Your preferred hook leads, it does not fill the set
The hook style you pick from the dropdown is used for the first introduction. The other 3 each take a different style from the list, and no style repeats inside a run.
That is deliberate. Picking "statistic" and getting 4 statistic openings would tell you nothing you did not already know when you picked it. Getting a statistic opening first, then a problem, a contrarian and a direct version of the same intro, tells you whether the number was actually the strongest way in. Often it is not. The preference orders the set; it does not define it.
If you genuinely want 4 variants of one hook, this is not the tool for that. It refuses by design.
The 6 hooks a blog intro generator uses and when they fit
Question opens with the question the reader is already asking, in their words. Strongest on search traffic, where you know the query that brought them.
Statistic opens on a concrete, checkable number. Powerful when the number is surprising, dead when it is the kind of stat every post in the category recycles.
Story opens mid-scene with a specific person, moment, or decision. No setup, no "let me tell you about the time." It starts inside the scene.
Contrarian opens by contradicting advice the reader has already been given. It needs a real default belief to argue with, not a strawman.
Problem names the exact failure state the reader is stuck in. This is the workhorse for how-to and troubleshooting content.
Direct leads with the answer or the claim and skips setup entirely. Best when the reader arrived impatient and the post has something specific to say.
The output panel gives a shorthand worth keeping: problem and question hooks suit search traffic that arrived stuck, while contrarian and story hooks suit a newsletter or social audience that already trusts you.
What the 5 fields control in an introduction builder
Post topic (required) is the actual subject. "How to run a content audit" gives the model a scene to write into.
Target keyword (required) is the phrase that has to appear naturally in the copy. It gets worked into all four, and lands in the first sentence of at least 2.
Audience shifts vocabulary and assumed knowledge. "In-house SEO leads at B2B SaaS" produces an opening that skips the definition of a content audit.
Tone is the standard tone list, defaulting to conversational. It changes rhythm, not structure.
Preferred hook seeds the first introduction, as described above.
Why 60 to 110 words is the working range
Every introduction targets that band, and the card flags anything outside it.
Under 60 words you have not established stakes. The reader gets a claim with no reason to believe it and scrolls past the first H2 looking for the actual answer. Over 110 words you have pushed the first heading below the fold, which on mobile means a reader who is deciding whether to stay sees only preamble.
The count on each card is the model's own number when it supplies a usable one, and a local word count of the paragraph when it does not, so the figure always matches the text you are looking at.
Keyword placement in the first sentence
Every card shows whether the target keyword lands in the first sentence, and at least two of the four are written so it does.
That flag matters more than a density number. The first sentence is what a reader checks against the query they just typed. If they searched "content audit checklist" and the first sentence talks about content strategy in general, the mismatch registers immediately even if the keyword appears twice in paragraph 3.
The rule the tool works under is that the sentence comes first. It will place the keyword early when the sentence still reads naturally and skip it when placing it would bend the line. 2 of 4 with the keyword up front, written well, beats 4 of 4 written awkwardly.
What the blog post tool refuses to write
There is a banned-opener list, and it is enforced. "In today's digital landscape," "in the ever-evolving world of," "whether you're a beginner or a pro," "let's face it," "let's dive in," "picture this," "it's no secret that," "we've all been there," and "in this article, we'll explore" are all blocked, along with near-variants.
One more rule does most of the work: it is not allowed to open by explaining that the topic is important, popular, or growing. That single move accounts for a large share of forgettable introductions. Open on the specific thing, never on the category.
It also will not summarize what the article covers or close on "here's everything you need to know." An introduction earns the next paragraph. It does not preview a table of contents.
The one thing you have to check yourself
The statistic hook is the risk. The tool has no live data, no index, and no search access. It is instructed to use only figures it is confident about and to keep them checkable, but "confident" is not "verified."
Treat every number in a statistic opening as unsourced until you have found it. If you cannot find it in 5 minutes, use one of the other three introductions. This is the single most common way an otherwise good AI-assisted draft ships something wrong.
The same caution applies more mildly to story hooks. A named person or company in a story opening is invented unless you supplied it.