"Write me a blog post" — why this prompt fails and how to fix it
A blog post prompt that returns generic filler, the four things it never specifies, and the rewritten version that produces something usable.
The prompt, as typed
Write me a blog post about our product. Make it engaging and high quality.
What it leaves the model to guess
- Who the reader is and what they already believe
- What the post is supposed to make them do
- Length, structure and section count
- What angle to take when several are possible
Graded before and after
Both versions below were scored by the same rubric the free prompt grader runs.
As typed
45/100 · grade D-
| Clarity of context | 2 |
|---|---|
| Structure | 3 |
| Specificity | 0 |
| Reusability | 0 |
| Decision logic | 0 |
Rewritten
66/100 · grade C-
| Clarity of context | 10 +8 |
|---|---|
| Structure | 6 +3 |
| Specificity | 10 +10 |
| Reusability | 6 +6 |
| Decision logic | 0 |
The rewritten prompt
You are writing for {{audience}}, who currently believe {{current_belief}}.
Topic: {{topic}}
Product context: {{product_context}}
Goal of the post: move the reader from {{current_belief}} to {{target_belief}}.
Write a 900-1100 word post with:
- An opening that states the reader's problem in their own words, no throat-clearing
- Three sections, each with a concrete example
- A closing that names one specific next step
Tone: {{tone}}. Channel: {{channel}}.
Constraints: no statistics you cannot source, no "in today's fast-paced world",
no more than one sentence of product mention before the final section.
If several angles fit, choose the one that most directly addresses
{{main_objection}} and explain in one line why you chose it.
This post succeeds if a reader who holds {{current_belief}} finishes it able to
state {{target_belief}} in their own words. Before you finish, check the draft
against that test and say whether it passes. Why the rewrite works
The rewrite names the reader, the belief change it is aiming for, the length, the section count and the angle rule. That removes four decisions the model was previously making on your behalf — which is why the first version reads like it could be about any product.
Common questions
- Why does this prompt score badly?
- It scores 45 out of 100 (D-) because it leaves the model to decide 4 things you already know: who the reader is and what they already believe; what the post is supposed to make them do; length, structure and section count; what angle to take when several are possible.
- Does a longer prompt always score higher?
- No. The rubric rewards named inputs, a fixed output format and an explicit rule for choosing between options. Extra prose without those adds length and no score.
- Will this rewritten prompt work in ChatGPT and Claude?
- Yes. It is plain text with placeholders you fill in before running, so nothing in it is specific to one model or vendor.