SeitenreportSEO & website analysis since 2007

Free · without registration

WDF*IDF analysis with own reference texts

Compare the terms of your text with three to ten self-selected reference texts. You see word occurrences, weighted frequencies, and the spread of a term in the comparative corpus.

Define text and comparative sources

References: one text or one URL (3–10) each
Word filter

No automatic Google query. Inserted texts remain local. For URLs, we load the static HTML; JavaScript content is not reloaded. A maximum of 150,000 characters and at least ten words per document.

A WDF*IDF analysis with traceable sources

This tool starts with your documents. Choose references to answer the same question or task, such as three different guides on the same topic. Texts with a different search intention, very short product descriptions or almost identical copies distort the comparison. The evaluation therefore does not replace an editorial selection of the sources.

A term that appears in multiple references and is missing from your text can point to an aspect you want to examine. But it can also belong to a different focus or to a foreign brand. Based on the content, decide whether a supplement makes sense for your readers.

This is how it's calculated here.

For a term with frequency f and a document with L Words apply here WDF = ln(1 + f) / ln(1 + L). Off N Reference documents and their document frequency df is: IDF = ln((N + 1) / (df + 1)) + 1. The table shows the product of both values. The reference average includes all references, including those without the respective term.

The weighting is based on the small, self-chosen corpus. Other WDF*IDF tools can use different formulas, corpora, and text filters to provide different numbers. An absolute value is not a quality measure and not a Google ranking factor.

Which words are compared?

Capital and lower case will be standardised. Terms from three characters appear in the table; frequent filler words can be hidden. For the document length, all recognized words count before this filter. Word stems and synonyms are not merged: “Website” and “Websites” remain different terms.

For URLs, the text is preferred from main It is used, otherwise the body. Navigation, head and foot areas and scripts are omitted. That’s a static approximation of the main content, not a rendered Google crawl. Check the word numbers and insert the relevant text yourself if necessary.

From Comparison to Revision

  1. Check sources and word numbers for thematic comparability.
  2. Substantially classify terms with broad reference coverage.
  3. Missing answers, examples or explanations complement, if they belong to the page target.
  4. Shorten unnecessary repetitions and re-read the text.
  5. With the Readability checker the comprehensibility and then check the actual development with Search Console.

Frequent questions

Does the tool ask for Google search results?

No. You select three to ten reference texts or public URLs yourself. There are no chargeable SERP queries.

What is the ideal WDF*IDF value?

There is no universal ideal value here. Average and maximum describe only the selected references and are not specifications for the search engine.

What happens to non-retrievable reference URLs?

The affected source is named with error. The comparison only starts when all selected sources can be evaluated; this does not change the corpus unnoticed.

Will my inserted texts be saved?

The local analysis does not store and transfer inserted texts. Only explicitly entered URLs are sent to Seitenreport for retrieval.

Can the tool be used for other languages?

Unicode word recognition also processes other languages. The optional filler word list is in German and English; turn it off if necessary.