Competitor research, without the fake certainty

No. Without a licensed traffic/ranking dataset, GapForge does not claim a traffic-ranked leaderboard. The list is a research shortlist based on detected subject/site type, market hints and optional live-search candidates.

A category leader can be useful as a quality benchmark. For direct competitive decisions, prefer sites serving a similar audience, task, country and business model.

It means the competitor did not return a readable public-HTML sample, so GapForge does not assign a fake low or high relevance percentage. Retry the competitor, use another candidate, or treat the domain as an unverified research lead until evidence is readable.

They solve different primary tasks. A resume builder creates CVs, templates and exports; a jobs portal focuses on vacancies, employers, recruiting and job search. GapForge treats these as separate precision subjects so job-board candidates such as Indeed should not be the default shortlist for a dedicated resume builder.

GapForge takes repeated phrases from your verified own-site titles and content headings and uses them to prioritize relevant same-site competitor URLs during the bounded crawl. The focus phrases are visible in the sample diagnostics. They influence page selection only; they do not change the measured checks or invent traffic/search data.

The percentage uses completed mechanical pass/fail checks. Review items are excluded from the score, so a site can pass every scored check while still having questions that need manual inspection.

No. Review means the crawler cannot responsibly decide automatically. GapForge v168 separates review-only signals from confirmed failed checks.

No automatic publishing. GapForge only promotes a BUILD phrase when it is supported by at least two sampled competitor pages and the competitor-fit gate allows topic action. You still need to validate relevance, intent, real demand and overlap with existing pages.

Yes. Navigation/boilerplate terms, brand noise and generic loose words are filtered. BUILD research now prefers 2–3 word title/H1/heading phrases supported by at least two competitor pages. Single-page observations can be preserved as suppressed evidence but are not promoted to BUILD.

The older vocabulary layer was mainly unigram-based. Those words can recur across headings without representing a real search/user task. The phrase-based evidence layer fixes this by requiring stronger multi-word and repeated-page support.

READY means the bounded sample, relevance rules and repeated phrase support are strong enough for research. VERIFY means inspect the competitor/audience fit before acting. PAUSED means the sample or fit is too weak for topic decisions. It is a rule-based evidence gate, not a traffic score.

GapForge does not bypass security challenges. If the SocialKit API itself is intercepted, the browser now identifies challenge HTML and reports a security-verification error. Complete the verification or adjust your own WAF rule, then retry.

The report stays partial or unavailable. A blocked competitor is not assigned fake weakness, gap or relevance evidence. Choose another readable competitor or use an authorized/rendered data source if you control it.

The interactive tool is intentionally bounded for server load, fairness and transparency. Saved Pro projects can run deeper scheduled samples, while JavaScript-rendered content requires the optional render bridge.

No. Public HTML cannot reveal private analytics, full rankings, all backlinks, brand demand or conversions. Use GapForge to prioritize investigation, then validate with Search Console, analytics and appropriate third-party datasets.

Fix confirmed failures, verify review items manually, improve important existing pages, research high-fit topic gaps, build original content, then measure real outcomes.

Yes. Discovery is a shortlist, not proof. GapForge re-checks fetched pages. If the fetched sample is ambiguous, the result becomes Niche unverified and BUILD stays paused. Only strong contradictory evidence becomes Niche mismatch — blocked.

A bounded crawl can land on generic pages, navigation-heavy pages or pages outside the competitor’s core task. Calling that a proven mismatch would be false certainty. v191 requires stable exact-niche evidence before blocking and uses an unverified state when the evidence is inconclusive.

GapForge recognizes broader task evidence such as replacement cost, cost to install, cost to repair, installation cost, labor/material cost, project estimate and cost guides. It does not require two sites to use the exact same marketing phrase.