panel shows

The overview Gap table exposes keyword, source-domain position, and monthly search volume, and keeps the direction in its caption. Stored rows can also contain CPC, paid competition, keyword difficulty, intent, ranking URL, ETV, monthly demand history, bid range, SERP item types, core keyword, detected language, language mismatch, and top-result backlink averages. Compact presentation does not erase those source fields.

The module distinguishes pending, failed, not requested, available but empty, and truncated coverage. Empty for a valid pair means no rows were returned within this provider request and scope. Not requested means no pair-specific gap was bought. Failed means the attempted module did not yield usable rows. These states are not interchangeable.

Selection is limited to valid loaded candidates

Keyword and Gap checkboxes work on rows already loaded in the browser. Candidate preparation trims terms, deduplicates them case-insensitively, rejects empty or overlong values, and retains only finite metrics inside the import contract's ranges. In the ranking overview, Select All follows the currently visible filtered candidates. In the compact Gap panel it follows valid candidates from all loaded gap rows.

Selection survives local sorting and further cursor loads by normalized keyword key, but it is not a saved workspace object until an explicit import succeeds. A selection count is not proof that every stored or provider-total row has been reviewed.

Database import and research handoff do different work

Add to Keyword Database creates or reuses a bounded list named with the Domain prefix and normalized domain. The non-billable bulk-import request sets enrich and force to false. It imports selected normalized terms plus valid search volume, difficulty, CPC, paid competition, intent, and detected language. The source rank, ranking URL, and ETV are not imported as current Keyword Database metrics by this handoff. Duplicate outcomes are reported by the import response.

Use in Keyword Research passes the selected normalized terms as seeds and is shown only when both Keyword Discovery and Keyword Database features are enabled. The selection must fit the current seeds-per-run limit. Opening the research route does not execute a paid run, approve every seed, or preserve all Domain Overview metrics as research evidence.

Do not turn bounded or missing evidence into a page verdict

A short table can reflect pagination, a plan limit, provider total versus persisted coverage, filtering, legacy fields, a compacted run, a not-requested module, or a failed module. Read the manifest and coverage caption before concluding that the domain has no other rankings, pages, or gaps.

Use Domain Overview to select a reproducible investigation. Use Keyword Database to own imported terms, Site Audit to verify technical page evidence, first-party analytics and Search Console evidence to assess owned outcomes, and a human decision record to connect observation, action, and result.