A short scene, with people doing and saying things at a stated time, written as fact but with no detail that only someone who was there could know. The free checker at klype.ai/check flags it as "Reads like an invented story". This is the heaviest finding the checker makes: it deducts three times as much from the score as a formatting tell.
What it looks like
- Last month a founder messaged me at 11pm. He had spent $20,000 on ads and had nothing to show for it.
- She opened her laptop, searched for her own company, and stared at the screen. Nothing.
- Within two weeks, the team found three buyers asking for exactly what they sold.
Example text on this page is written to show the pattern. Figures and people in it are illustrative, not reported.
Why readers distrust it
A story is a claim that something happened. When the scene is tidy, the dialogue is neat and the turning point lands in the third sentence, readers who have seen generated posts recognise the shape. The missing piece is always the same: the odd, specific detail that a real memory keeps and a model does not supply. Without it, most readers decide the author asked for the story rather than lived it. This is the heaviest finding the checker makes, because an invented story is a false statement about the author's own life.
Why it shows up in generated drafts
Models have learned that LinkedIn posts with stories perform, so a request for a post produces one. The characters are generic, the timing is round and the resolution proves the point. None of it occurred.
How the free checker spots it
The checker splits the draft into sentences and looks for four narrative signals: an actor followed by an action verb ("the client asked", "she opened", "we found"), an elapsed-time phrase followed by a pronoun ("in seven minutes, we"), someone being told or shown something ("showed him", "told me"), and story time stamps ("last month", "back in March", "three years ago", "I remember"). Two or more sentences carrying any of these raise the flag, and the first two are quoted. A true story trips it as well, so the finding is a prompt to check, not a verdict.
The checker runs in your browser; nothing you paste is sent anywhere.
How to fix it
- If it happened, keep it and add one detail only you would know: the tool on the screen, what you got wrong first, what the other person actually said rather than the line that makes the point.
- If it did not happen, cut the scene and make the point directly. A plain claim you can stand behind beats a story you cannot.
- Never present a composite as a single event. "A pattern I see in client calls" is honest; "last Tuesday a client" is a claim about last Tuesday.
Before and after
Last month I asked a client to show me his best post. He couldn't find one. So we opened a tab and searched together. In seven minutes, we found three buyers asking for exactly what he sells.
Most founders I work with cannot point to a post that brought in a buyer, not because the posts are bad but because nobody checked. I now ask for the LinkedIn analytics export before the first call. It settles the question in about ten minutes.
The rewrite removes the pattern without adding anything the author could not stand behind.
Test your draft
The checker reports every pattern it finds, not only this one.
Common questions
My story is true. Why was it flagged?
Because the checker cannot tell. It sees the signals a scene produces, such as an actor and a verb or a time stamp, and flags any draft with two or more of them. Add a detail that proves you were there, then keep the story.
Can I use a made-up example in a post?
Yes, if it is labelled. "Imagine a founder who has spent $20,000 on ads" is a hypothetical the reader can judge as one. "Last month a founder messaged me" is a statement of fact. The problem is invention presented as memory.
What makes a real story read as real?
Detail that does not serve the point: the time of day, the wrong first guess, a line the other person said that was slightly off topic. Generated scenes include only what proves the lesson.