Draft short social posts with facts, limits and one clear action

Build a small set of posts from one offer without manufacturing scarcity, discounts or customer quotes.

Workflow guideEditorial update 2026-10-15

Tools for this workflow

ChatGPT, Claude

Start with this input

Fictional business: 陶屿工作室,零基础成人周末陶艺课。每班最多6人,每节2小时,周六周日各两班,198元含材料和烧制。老师示范并一对一修坯,作品由工作室上釉烧制,三周后可取。
禁止:保证作品不开裂、折扣、限时优惠、名额紧张、学员评价,以及“治愈”“绝绝子”“天花板”。
CTA:私信预约周末班次。
请写3条小红书文案,每条不超过200字;每条只围绕一个事实展开,不添加话题标签或emoji。

Follow the process

  1. Choose Social posts, your output language and the actual tool budget. Decide on the platform and length limit before generating.
  2. Give the model the exact price, schedule, included items and forbidden claims. Ask for three posts that each centre on a different fact.
  3. Count the complete text of each post, including punctuation, spaces, hashtags and emoji if present. Platform counting rules can differ; inspect the actual composer before publishing.
  4. Highlight any scarcity, discount or testimonial. Remove it unless the brief contains evidence and authorisation for that exact statement.
  5. Keep one action in each post and save the approved copy with the brief. Publishing and model generation are separate steps.

Reference output and acceptance criteria

Editorial reference post, not a model result:
周末想动手做点东西?陶屿工作室的零基础成人陶艺课,每班最多6人,每节2小时。老师示范并一对一修坯,198元含材料和烧制,作品三周后可取。私信预约周末班次。
Acceptance: price, class size and duration match the brief; no guarantee, discount, scarcity or testimonial was added.

If it goes wrong

If a post exceeds the limit, shorten introductions before dropping the fee or conditions. If three posts repeat the same angle, assign one to the class size, one to the making process and one to collection. Do not add urgency to force variety.

Time, cost and evidence

Use an eligible existing plan and check its current allowance. No publishing reach, engagement or generation cost has been measured for this editorial example.

This guide uses a fictional offer and an editor-written reference post. It is not evidence that any model performs better or that the post will attract buyers. Any measured product run is shown separately with its exact conditions.