BM25 Drafts is a RankGear workflow that scores your local draft files against the pages currently ranking for a target query, then estimates where each draft would land inside that comparison set.
| Term | BM25 Drafts |
|---|---|
| Category | Product and Workflow |
| Also known as | BM25 Draft Tool |
| Where it appears | Create → BM25 Drafts |
What it means in RankGear
The workflow lives under Create → BM25 Drafts. You point it at one or more draft documents you are still writing and at a set of live pages ranking for the query you are targeting. RankGear applies the BM25 relevance function — the term-frequency and document-length scoring used by classic retrieval systems — to weigh each draft against that corpus, then reports where the draft would sit relative to the pages already competing. The point is to compare a work in progress with real competitors before you publish it.
How to interpret it
Because this is a workflow rather than a single metric, read the whole run: confirm which drafts you fed in, which live pages RankGear scored them against, and the ranked comparison it produced. A draft’s BM25 placement is a comparative indicator against the specific set of pages you supplied, so it moves if you change the query, swap the competing pages, or edit the draft. Read a low placement as a prompt to look at topic coverage and phrasing against the leaders, not as a fixed grade.
Example
Say you are drafting an article targeting “commercial lease audit” and RankGear pulls the ten pages currently ranking for it. You load your draft into Create → BM25 Drafts and run the scoring, which places your draft eighth of eleven in the set. Opening the comparison shows the leading pages cover several sub-topics your draft skips entirely. You add that coverage, tighten the wording, re-run the workflow, and watch the estimated placement climb.
Important considerations
- BM25 Drafts describes RankGear behavior and a comparative estimate — it is not a Google ranking signal, and a strong BM25 placement does not mean the page will rank in search.
- The score is relative to the corpus you provide. Correlation with what those pages do is not causation, and the estimate only speaks to the pages inside that comparison set.
- Results depend on the draft text and the competing pages at the moment you run it, so re-running after edits or against a different set of competitors will change the placement.
Related terms
Part of the RankGear glossary · how RankGear measures · the 870 factors.