The Pre-Publish Checklist: What to Check and How to Measure the Effect

A story is not finished when it is published; that is when it starts. Most texts that are well written and still fail break in one of two places: before publishing, where nobody reviewed them through a reader's eye, or after publishing, where the writer looked at one number without context and drew the wrong conclusion.

This article belongs to the story is everything series and has two jobs: a short checklist that catches the recurring errors before every release, and a way of reading feedback and metrics without fooling yourself.

One note: there are no figures or percentages here. Metrics mislead without context, and public content averages have nothing to do with your reader. What follows is the relation between a goal and a signal; the number you take from your own data.

The pre-publish checklist: six questions in the reader's order

Ask these six in the order the reader moves. If one answer is no, stop there; moving to the next question only hides the error.

• Title: does it make a promise or name a topic? A title that names a topic invites nobody to read a text they already know.
• First sentence: does it put the reader inside a situation, or start with a definition and a warm-up? If removing the first sentence leaves the second one meaningless, the first was a warm-up.
• One change: is there exactly one transformation? Texts that carry two or three changes finish none of them.
• Detail: is there one detail you can touch — a name, a time, a small number, a line somebody said — or only large adjectives? "A fast service" is not a detail; "the receptionist called every evening" is.
• Ending: does it close on a decision or a turn, or on a repeated summary? If you delete the last paragraph and nothing is lost, delete it.
• Ask: one clear step, or three vague ones? "If you try this, tell me what happened" is an ask; "get in touch with us" is not.

A useful habit: read the checklist bottom to top. Starting from the end means your mind is no longer inside the writing, and it sees the text more like a reader.

Reading aloud: the ear test

Read the text out loud, ideally to one person. The ear catches what the eye skips: nested sentences, hidden repetition, extra politeness, a flat rhythm. Every place you pause or run out of breath is an edit point.

Mark three things while reading:

1. The sentence that gets lost halfway: if the listener cannot remember where it began, split it or move the verb forward.
2. Pointless repetition: if a word or idea appears twice in one paragraph and the second time adds nothing, the second one goes.
3. The manufactured joint: a sentence that exists only to connect two parts ("as we said earlier") and whose removal breaks nothing.

The same test works for audio: if you are preparing a script for a podcast, reading aloud is the only way to check the rhythm. The differences between writing and speech are covered in platform style.

The cut test: what can go while the story stays intact

Cutting is the most common editing tool and the least used, because the fear of losing something blocks it. The method: from each section, remove one paragraph and see whether the meaning breaks. If it does not, that paragraph was decoration.

Three things must survive every cut: the change, the detail that makes the change believable, and the decision. Everything else is negotiable. In practice, most cuts come from five categories:

• Explaining what the reader already understood.
• Large adjectives standing where a small detail should be.
• Warm-up before the scene arrives.
• A closing summary with nothing new in it.
• A sentence pointing at itself rather than the subject ("in this section we will see…").

Another cutting rule: read each sentence alone and ask "what would change if this were gone?" If the answer is unclear, the sentence was probably there for the writer's comfort, not the reader's.

Title and image: the promise of the story, not its summary

Title and image together make a promise, and the text keeps it. A title that summarises everything removes the need to read; a title that says nothing gives no reason to. The middle ground: the title creates a question or a tension the text answers.

Three title tests:

1. Standing alone: read it by itself. Does it say something, or only mean something next to the text?
2. Honesty: does the text deliver what the title promised? A title that promises more than the text spends the trust of the next reader.
3. Specificity: does the title hold one concrete detail, or is it built from generic words?

The image is part of the narrative too, not decoration. It should be a frame from the same scene, not a generic and unrelated photo. If you swap the image and the text is unchanged, the image was doing no narrative work. And if the image gives away what the text saves for the end, pull the image back.

Measurement: what to watch, and what misleads

A good metric has one property: it connects to a decision. Before looking at any number, ask: "if this goes up or down, what will I do?" If there is no answer, the metric is decorative and is better ignored.

Healthy signals. Completion, or reaching the last section, on long texts; saves, which signal future value; replies that quote a specific line; brand-name search; the same reader returning; and word of mouth — someone saying in a meeting "I read that piece". These connect to real reading.

Misleading signals. Likes and views without context, follower counts, shares without reads, and comparing two texts with two different goals. A short entertaining post and a deep technical article do different jobs; comparing their raw numbers says nothing. Comparison only means something between two versions of one text with one goal.

One more point: dwell time on a long text is naturally higher. That difference is about format, not quality. Always interpret a metric against its container, never alone.

Rewriting the story from feedback

Feedback has three useful shapes and one useless one. Useful: the place where a reader stopped; a question several people asked separately; a misunderstanding they took from the text. Useless: "it was good" — that is not feedback, it is politeness.

Rewriting from those three follows three rules. First, repair in the right place: if the reader left at the third sentence, the problem is usually in the first, not the middle. Second, do not explain the missing part; turn it into a scene. Third, keep the previous version; comparing two versions is the only way to know whether the change genuinely helped or merely made things different.

And a warning: pleasant feedback does not prove the text is right, and thin feedback does not prove it is worthless. If the piece was written for a small audience, quiet numbers are natural. Trust the qualitative signals from that small audience, not the general noise. To keep the story aligned with a sales goal, see storytelling for sales.

The metrics table: goal, healthy signal, misleading signal

Fill this table in for every text; if you cannot fill the second column, the goal is probably unclear.

The last row matters for a series: if the reader does not move from the first article to the second, the map is not working, even when each article is good on its own. Inline linking and publishing order are covered in from long text to a series.

The thirty-day programme

Thirty days is enough to see a full loop: write, publish, gather feedback, rewrite. Start the programme with one text, not five.

1. Week one — build: write one reference story that holds every accurate detail, then make two native versions of it. Run the six-question checklist on both.
2. Week two — publish: ship the first version and hold the second for another day. This week, collect only qualitative feedback: where people stopped, what they asked.
3. Week three — rewrite: rewrite the title and the first sentence from week two's feedback, and publish the second version carrying that lesson. Put the two versions side by side.
4. Week four — review: fill the metrics table, write one decision per metric, and choose a single thing for next month: the title, the ending, or the ask.

At the end of the month what you have is not a better text; it is a habit: writing from a map, reviewing by ear, cutting without fear, and reading metrics with one clear question. That habit is what separates content that gets published from content that gets read.