Content Metrics and KPIs

For every article Marfeel computes and makes available across the data warehouse all the metrics listed below. These metrics can be used as dimensions for qualitative analysis on the content production from Explore, Optimize or Content Opportunities reports.

Some of the content metrics are inspired by The State of Content Marketing report by Semrush which reveals findings to build winning content strategies based on data.

Article Body

Given a URL Marfeel automatically extracts its content using Readability which uses a set of heuristics to differentiate the content from UI elements.

Depending on the html markup Readability might incorrectly keep or remove text from an article causing inaccuracies into how the metrics are calculated. You can fine tune Readabilty from Organization Settings > Metadata settings whitelisting different html elements via CSS selectors. Whitelisted elements won’t be removed when processing articles.

TIP:
Use Metadata settings to fine tune which elements Marfeel detects as part of the article body.

H1 length

The H1 length in number of words.

Heading distribution

Presence or absence of H2, H3 and H4 headlines. The possible values are:

  • No H2
  • H2 only
  • H2 + H3
  • H2 + H3 + H4

Image count

Number of images in the article.

List count

Number of lists <li> for every 500 words of plain text.

Video count

Number of videos in the article.

HTML Language

Language of the article as specified in the <html lang="en"> tag. The language of the article is used to perform language-specific calculations like the number of syllables and can also be used as a dimension to analyze sites with a multi-language strategy.

Number of words

Word count of the article. We internally use phpSyllable which basically splits a text in spaces and counts the parts.

Number of sentences

Number of sentences in the article.

Number of syllables

Number of syllables in the article. We use TEX hyphenation patterns for most of the languages.

Flesch Index

In 1946, lawyer, author and writing consultant Rudolph Flesch published a readability formula in his dissertation, “Marks of a Readable Style.” That formula, the Flesch Reading Ease index, was the original Flesch test. The formula is:

206.835 – (1.015 x words per sentence) – (84.6 x syllables per word) = reading ease

Flesch’s work with the Associated Press helped bring the reading level of front-page newspaper stories down by five grade levels. Publishers increased readership by 40% to 60% with the formula.

Scores range from 0 to 100. The higher the score, the easier your message is to read.

Flesch Reading Ease 2

Score Level Words/sentence Syllables/word Estimated school grade % of adults who can read at this level
90-100 Very easy 8 or fewer 1.23 or fewer 4th 93
80-90 Easy 11 1.31 5th 91
70-80 Fairly easy 14 1.39 6th 88
60-70 Standard 17 1.47 7th or 8th 83
50-60 Fairly hard 21 1.55 Some high school 54
30-50 Hard 25 1.67 High school or some college 33
0-30 Very hard 29 or more 1.92 or more College 4.5

Aim for 60 or higher. To increase your score, reduce the length of your sentences and words.

The wikipedia definition is quite interesting.

The Flesch Index formula uses different content metrics and multipliers. These multipliers change depending on the language. If Marfeel can’t detect the language it falls back to English multipliers.

const FLESCH_SCORE_MULTIPLIERS = array(
	'de' => [180, 1, 58.5],
	'en' => [206.84, 1.015, 84.6],
	'es' => [206.835, 1.015, 60],
	'ca' => [206.835, 1.015, 60],
	'fr' => [207, 1.015, 73.6],
	'it' => [217, 1.3, 60],
	'nl' => [206.84, 0.77, 93],
	'pt' => [248.835, 1.015, 84.8],
	'ru' => [206.835, 1.3, 60.1],
);

[$a, $b, $c] = FleschScoreMultipliers::get($language);

$fleschIndex = $a - $b * $wordsPerSentence - $c * $syllablesPerWord;
TROUBLESHOOTING:
There are several factors that could affect the Flesch Index value compared to other platforms:

Has Affiliation

If the body of an article contains an affiliation link Marfeel will set hasAffiliation=true based on these link detection rules