LinkedIn post ideas that could only have come from your week
Paste a call, an article or a case study. PublishFlow finds the three to five moments in it that nobody else could have written, ranks them against each other, and tells you why the winner won.
- 1
The prospect went quiet when I asked what they had already tried
We tried that for six months and it did nothing.
Top of the five. A real moment, a quote nobody else has, and a claim a reader can disagree with.
- 2
We lost the deal on the question we did not ask
Strong, and it argues the same point as the first one from further away.
- 3
Nobody says “we tried that” about something that worked
Good line, thinner evidence. The moment behind it is one sentence long.
5 lessons from a discovery call
Topic, not a momentAnyone in your field could have written this without ever seeing your call.
The order is the product. It is not five scores for you to compare yourself.
You have read the lists. Fifty-five post ideas, a hundred prompts, thirty types of post that get engagement. You picked one, you wrote it on a Sunday night, and it read like something anyone in your industry could have posted. Because it was. The list had never met you, and it did not know what happened in your week.
A list of ideas cannot know what happened on Tuesday
Search for LinkedIn post ideas and you get lists. The whole category answers the question with volume: more prompts, more templates, more angles, ten ideas from your job title. It is a reasonable answer to the question people ask.
It is the wrong answer to the problem they have. A prompt that works for you works for everyone in your field, and a post that anyone in your field could have written is the exact post that does nothing for your reputation. The supply problem was never quantity of topics. It was that the good stuff, the thing you said on a call at 3pm that made a client go quiet, gets thrown away every week because nobody wrote it down.
PublishFlow starts from your material and throws away the topics. The test it applies to everything it finds: could a competent professional in your field have written this without ever seeing this? If yes, it is a topic, and it goes to the bottom no matter how strategic it sounds.
How it works
Give it something that actually happened.
A call transcript, an article you wrote, one you disagreed with, a case study, a link. Paste it and it works out what it is, and whether you were there or are reading it from outside. Roughly an hour of transcript fits.
It comes back with moments, not topics.
Three to five, each one holding what happened, the words that were actually said, and the part that is uncomfortable or contested. If the material only holds two real ones, it returns two rather than padding the list.
It ranks them against each other and says why.
Not five scores out of ten to compare yourself. One order, best first, with a line on each explaining what beat what. Ideas that turned out to be topics in disguise sink to the bottom however well they scored.
What one idea holds when you open it
Example- The moment
- Forty minutes in, you asked what they had already tried. They stopped, and the answer took a while to come.
- What was actually said
- “We tried that for six months and it did nothing.”
- The tension
- They had done the work. It still failed. Most advice in this field assumes the opposite.
- Who it lands on
- The reader who has already tried the obvious thing and has been told to try harder.
It flags the ones that are not yours to publish
When a moment rests on something a named client said, or on a conversation the other person thought was private, it gets marked before you write it. The flag does not change the ranking. It is there because that call is yours to make, and nobody can make it if nobody tells them.
Why the top one is at the top
Finding ideas is easy and worth nothing. What decides whether this is useful is what it throws away, so here is exactly how that is decided.
The discard test, asked twice
Every idea faces one question before anything else: could a competent professional in your field have written this without ever seeing your material? Then the same question is asked of the argument, because a specific story can still carry a conclusion everyone already agrees with. Fail either and it sinks, whatever its scores said.
Ranked, not scored
Scores out of ten cluster between 7 and 9, and an average of five of them to one decimal place is noise dressed as precision. So the ideas are compared to each other in one pass and put in an order, and the score you see summarises that decision rather than making it. You get told what beat what, and why.
It cannot make up a quote
Every quote it hands you is checked against your source before it reaches the screen, in code rather than by asking the model nicely. If the words are not in your material, they do not appear. That is the same check the writing side uses, and it is why a post can carry your actual phrasing instead of a smoothed-out version of it.
“Share a lesson you learned from a failure.”
A fine prompt. It is also sitting in front of every other person in your industry who read the same list this morning, which is what makes the post it produces interchangeable.
The prospect went quiet when I asked what they had already tried.
We tried that for six months and it did nothing.
One of these has a quote in it that only one person on LinkedIn can use.
Every idea, quote and verdict on this page is an example rather than customer data. This part of the product is young, and we would rather show you exactly how the ranking is decided than a number we made up.
What it will not do
The limits, stated plainly, so you can decide before you sign up.
It will not invent ideas from your job title.
No material, no ideas. There is a hundred character floor and it is enforced. If what you want is ten prompts from a role and an industry, our free idea generator does that, and it is a different job with a different result.
It does not watch trends, your feed, or anyone else’s posts.
It reads what you give it and nothing else. There is no scanning of what is working for other people this week.
It cannot quote what you did not say.
Which also means a thin source gives thin ideas. Ten minutes of small talk produces two weak moments, correctly.
It will not decide whether publishing is fair to the other person.
It flags the moments that rest on someone else’s words or a private conversation. You decide.
Saved ideas keep the argument, not the transcript.
Park an idea for three weeks and you can still write it, but the version you get on the day, with the whole source still in front of the writer, is the stronger one. Write the good ones while they are warm.
Anything past about an hour of transcript is trimmed.
Thirty thousand characters go in, the same window the writer gets. Long recordings are better split.
The ranking is a judgement, not a prediction.
It is an opinion about which post is worth your name on it. Nothing here forecasts engagement.
Works with
Meeting notetaker
Where most of the material comes from, without anyone having to remember to record.
The content engine
The same reading, running on a schedule, so Monday arrives with the week already drafted.
Voice notes
Two minutes out loud, for when the moment happened away from a call.
LinkedIn post generator
What happens after you pick one: three arguments from the single idea, in your voice.
Questions
Where do the post ideas come from?+
From material you provide: a meeting transcript, an article, a case study, a link. PublishFlow reads it and pulls out the specific moments a post can be built on. It does not generate ideas from a topic or a job title.
How many ideas do I get?+
Three to five per source, usually five. If the material genuinely only holds two good moments, it returns two. Padding the list with generic topics would defeat the point.
How does it decide which idea is best?+
It compares them against each other in one pass and puts them in order, then tells you why each one placed where it did. Ideas that could have been written by anyone in your field, without ever seeing your material, are pushed to the bottom regardless of their scores.
Can it make up a quote or a statistic?+
No. Every quote is checked against your source text before it is shown, and anything that is not literally there is removed. The check runs in code, not in the prompt.
What if the material involves a client?+
Ideas that rest on a named client, a prospect, or a conversation the other person thought was private are flagged before you write them. The flag does not change the ranking, and PublishFlow never publishes anything without your approval.
Can I save an idea for later?+
Yes. Saved ideas sit in your library until you write them. They keep the argument and the source quote rather than the whole transcript, so the strongest version of a post is still the one you write while the material is in front of you.
Do ideas count against my monthly limit?+
No. Finding and ranking ideas is unmetered. The monthly limit applies to the posts you generate from them.
The list of 55 ideas has never met you
Fourteen days free. No credit card. Paste one call and see what it finds.
Start freeFinding and ranking ideas does not count against your monthly posts