Opinionated Thought Leadership Content Should Cost You Followers

2026-09-25 · The Prowir Team

Someone unfollowed you after that post. That is not a failure state. That is the instrument working.

Most professionals read a drop in follower count as a verdict on the post. It is closer to a receipt. Opinionated thought leadership content works by subtraction: it tells part of your audience, clearly, that you are not for them. The people who leave were never going to hire you, refer you, or argue with you usefully. They were inventory.

The trouble is that every dashboard you look at is additive. Followers, impressions, reactions, profile views — all of them count arrivals. Nothing counts departures, and nothing tells you whether a departure was good. So the feedback loop quietly rewards the post that offends nobody, which is also the post nobody remembers.

The safest post in your feed is the one nobody can disagree with

Look at what that incentive has produced at scale. Originality.AI sampled 5,000 public LinkedIn posts of at least 100 words in July 2026 and classified 81.2% of them as likely AI-generated, up from roughly half of sampled long-form posts in late 2024. The interesting part is not the volume. It is the narrowing. Models are tuned to be agreeable, so the median professional post has converged on a register that is fluent, balanced, and completely unfalsifiable.

LinkedIn has noticed. The platform is rolling out a "Seems Like AI Slop" option in the three-dot menu alongside "Not interested" and "Report post," and its chief product officer has called giving members that signal a top priority. Readers are being handed a button for a feeling they already had.

Here is what that means for you. A post nobody can disagree with costs you zero followers and buys you zero standing. It is free, and it is worthless. The unfollow is the price of being worth something, and it is cheap.

Opinionated thought leadership content has to be arguable

A working test: could a competent, informed peer in your field read this and say, flatly, "no, that's wrong"? If not, you wrote a description, not a position.

What that looks like across different kinds of work:

• A physical therapist: most running knee pain is a hip problem, and the imaging pathway patients get routed through has no incentive to say so.
• A supply chain lead: dual-sourcing is theater when both suppliers sit inside the same 200 miles and clear the same port.
• An architect: open-plan was a cost decision wearing a culture costume, and the profession agreed to pretend otherwise for a decade.
• A data engineer: most of your dashboards are load-bearing decoration. Name one decision that changed because of one this year.

Each of those loses somebody. The imaging clinic. The procurement director who just signed the second contract. The firm that built its practice on open floor plates. That is the mechanism, not a side effect. You cannot become known for a position without being unpopular with the people that position implicates.

Not every unfollow is a compliment

This is where the advice usually goes wrong, so handle it carefully. Three different things cost you readers, and only one of them is worth paying for.

Filtering. You got specific, and the wrong-fit audience left. Pay this gladly.

Friction for sport. Contrarianism with nothing behind it — a take you would abandon the moment the room turned. This costs more than followers. It costs the credibility you were trying to build, and experienced readers can smell it in one line.

Collateral. You aimed at a person instead of a practice, or you were careless with a claim you could not support. That is not nerve. That is a cleanup.

The diagnostic is simple. Could you defend the post, in the same words, in a room containing the people it is about? If you would soften it in person, it was a performance. If you would say it exactly that way, you have a position, and the unfollows are just the bill.

Why this became load-bearing

Two years ago, a competent and agreeable post still signaled effort. Somebody sat down and wrote it. That signal is gone. In a feed where four out of five long posts read like machine output, fluency proves nothing on its own, and the presence of something contestable has quietly become the evidence that a person made a decision.

This is also the exact line AI cannot cross on your behalf. A model will draft, structure, tighten, and help you distribute, and it is genuinely good at all four. It cannot decide what you are willing to be wrong about in public, because that decision requires having something at stake, and it has none. We say this plainly because the alternative is pretending otherwise, and the feed already shows what pretending produces.

So stop grading a post on what it added. Ask what it cost you, and whether the people it drove off were people you wanted.

If the hard part is finding time to turn positions you already hold into published work, that is the problem we built Prowir to handle. The beta is open.