Your analytics tell a story you do not want to hear. The post that took you a full afternoon has an average time on page of 44 seconds. You pictured readers nodding along at paragraph eight. They never reached paragraph three, and the ones who did skim straight to the bottom, hunting for something they could not find.
The temptation is to blame the audience, the algorithm, attention spans. Usually the cause is structural. A post gets finished when its shape carries a reader from promise to payoff without a predictable stretch in between, and nearly every abandoned post failed at one of three joints: the title, the opening, or the subheadings. None of this requires talent. It requires treating structure as the craft.
The title is a contract
A title makes a promise, and the post has exactly one obligation: keep that promise, completely, with nothing else in between. "How to Fix a Leaky Faucet in Four Steps" commits to a procedure, four steps, and a leaky faucet. A reader who lands and gets four hundred words of plumbing history leaves because the contract broke, not because she is impatient.
| Title shape | The promise | How posts break it |
|---|---|---|
| "How to X in 4 steps" | A procedure, numbered, complete | Burying step one under 400 words of background |
| "Why X fails" | A diagnosis with evidence | Pivoting halfway into a product pitch |
| "7 tools for X" | Seven tools, actually compared | Six thin entries and a padded seventh |
| "X vs. Y" | A framework for deciding | Describing both and refusing to recommend either |
Write the working title before you draft and let it discipline the piece. Any section that is not part of the promise is a candidate for cutting or for its own post. When the draft wanders, which drafts do, rename the title afterward so it describes the post you actually wrote rather than the one you planned.
The same promise logic extends to the address bar. A title like "Price Your Freelance Work by the Project" should produce a URL carrying those words, not "blog-post-17" or the phrasing of an abandoned earlier draft; a slug generator keeps the URL and the promise in step. Our guide to URL slugs and SEO covers why the search engines care too.
The intro has one job: earn paragraph two
Nobody owes you a second paragraph. A reader decides within one screen whether this post is for them and whether it will deliver, and the Nielsen Norman Group's eye-tracking studies have shown for years that web visitors scan in an F-pattern: the first two paragraphs get real reading, then the eye drops down the left edge hunting for a reason to continue.
So do not warm up in public. Cut the dictionary definition, the history of the topic, the sweeping claim about modern life. Open on the reader's situation or a concrete scene that implies it.
Openings matter enough that we have a full piece on compelling opening paragraphs. One honest caveat: if you struggle to write intros, stop writing them first. Draft the body, then return and write the opening that matches what the post turned out to be.
Your subheadings should tell the story on their own
Here is the most useful test in web writing, and almost nobody runs it. Copy only your H2 headings into a blank document and read them top to bottom. They should form a coherent outline of the entire argument, one a skimmer could follow and act on without reading a single body paragraph.
"Tips" and "Conclusion" carry zero information. If a subheading would work equally well above somebody else's article, rewrite it. This is not decoration for diligent readers. For skimmers, the subheadings are the post, and for finishers they are the progress markers that say how far away the payoff is.
Open loops, and where the payoff goes
An open loop is a question you raise early and answer later, with a visible promise that the answer is coming. "The exact script for the pricing conversation is at the end of this section, but first you need the number" gives readers a reason to stay through the setup instead of asking them to trust that you will get to the point eventually.
The order of payoffs matters as much as the loops. Newspapers inverted the pyramid because readers historically left mid-article: most important fact first. A blog sits between pyramid and serial. Deliver one genuinely useful thing in the first third, which proves the post keeps promises, then escalate so each section is worth more than the last and quitting gets progressively more expensive. A post that spends its best tip in paragraph two and coasts afterward trains readers to leave.
Never bury the promised answer below the fold without a marker that it exists. That is precisely how 44-second averages happen.
Cut everything the reader can predict
Finishing is a subtraction problem more than an addition problem. The sentences readers skip are the ones they could have written themselves: "In this post, we'll explore..." "Without further ado..." "There are a few things to keep in mind." A sentence that contains no information the reader lacked is a toll booth on the road to your point, and our guide to cutting filler words keeps the full hit list.
Which brings us to the honest word about word count: the number itself means nothing. A 400-word post that keeps every promise gets finished more often than a 2,500-word post that pads. The right length is whatever the promise requires plus zero predictable sentences. Use a word counter only to check that your draft has not wandered to twice the size of its promise, because at that point you do not have a long post. You have two posts holding hands.
Rhythm carries readers through whatever length the promise legitimately needs. Uniform sentence length flattens prose into a hum the eye skims, while alternating long builds with short punches keeps attention sentence by sentence, which is the whole subject of our piece on sentence length and rhythm. A readability checker will surface the worst offenders: the forty-word sentences with three subordinate clauses each.
End with one action, not a summary
The ending is the highest-trust moment in the entire post. The reader who arrives there has verified that the title, the intro, and the subheadings all kept their word. Do not spend that moment recapping what they just read. A summary tells readers the post is over. An action tells them what it was for.
Choose the single next step you would hand a friend who read the whole thing: "Open your analytics right now and find the time on page for your three most recent posts." One verb, one object, ten minutes of effort. A reader who takes the action remembers the post, while a reader who reads a recap remembers nothing, because the recap competed with everything above it and added nothing.
Frequently Asked Questions
How long should a blog post be?
As long as its promise requires. A faucet-repair post can deliver in 600 words; a genuine comparison of four systems needs more. The failure mode is never length itself but predictable sentences inside the length. Draft toward the promise, then subtract whatever the reader could have written themselves.
Should I write the introduction first?
Usually last. The intro's job is to match the body, and you do not know precisely what the body became until it exists. Write a working title and one sentence of promise to get moving, then return and draft the opening that the finished piece earned.
How many subheadings do I need?
One per beat of the promise, roughly. Most posts between 800 and 2,000 words land between four and seven. If two subheadings cover the same ground, merge them; if one subheading hides three distinct arguments, split it.
Do readers actually reach the end?
Fewer than you would like, which is exactly why the end should hold one clean action rather than a recap. The skimmers who arrive at your final section by way of the subheadings are often your most engaged readers, because they went looking for the payoff.
What is the fastest fix for a post nobody finishes?
Run the subheading test. Paste the H2s into a blank document and read them as a paragraph. If they do not tell the story, no amount of sentence polish will save the draft; restructure the sections first, then rewrite the transitions.
The subheading test
Before you publish anything else, run one check on the draft currently sitting in your editor: copy its subheadings into a blank document, read them top to bottom, and ask whether they alone would teach a stranger something. If they would, your structure is carrying the reader and everything from here is polish. If they would not, you now know why the analytics say 44 seconds, and you know exactly which layer to fix first.