I said something in a meeting in September that turned out to be wrong. So I pulled the numbers.
On September 22, 2026, our leadership team was talking about LinkedIn. Brandon Checketts said automated posting had done nothing for his team, while his own posts got replies from people he actually knows. I agreed, and added my own proof: my original post on Amazon blocking AI agents was my most-seen post ever. Reposts and generic AI content, I said, get very little.
Nobody checked. It sounded right.
On November 3, I opened my LinkedIn analytics and pulled my 50 most-seen posts of the past 365 days.
That post on Amazon blocking AI agents got 398 impressions. My top post of the year got 15,332.
Where the reach actually went
I sorted all 50 posts into four groups by what they were about:
| Type of post | Posts | Median impressions | Engagement rate |
|---|---|---|---|
| Events and people I met in person | 13 | 489 | 3.5% |
| Built on something only we had (our own data, a customer’s question, a bug I paid for) | 17 | 403 | 1.5% |
| Generic AI takes and news hooks (excluding the top two) | 16 | 302 | 1.1% |
| The top two: AI news headlines | 2 | 24,124 combined | under 0.1% |
Engagement rate here is reactions plus comments, divided by impressions.
My two biggest posts of the year were both riffs on AI news headlines. Together they got 24,124 impressions. They got 22 reactions and comments between them.
The post on Amazon blocking AI agents landed below the median for reach. It also drew 8 comments, more than the post with 15,000 impressions.
So Brandon and I were each half right. Generic content can win the reach lottery. It just doesn’t start conversations.
The lottery may be closing
Both of those big posts are about eight months old. They went out before LinkedIn started detecting AI-generated content and limiting its reach in May 2026. Since then, my best-reaching post is a first-hand one: 3,532 impressions in its first week, built on our own numbers.
My data can’t prove the crackdown caused that. Other people’s data points the same way, though. LinkedIn has reportedly said posts it classifies as AI slop get about 40% fewer views. LigoSocial compared 134,522 posts and found that the ones that sounded AI-written got 22% less engagement than the same author’s other posts.
Alex Hormozi put the alternative in six words: “We have to focus on proof.” His point is that anyone can produce content now, so what wins is evidence of real work that nobody else could have shown you.
What starts a conversation
Look at what my best-engaging posts have in common. A question a person asked me at a lunch-and-learn. The exact error screen an AI shows when it can’t read your Amazon listing. What one customer pays every month for a reporting dashboard. Photos from a room full of people I’d just met.
None of those could have been written by someone else. That is the whole difference.
It is not about whether AI touched the draft. Most of my posts start as an AI draft. The ones that fail are AI drafts with nothing of mine in them.
If you’re an Amazon agency or seller posting for leads
You already know this pattern from your ad account. Impressions are not orders. A campaign with a huge impression count and no sales is not a winning campaign, however good the top line looks in a client report.
LinkedIn works the same way. Three things to check:
- Look at engagement rate, not impressions. Pull your own top posts and divide reactions plus comments by impressions. If your biggest posts are your lowest-rate posts, your reach is coming from strangers who scroll past.
- Run the “only us” test before you post. Does this post contain something only you could have written: a client result, a question from a sales call, a number from your own account data? If you could swap your logo for a competitor’s and nothing would change, it won’t start a conversation.
- If you pay for an auto-posting tool, measure it like an ad. What did it produce in replies, DMs and booked calls, not in posts published?
The number that matters is the one closest to money. On Amazon, that’s orders. On LinkedIn, it’s the conversation that turns into a call.
I said my original posts beat the generic ones. On reach, I was wrong. On everything that leads to a client, the data says I was right.
