AI Titles and Topic Picking: Four Formulas and a Copyable Prompt

By the end of this article you'll be able to produce a batch of candidate titles with AI in ten minutes, and pick the one most likely to get tapped using a repeatable standard.

Titles drive click-through, but don't become clickbait

Readers see the title and cover in the feed first, then decide whether to tap. For the same piece of content, a weak title can halve the clicks. But keep two things separate: a title's job is to state accurately what the reader will get, not to trick them into arriving. Traffic you tricked in bounces immediately, and over time it drags down what the platform recommends you to.

Four reusable title formulas

• Result first: lead with what the reader gets. "Build a week of topics in three minutes" beats "Let's talk about my topic selection method."
• Audience + situation + method: say who it's for and when to use it. "First time making a RED cover? Fix it in this order."
• Counterintuitive claim + evidence: open with a judgment that runs against instinct, then explain why. "Posting daily isn't discipline, it's a topic problem."
• Specific number + boundary: the number supplies certainty, the boundary supplies credibility. "I use AI for titles, and it only works in these three cases."

The point of a formula is to guarantee the information is complete, not to make every post sound the same. Run the same sentence pattern week after week and readers tire of it.

Ask AI for ten at a time

Don't ask for one, and don't ask for fifty. Ask for ten, split into two groups: five on the safe track, clear and restrained; five on the sharp track, with a point of view and a bit of friction. Then pick one or two from each group and merge them by hand.

There are only three criteria: can it be understood at a glance, does it carry specific information, does it fit the account's voice. Anything that passes all three goes to the next round.

A prompt you can adapt directly

Copy the block below into the chat and replace the bracketed parts:

You are an editor for WeChat and RED. My account is aimed at [target readers] and covers [domain]. The core of this piece is [one-sentence thesis], and after reading it the reader should be able to [specific outcome]. Write 10 titles: 5 that are clear and not exaggerated, 5 that carry a point of view and a little friction. Keep each under 20 characters, and avoid words like "shocking" or "must-read." Then, on a new line for each, explain what structure it uses and which platform it suits.

That last instruction matters: making it explain the structure lets you spot the ones that were forced together.

Topics: a niche base, with trends on top

Even a great title fails on a bad topic. Stable accounts usually keep about 70 percent of content inside their domain and borrow trends for the other 30. The test for whether a trend is worth riding is "what does this have to do with my readers," not "is this trending." Force it and the people who arrive aren't your readers, and they won't come back.

Test instead of arguing

When you can't decide between two titles, don't hold a meeting. Test them. Use different titles for the same piece on different platforms, or republish later with a new title. Before republishing, check the platform's rules on duplicate content and follow the official guidance (as of 2026-11). Log impressions and click-through each time; after a dozen entries you'll see the pattern for your own account, which beats any generic advice.

Three common traps

• The body can't deliver what the title promised, and readers feel cheated.
• Shoving in keywords turns the title ungrammatical. Search may bring impressions, but nobody clicks a title that doesn't read.
• Using one title everywhere. WeChat and RED differ a lot in length and tone.

Title and cover work as a set. For cover guidance, see the tutorials. For how to adapt a finished piece into each platform's version, see this site's article on repurposing one draft.