After a year of collaborating with AI on content, my biggest takeaway is: the outline determines the upper limit of article quality, AI is only responsible for expanding, not thinking. Let AI write without an outline, and the output is always “correct but mediocre” content — identical structure, missing viewpoints, unclear data. Later I set a rule for myself: spend ten minutes filling out an outline before writing, think through four questions clearly before starting.
Figure: Content Outline Template: 10 Minutes Before Writing in the AI Collaboration Era (compiled by YunyingGO)
Four questions the outline must answer
My outline template is very simple, just four blocks: who it’s written for, why they should trust it, what it says, how it’s arranged.
“Who it’s written for” is the target reader of this content — for example, foreign trade operations newcomers, small team leaders. The more specific the reader profile, the more targeted the content. “Why they should trust it” is the unique angle and credibility source of this content — points others haven’t covered, or your own practical cases. This is what AI can’t provide. “What it says” is the core conclusion — one sentence making clear what this article conveys. “How it’s arranged” is the structural skeleton — H2/H3 list, what each paragraph covers.
Among the four questions, “why they should trust it” and “what it says” are the most critical. If these two aren’t filled well, no matter how well AI writes, it’s just a流水账.
How to find a unique angle
This is where many writers get stuck, I’ve总结出 four methods. Switch stance: others talk about “how to do it,” you talk about “what happens when you do it wrong.” Switch subject: others talk theory, you talk about “our actual operational process.” Add constraints: “beginner,” “foreign trade,” “small team,” “limited budget” — the more constraint words, the more unique. Counterintuitive: “don’t touch high-search-volume keywords first” — counterintuitive conclusions are easier to remember.
Before writing each time, I first search the target keyword, see what angles the top-ranking articles use, then deliberately avoid them — without overlapping angles, there’s a reason to be seen.
Handover spec from outline to draft
After the outline is filled, how to hand it to AI also matters. My approach: directly attach the outline in the instructions to AI, clearly state “expand each H2 to 150-200 words, don’t add new sections, don’t invent data.” AI is the executor, the outline is the baton — write instructions clearly, and the first draft output is basically usable.
For data, I explicitly don’t let AI provide it — have it mark placeholders where data should go, and I fill in real data afterward. Viewpoints, conclusions, and cases — these three are always written by myself, never handed to AI. Opening and closing also written by myself, to keep the human touch.
Outline template you can copy directly
Make the four blocks into a fixed table, fill in the blanks before each writing session — much faster than rethinking the structure every time. The “why they should trust it” column in the table is the easiest to skip, but it determines whether the article has human touch — must fill it yourself.
| Block | Question to answer | How to fill |
|---|---|---|
| Who for | Target reader profile | More specific better, e.g. foreign trade beginner |
| Why trust | Unique angle or case | AI can’t provide, write yourself |
| What to say | One-sentence core conclusion | Conclusion first, then expand |
| How to arrange | H2/H3 structural skeleton | List what each paragraph covers |
Practical process to fill in ten minutes
In practice, my process is: first search the target keyword and look at results pages, two minutes; set target reader and unique angle, three minutes; list structural skeleton and core conclusion, three minutes; mark data points to use, two minutes. Complete within ten minutes, then start writing or hand to AI.
A few times I got lazy and didn’t fill the outline, started writing directly, and regretted halfway through — mediocre angle, scattered structure, data not prepared either. After that I never skipped the outline step again.
Three pitfalls to avoid in outlines
Pitfall one: only list keywords not conclusions — AI expands without viewpoints, after reading the whole article you don’t know which side you’re on. Pitfall two: leave “why they should trust it” blank — the whole article becomes a correct but mediocre summary, anyone could write it. Pitfall three: structure written as you think — eventually sections overlap heavily, readers leave after two paragraphs. After filling the outline, read it yourself — if someone else could follow it to write an article on the same topic, it’s qualified.
- Among the four questions, “why trust” and “what to say” are most critical, fill first.
- Have AI leave placeholders for data positions, fill real numbers yourself afterward.
- Write opening and closing yourself, keep human touch, don’t hand to the model.
Relationship between outline and writer brief
The outline is your own thinking framework, the writer brief is the execution instructions handed to collaborators or AI — the two are一脉相承. Expand the four outline blocks directly into a brief: reader profile for whom, unique angle clarifying the切口 different from top-ranking articles, core conclusion as one sentence, structural skeleton broken into paragraph-by-paragraph key points. The more specific the brief, the less likely the person taking over drifts, first draft rework rate drops from 50% to 10%. Build the habit of outline first then brief, and the ceiling of content quality gets raised a layer first.
- Mark word count limits per paragraph in the brief, prevent AI from bloating any section.
- List real data to use separately as attachments, don’t let the model fabricate.
How to验收 after the outline is filled
Don’t rush to hand it over after writing the outline. Play the reader yourself and read it once: can you say in ten seconds who this article is for, what it answers, why they should trust it. If it can’t pass this, rework the outline — saves three times the time vs. tearing down after the first draft. Nail this验收 action into the process, and the floor of team output immediately gets raised — even the worst writer can’t produce a completely off-topic draft. The验收 checklist is just three items: specific reader, clear conclusion, angle doesn’t collide with top-ranking — missing any one, send back for rewrite.
Spending ten minutes filling an outline before writing equals installing a steering wheel on AI. An AI car without a steering wheel, no matter how fast it drives, won’t reach where you want to go.
