Do not index
"My reach fell off a cliff and nobody can tell me why. Did someone flag my posts?"
That question started landing in my inbox the same week LinkedIn shipped its AI slop reporting option. The short answer is that yes, a flag costs you distribution, and no, the flag is not your real problem. According to Forbes reporting on the rollout, "anyone can now report your LinkedIn post as AI slop, and a flag costs you reach." What that sentence hides is the part that matters. The people pressing that button are not running a detector. They are reacting to a feeling. Bad human writing produces the exact same feeling as competent machine writing.
So distribution on the largest professional network in the world is now partly decided by whether strangers think your writing sounds like a machine. Not whether it is. Whether it sounds like it.
I run a content operation and I edit human writing every day. Drafts from founders, drafts from ghostwriters, drafts from operators who type fast and think faster. From that seat I can tell you the markers people flag are almost never proof of AI. They are proof of a writer hedging, padding, or reaching for a shape they have seen work for someone else.
This matters most for agency owners between $200k and $2M in revenue who post under their own name and under client names at the same time, and for ghostwriters charging $5k to $30k per month who now carry a reach risk that shows up in no dashboard and gets blamed on them personally. It matters for founders running personal-brand content where a single account sits at the top of the pipeline.
Skip this if you post twice a month for the record and do not care about reach. Skip it if your pipeline is entirely outbound and LinkedIn functions as a credibility check rather than a channel. And if you are still buying engagement or running a generator that produces five posts from one prompt, this article will not change your model, because the flag is the least of what is wrong with it.
The Fingerprint Test
What I use with my team is what I call the Fingerprint Test. Before a draft ships, it has to carry at least one thing that could only have come from the person whose name is on it. A number from their actual P&L. A client objection phrased the way the client phrased it. A decision they regret with the month attached. A price they charged and the reason they raised it.
The test is deliberately narrow. It does not ask whether the writing is good. It asks whether the writing is traceable. A machine can produce a clean sentence about retainer churn. It cannot produce the sentence about the $18k retainer you lost in March because a junior writer handled the client's second quarter and nobody caught the tone drift until the founder did.
Traceability is what readers register as human. Not sentence length. Not the absence of the word leverage. When a post carries one traceable detail in the first four lines, nobody reaches for the report button, because the post has already answered the question the button exists to ask.
What actually gets flagged on LinkedIn
Here is what my team strips out of drafts, and it will look boring, because the real markers are boring. Symmetrical three-part lists where the third item was invented to complete the pattern. Openings that restate the title. Sentences that qualify a claim into meaninglessness. Transitions that announce a turn instead of making one. Conclusions that summarize rather than land. None of those are AI signatures. They are the signatures of a writer who had nothing specific to say and filled the space anyway.
What we replace them with is smaller and narrower. One example instead of three. A number instead of an adjective. A named situation instead of a client. The draft usually gets 30 to 40 percent shorter and reads twice as confident, because confidence in writing is mostly specificity that has not been sanded down.
This is the same discipline that keeps client work from drifting, and it is why quality control on a content retainer has to live at the draft stage rather than the approval stage. By the time a post reaches the approval queue, the traceable detail either exists or it does not, and no editor can add it after the fact without inventing it.
The reporting button is a symptom. What sits underneath it is that a large share of professional writing has become interchangeable, and readers have gotten fast at spotting interchangeable. Platforms are now handing them a lever to act on it. That lever will get more sensitive, not less.
For anyone whose revenue depends on a personal account, that changes the shape of the bet. Volume was the winning strategy for roughly four years, because distribution rewarded frequency and the cost of a mediocre post was zero. The cost is no longer zero. A flagged post suppresses the account that published it, which means your worst post is now doing damage to your best one.
The operators who come out of this ahead will cut output and raise the traceability floor. Fewer posts, each carrying something that could only have come from one person. That is a harder operation to run and a much harder one to copy, which is exactly why it will still be working after the next three platform rule changes have come and gone.
