LinkedIn · Social Analytics

Platform · Personal

Every LinkedIn post, how it performed, when it performs best, and what the LinkedIn specialist recommends.

Data as of Sep 25, 2026, 9:48 PM

Engagements

149+893%

90D

Per post

1.7-11%

Engagements per post

Engagement rate

2.0%-50%

Average, on impressions

Posts

89+1013%

Last 90D

Impressions

11.8K+2563%

7.5K reach

Followers

3,5641 account

Current total

LinkedIn specialist

Go back to posts about AI skills and careers for non-developers. Your 7 posts on that topic had a median of 255 impressions and 3.9 engagements each. You haven't posted one since Aug 10, and impressions per post fell to 80 in September.

All advice

1
Post twice a week about AI skills for operators who aren't developers
Write about what an operator or owner should learn, which parts of a job AI changes, and what the Claude certification paths mean for a team. Frame each post around a buyer's decision, such as what your team should learn first or what should stay with a person, so the posts reach Arqentia's buyers and not only job seekers. When Anthropic or another vendor announces something about skills or certification, post about it within 24 hours.
2
Open with your position, not a survey stat or your credentials
Make the first line a claim about a decision the reader faces, as in 'The most valuable AI skill right now is not prompting. It is knowing what to leave alone.' Mention your hardware background in one line in the middle of the post, and word it differently each time. Stop re-posting earlier frameworks as 'I published my ... on this feed' posts. If a framework deserves a second run, rebuild it as a document post.
3
Post your best work Friday to Monday, and never more than once a day
Keep posting at 11am ET. Put the week's AI-skills post on Saturday or Sunday, post on Friday and Monday, and use only one day from Tuesday to Thursday. Don't add a second post on the same day.

From 97 posts

Best times to post

Full map

1Mon 12pm–3pm2 posts1.6×
2Fri 12pm–3pm3 posts1.4×
3Sat 12pm–3pm2 posts1.3×

Lift vs this account's usual post · times in America/New York

Engagement over time

Engagements by publish week · last 90D

@Rafael Schwart , PMP

89 posts in range

All LinkedIn posts

Post	Published	Impr.	Reach	Likes	Comm.	Saves	Eng.	ER
Is your AI writing the work, or checking it? Almost everybody picked writing. Almost nobody decided it. You have one model and two places to put it. In front of the person, or behind them. In front means the model draft… Image	Sep 25, 202611:16 AM	30	19	1	0	0	1	3.3%
87% of AI decision-makers say their teams keep re-checking context their agents already have. Not the answers. The context. That is not a trust problem. It is an expiry problem. Somebody set the agent up once. Gave it t… Image	Sep 24, 202611:17 AM	29	22	0	0	0	0	0.0%
The AI demo you are watching was built to work. That is not a criticism of the vendor. It is the job of a demo. The mistake is treating it as evidence. Every AI demo I have sat through runs on the clean version of the p… Image	Sep 23, 202611:16 AM	66	45	1	2	0	3	4.5%
You do not know which steps of your automation are safe to run twice. Almost nobody does. Then it fails halfway on a Tuesday, somebody hits run again, and the customer gets the same email twice. Every step in a workflow… Video	Sep 22, 202611:16 AM	61	42	0	0	0	0	0.0%
Every automation has the same backup plan: we will just do it by hand. Nobody has actually done it by hand since the day it went live. That is the risk, and it does not show up on any dashboard. I take hardware from pil… Image	Sep 21, 202611:10 AM	29	16	1	0	0	1	3.5%
How much of your process is anyone actually working on? Almost never as much as you think. The rest of it is sitting in somebody's queue. That is why the AI pilot saved 20 minutes and changed nothing. A process has two … Text	Sep 20, 202611:07 AM	136	89	0	1	0	1	0.7%
An automation does not break. It drifts. And drift makes no noise. You costed the build. The prompt, the connector, the pilot, the demo that went well. One number, approved once. Nobody costed the run. On a factory floo… Image	Sep 19, 202611:08 AM	33	22	0	0	0	0	0.0%
75% of companies say their AI agents are secure end to end. That group had security incidents at 88%. Everyone else: 87%. One point. The confidence was not measuring anything. I spent years running Gauge R&R on factory … Image	Sep 18, 202611:20 AM	23	17	1	0	0	1	4.3%
The first time I sat in an 8D, I had the order wrong. I thought you find the cause, then you fix it. The room went somewhere else first. Not what caused it. How many already shipped. Ten years on factory floors since, a… Image	Sep 17, 20263:19 PM	85	60	0	0	1	1	1.2%
You do not have a quality gate. You have a checklist that somebody signs. Adding a model to it does not change that. A checklist records that someone looked. A gate decides whether the work moves. Most operations only e… Video	Sep 14, 202611:13 AM	39	28	1	0	0	1	2.6%
Most operators get the right answer out of Claude exactly once. Then they ask the same question next month and get a different number. They call the first one a win and move on. On a factory floor that is not a win. A g… Image	Sep 13, 202611:11 AM	31	20	3	0	0	3	9.7%
Revoking an agent's access and the agent losing access are two different events. Most teams only ever do the first one. You flip the toggle, the ticket closes, everybody moves on. Nobody checks whether the thing actuall… Text	Sep 12, 202611:08 AM	81	54	2	0	0	2	2.5%
Where is the last good answer Claude gave you? If the honest answer is "in a chat somewhere," that was not work. That was reading. Operations do not run on answers. They run on records. A work order. A ticket. A row in … Image	Sep 11, 202610:31 AM	85	64	1	0	0	1	1.2%
Someone is going to tell you a process is 80% automated. That is not 80% of the work. XPeng commissioned its humanoid robot line in Guangzhou yesterday, and the first IRON walked off it on its own. The published spec: o… Image	Sep 9, 202610:07 PM	179	143	1	0	0	1	0.6%
The problem is not that Claude gets it wrong. It is that you cannot tell when it did. You asked because you could not work it out yourself. Then it came back clean, confident, formatted. And now you are supposed to grad… Image	Sep 8, 202611:10 AM	129	111	0	0	1	1	0.8%
The automation that breaks is not the one that costs you. It is the one that keeps running. Aerospace, surgical robotics, AR. Eight-plus years of hardware ramps, and the expensive failures were never the loud ones. A te… Video	Sep 7, 202611:10 AM	63	51	0	0	0	0	0.0%
94% of companies are confident their AI agents don't hold more access than they need. 32.7% actually provision least privilege. Same survey, published last week. 202 enterprise technology and security leaders. And 65% o… Image	Sep 6, 202611:09 AM	97	74	0	2	1	3	3.1%
One clean AI run proves possibility. It does not prove the workflow is ready to operate every day. The gap is promotion discipline. Before a useful experiment becomes an operating capability, it should pass five gates. … Image	Sep 6, 20269:01 AM	0	0	0	0	0	0	0.0%
You are pasting the summary into Claude and asking it to find what the summary hid. I spent eight-plus years taking hardware from pilot build to full ramp, and it went the same way every time. The number on the dashboar… Text	Sep 5, 202611:05 AM	323	231	2	0	0	2	0.6%
A factory line has no station called "build the radio." It has dozens of stations, and each one gets automated on its own terms. Ten years on factory floors, not in slide decks, and I have never once seen that rule brok… Image	Aug 30, 202611:08 AM	128	97	3	0	0	3	2.3%

Showing 20 of 89

What people do

Engagement mix

LinkedIn149

likes 108comments 20shares 4saves 17

LikesCommentsSharesSaves

Engagements per post

What works

By format · posts

Image1.9 · 55
Carousel1.7 · 3
Text1.5 · 17
Video1.1 · 14

By weekday · best: Sat

Mon12

Tue11

Wed11

Thu15

Fri15

Sat12

Sun13

Engagements are likes + comments + shares + saves as each platform reports them through Zernio. Periods compare posts published in the last 90D against the 90D before; recent posts are still collecting engagement.
Engagement rate is engagements ÷ impressions, as Zernio reports it. Impressions mean different things per platform (Instagram counts repeat views), so rates are compared within a platform, never across.
Personal LinkedIn: LinkedIn only returns analytics for posts published through Zernio.
X @arqentia: Zernio returns no posts yet, so it shows followers only.
Missing values show as a dash — not available is not zero.