The pre-publish check for a how-to blog post: readability, keywords, meta tags and reading time

A finished draft is not the same as a publishable post. Between the last sentence and the publish button there are four quick checks that catch most of what makes a how-to article hard to read or easy to skip in search: is the prose pitched at the right level, is the target phrase used enough (or far too much), will the title and description survive Google's truncation, and what reading time should the header promise. Each has a tool on this site, and they work best in that order, because each edit changes the inputs to the next check.

This guide runs one real draft through all four. The post is a 1,240-word how-to on descaling a home espresso machine, aimed at the phrase "descale an espresso machine". The numbers below come from running the actual text through each tool, first as drafted and then after editing.

The draft, as written

Here is the opening paragraph exactly as the first draft had it:

"If you want to descale an espresso machine, you need to know that learning how to descale an espresso machine is one of the most effective maintenance procedures open to home baristas, because mineral deposits from hard water gradually accumulate inside the boiler and the group head and eventually compromise temperature stability, water flow and ultimately the flavor of every shot you pull. Many owners postpone the descaling process indefinitely, which is no surprise considering that manufacturers' instructions are frequently ambiguous, but an espresso machine that is never descaled will eventually require professional repairs that cost considerably more than a carton of descaling solution."

Nothing in it is wrong. It is just two sentences carrying 104 words, with the target phrase forced in twice in the first line. The second paragraph, the actual instructions, ran four sentences and 141 words in the same style. Those two paragraphs (245 words, six sentences) are the sample used throughout.

Pass 1: readability consensus

The Readability Consensus Score runs nine formulas over the same text and reports the median as a single grade, plus the spread between the highest and lowest. The spread is the part people skip and shouldn't: it tells you whether the formulas agree on why the text is hard.

Formula	Draft	Edited
Flesch–Kincaid	21.4	6.1
Gunning Fog	24.5	8.2
SMOG	20.7	8.8
Coleman–Liau	15.1	8.0
ARI	24.3	6.2
Linsear Write	46.5	7.2
FORCAST	12.5	9.8
Dale–Chall (band grade, from raw score)	16.0 (raw 10.8)	9.5 (raw 7.3)
Flesch Ease, as a grade	17.2	4.8
Consensus (median)	20.7	8.0
Spread	34.0, low agreement	5.0, moderate

The draft averaged 40.8 words per sentence and 1.79 syllables per word. Flesch Reading Ease came out at 14.1, deep in the "very difficult" band. The spread of 34.0 grades is mostly one formula. Linsear Write here scores only the sentences that fit in the first 100 words, and in this draft that window held a single 63-word sentence, so it shot to 46.5. (FORCAST also samples, using the first 150 words.) An outlier like that is a signal to look at sentence length first, and the lengths confirm it: 63, 41, 37, 26, 42 and 36 words.

The edit was mechanical. Split every sentence at its "because", "which" and "so that". Swap "accumulate" for "builds up", "approximately" for "about", "sufficient time to dissolve" for "time to soften". Drop the throat-clearing opener. The rewritten opening reads:

"Hard water leaves scale inside your espresso machine. It builds up in the boiler and the group head, and over time it slows the water, upsets the brew temperature and flattens the taste of every shot. Most owners put off descaling because the manual is vague. That delay costs more in the end: a repair shop charges far more than a carton of descaler."

Across both paragraphs the sample went from 245 words in 6 sentences to 180 words in 13, with an average sentence of 13.8 words. The consensus dropped from grade 20.7 to 8.0 and Flesch Reading Ease rose to 75.8. A grade 7 to 9 range is a sensible target for a general how-to. The edited text now lands between 4.8 and 9.8 on every formula, a spread of 5.0, with FORCAST, Dale–Chall and SMOG at the top and the vocabulary-light Flesch conversion at the bottom.

Two things in this tool are worth knowing before you trust the median. Dale–Chall does not produce a grade directly, so the tool converts its raw score to the midpoint of the matching band on Dale and Chall's table before taking the median: the final draft's raw 7.3 falls in the grades 9 to 10 band and enters as 9.5, and the original's raw 10.8 (college graduate) enters as 16. That makes it the second-highest score on the edited text. And "Flesch Ease, as a grade" is an approximation (20 minus a fifth of the Flesch score, capped at 18), which is why the draft's real Flesch figure on the first paragraph alone, minus 9.5, shows here as 18. For more on which levers move which formula, see how to lower a readability score.

Pass 2: keyword density, after the rewrite

Run the Keyword Density Analyzer after the readability edit, not before. Plain-language rewrites change which words repeat, and this one changed them in a way that mattered.

Phrase (sample)	Draft	First edit	Final
"espresso machine" (2-word)	5×, 2.31%	1×	2×, 1.31%
"an espresso machine" (3-word)	4×, 1.72%	0×	1×
"descale" (single word)	3×	0×	1×
Top single word	4-way tie at 3.40%, all flagged	water, 5.05%, over-optimization warning	water, 4.85%, flagged

The draft over-used the phrase: five uses of "espresso machine" in 245 words, and "espresso", "machine", "water" and "descaling" all crossed the tool's 3% warning line. The first edit fixed that and overshot. It set off the tool's over-optimization warning on "water" (5.05%), a word nobody was targeting, and the word "descale" disappeared entirely, leaving only "descaling" and "descaler". The final version adds one natural use back at the start of the instructions ("To descale an espresso machine safely, first check the manual...") and nothing more. That edit nudged the readability consensus from 7.7 to 8.0, which is a fair trade.

Read the percentages with the tool's arithmetic in mind. For single words, the analyzer removes stopwords before counting, so the denominator is the 103 content words left, not all 180. "Water" appearing five times is 4.85% here but under 3% of the full text. It also matches exact forms only: descale, descaling, descaled and descaler are four separate rows, so add them up yourself if you care about the root. On a short sample, one word is worth about one percentage point, which is why the flags are noisy until you paste the whole post. Google's spam policies define keyword stuffing by intent and give no percentage; the keyword density guide covers what a sensible range looks like.

Pass 3: title and meta description width

The SEO Meta Tag Length Checker estimates how many pixels the title and description take up, measured in Arial at 20 px for the title and 14 px for the description, and compares them with the limits SERP-preview tools use: 600 px for a title on either device (mobile wraps a long title rather than cutting it), and 920 px for a description on desktop, 680 px on mobile. Three rounds:

Version	Characters	Pixels	Desktop	Mobile
Draft title: "How to Descale an Espresso Machine: The Complete Step-by-Step Guide to Descaling Your Espresso Machine at Home"	110	1,070	long	long
Draft description: "Learn how to descale an espresso machine."	41	274	short	short
Final title: "How to Descale an Espresso Machine at Home, Step by Step"	56	545	good	good
Long description (175 characters)	175	1,061	long	long
Final description (127 characters)	127	790	good	long

The final description reads: "Scale slows your espresso machine and dulls every shot. How to descale it safely, which descaler to buy and how often to do it." A 137-character version in between scored 847 px. Both are "good" on desktop and "long" on mobile, which is the case the mobile setting exists to catch: a phone may cut the final version around "which descaler to buy". That was accepted, because the words that sell the click come first; trimming it to fit 680 px means losing either the descaler advice or the "how often", and a 95-character version that does fit ("...How to descale it safely and how often.", 601 px) read worse.

These are estimates, and the tool is open about it. It sums Arial's own character widths, so a title heavy in capitals or "W" and "m" runs wider than its character count suggests, which is the whole reason to measure pixels. Google's own documentation says there is no length limit on either tag and that results are truncated "as needed, typically to fit the device width", and Google frequently writes its own snippet from the page content instead of your description. Use the checker to keep the important words inside the first line, not as a guarantee of what appears. The post on pixel widths explains the limits in detail.

Pass 4: the reading-time label

Do this last, because every earlier pass changed the word count. The Reading Time Estimator divides words by fixed speeds: 150, 200 and 250 words a minute for slow, average and fast readers, and 130 for speaking aloud.

Full post	Slow	Average	Fast	Spoken
Draft, 1,240 words	8m 16s	6m 12s	4m 58s	9m 32s
Edited, 1,020 words	6m 48s	5m 6s	4m 5s	7m 51s

The edit cut 220 words from the full post, about 18%. The two sample paragraphs lost a quarter of their length, but the later sections had less padding. The header label moves from "6 min read" to "5 min read". That figure assumes plain prose at a steady speed. It counts no time for photos, a parts list or a table, all of which a how-to usually has, and instructions are read slower than a story because people stop to do each step. If your blog platform computes its own label at a different speed, use theirs for consistency.

The four passes as a checklist

1. Consensus first. Aim for a median around grade 7 to 9 for a general how-to, and treat a spread above 5 as a pointer to one specific problem, usually sentence length.
2. Keywords second. Check the target phrase is still present after simplifying, once near the top and once more where it reads naturally. Ignore stray 3% flags on a short sample.
3. Meta tags third. Keep the title "good" on mobile and the description under the mobile limit, with the key phrase early in both.
4. Reading time last, on the final word count, adjusted up if the post is heavy on images or steps.

The whole loop took about twenty minutes on this draft, and most of that was the rewrite in pass one. The tools only measure. What they are good at is pointing at the sentence that needs splitting or the phrase that went missing, so you spend those twenty minutes on the right lines.