How to pick LinkedIn content pillars (and when to drop one)

LinkedIn content pillars are the three or four subjects you commit to returning to, and the practical way to pick them is to work backwards from what you get paid for rather than forwards from what you find interesting. Choose three or four, give each one at least eight posts before you judge it, score it against your own median rather than against anyone else, and drop it when the median stays below that line and the comments come from people who will never hire you. That is the whole method. Most advice stops at the choosing, which is the easy half.

What is a LinkedIn content pillar?

A content pillar is a subject you have decided to be repetitive about. Not a category on a spreadsheet. A promise: if a reader follows you, this is what they will keep getting.

That definition does more work than it looks like. A category can be broad, because a category only has to hold things. A promise has to be specific enough that someone could be disappointed by you breaking it. “Leadership” is a category. “What actually happens when you promote your best engineer” is a promise. Nobody can be let down by a leadership post. Plenty of people can be let down by a bad promotion post, which is precisely why the second one gets read.

The structural case for pillars was covered in our guide to LinkedIn content strategy, which argues for four angles on one territory rather than four separate topics. This page is the operations manual underneath that: how you find the candidate themes, how you run a fair test on each, and how you decide when one has stopped earning its slot.

How many content pillars should you have?

Three or four, and the reason is arithmetic rather than taste.

Take a normal cadence of three posts a week. A quarter is thirteen weeks, so roughly thirty-six posts after you lose a few to holidays and bad weeks. Split across four pillars, that is nine posts per pillar per quarter. Nine posts is about the point where a regular reader stops seeing individual posts and starts seeing a pattern.

If you post once a week, cut to two or three pillars. The number of pillars should be a function of your volume, not of how many interesting things you can think of. This is the single most common error: someone picks five pillars at one post a week, gives each theme two or three posts a quarter, and then concludes after six months that LinkedIn does not work for them. Nobody saw a pattern because there was not one.

Where do content pillars actually come from?

Not from a brainstorm. A brainstorm produces the themes you can articulate about yourself, which are usually the flattering ones and rarely the ones that draw money. Three extraction methods that work better, in the order you should try them:

1. The repeated explanation. Search your sent mail for the last five long emails you wrote explaining something to a client or a colleague. Themes you have explained four times are themes you can write about forty times, and the fact that you keep being asked is proof that demand exists.
2. The argument you keep having. What position do you find yourself defending in meetings? A pillar built on a disagreement has an engine, because there is always someone new to disagree with and always new evidence. Pillars built on agreement run out of material fast.
3. The thing you get hired to fix. Look at your last ten engagements or projects. What was the presenting problem? Not what you actually did, which is usually more interesting to you than to the buyer. The presenting problem is what the reader recognizes.

Run each candidate theme through one filter before it becomes a pillar: can you name ten specific posts inside it right now, without research? Ten, written out as titles on paper. If you get to four and stall, the theme is a topic you like rather than a territory you own. This test is unforgiving and it is worth doing badly rather than skipping.

What do people in your role actually write about?

We can answer this part with data rather than intuition. We split our cohort into role buckets by author headline and counted which themes each bucket writes about most. A post can match more than one theme, so the counts overlap and do not sum to the bucket total.

Two things stand out. First, leadership appears in the top four for every single bucket. It is the universal default, which means it is also the theme where your post looks most like everyone else's. Second, only the marketers bucket writes most about its own discipline. Founders write more about finance than about founding. Executives and coaches both write most about technology, which is a subject neither bucket is defined by.

The drift has a cause worth naming. Broad themes feel safer to write about because they require less exposure. A post about leadership commits you to nothing. A post about the specific thing you do wrong in your own discipline commits you to a claim someone can contest. The default themes are the ones with the lowest personal risk, and that is exactly why they are crowded. There is more on how role changes what works in our LinkedIn guide for founders and CEOs and the version for executives.

Which content themes are crowded?

Volume in our cohort is a rough proxy for how much these creators write about a theme. Set it next to the median engagement rate for posts on that theme and the pattern is not subtle.

The three highest-volume themes in the table (finance, leadership, technology) all sit at or below 0.57 per 1,000 followers. The four thinnest themes shown (personal branding, networking, job search, sales) all sit at or above 0.73. That is the crowding effect in one sentence: the more of these creators write about something, the less a median post about it earns.

Now the honesty. This is correlation with at least three confounds and you should not read it as a promise:

• Selection. The people who post about personal branding on LinkedIn are definitionally the people who are good at LinkedIn. Their high median says as much about who they are as about the theme.
• Audience fit. Finance posts here come mostly from accounts whose audience is not primarily on LinkedIn for finance content. A finance theme aimed at a finance audience is a different proposition.
• Volume drags the median. A theme with 2,795 posts includes a lot of routine posting. A theme with 173 includes mostly deliberate ones.

The usable version of the finding is narrower and still worth acting on: pick the theme your peers are least likely to have already covered this week. When you are choosing between two candidate pillars and one of them is the obvious industry default, that is a reason to take the other one, not a reason to take the safe one.

How do you turn a theme into a pillar you can write thirty posts about?

Most pillars fail at the specificity step, not the choosing step. A theme is too wide to write from and a single post idea is too narrow to repeat. What you want is the middle rung. Four levels, worked through with a real example:

The pillar rung is the one people skip. It is uncomfortable because it commits you to a position before you have written anything, and because it obviously excludes readers. Excluding readers is the mechanism, not a side effect. A pillar that could interest anyone will be remembered by nobody.

One shortcut for finding the pillar rung: write the theme, then add the word “why” or “when” and finish the sentence with something you have seen happen. “Hiring” becomes “why senior hires quit in month seven”. The addition forces a mechanism, and a mechanism is what a reader can disagree with.

How do you test whether a content pillar is working?

Here is where almost everyone goes wrong, in a specific and fixable way: they judge pillars on the best post each one produced. LinkedIn engagement is skewed steeply enough that this is close to meaningless. In our cohort the 75th percentile post earns about 3x the median and the 90th percentile earns roughly 5x the 75th. One outlier inside a four-post sample will make a dead pillar look alive.

The fix is to judge the median of a decent sample. Concretely:

1. Eight posts per pillar, minimum. At three posts a week across four pillars, that is about two months. Painfully slow, and shorter tests lie.
2. Hold format roughly constant. If one pillar gets all your videos and another gets all your link shares, you have tested formats, not pillars. Format differences are large enough to swamp topic differences on their own.
3. Score by comments, not impressions. Comments are the scarcer signal. The median top-decile post in our cohort earned 72 comments against 5 for the median bottom-half post, a gap of more than 14x, while reactions differ by roughly 3x.
4. Use your own median as the line. Not the cohort median, not a competitor. The question is only which of your pillars beats you.
5. Record who commented, not just how many. This is the column people skip and it is the one that decides whether a pillar stays.

The record itself can be five columns in a spreadsheet. It does not need to be a system:

After eight posts per pillar you will have one pillar that clearly beats your median, one or two that sit on it, and usually one that is quietly dragging. That is the normal distribution and it is the point of running the test. Benchmarks for what a strong post looks like at different follower counts are in LinkedIn engagement benchmarks.

What counts as a fair test?

Three things invalidate a pillar test and all three are common:

• Uneven effort. You wrote the pillar you like carefully and the other one on a Friday afternoon in eleven minutes. This is the most frequent confound and the hardest to admit to.
• Uneven timing. One pillar always landed on Tuesday morning and the other always on Friday evening. Timing effects are smaller than people think but not zero, and they are covered in the best time to post on LinkedIn.
• A structural break. If your reach fell off a cliff mid-test for reasons unrelated to the pillar, the test is void. Diagnose the account-level cause first, then rerun the pillar test from scratch.

Here is a real post from our dataset that illustrates why a pillar built on a live disagreement outperforms one built on general commentary. It is a position with a person attached, and the comment count relative to reactions tells you what happened:

When should you drop a content pillar?

Four conditions. Any one on its own is survivable. Two together and the pillar is finished.

1. The median is below your account median after eight fair attempts. Not the mean, and not the best post. If a pillar has produced one hit and seven duds, the one hit was about something else.
2. The comments come from the wrong people. A pillar can perform well and still be a liability if it draws peers, students, or people in a country you do not serve. This is the most expensive failure because it looks like success on every dashboard you have.
3. No conversation has traced back to it in a quarter. Not one DM, not one call, not one person who mentioned it. Nine or twelve posts with zero downstream is a real signal.
4. You dread writing it. Treat this as data rather than as a character flaw. Dread produces thin posts, thin posts produce bad numbers, and the pillar dies slowly and expensively rather than quickly.

Two bad reasons to drop a pillar, both extremely common:

• A slow month. Variance is enormous at post level. Three flat posts in a row inside a pillar that was working is normal, not a signal.
• You are bored of it. Boredom with your own material arrives roughly ten times faster than reader fatigue. By the time repetition feels unbearable to you, a large part of your audience has encountered the point once. This is the same argument made at length in LinkedIn personal branding, and it is worth internalizing before you rebuild your pillars for the third time in a year.

What do you replace a dropped pillar with?

Not with something new. Replace it with a split of your strongest pillar.

A pillar that beats your median has usually got two or three distinguishable claims inside it that you have been treating as one. Look at the eight posts and ask which of them were actually about different things. The best performing pillar in a test almost always subdivides cleanly, and the subdivision inherits the audience you already built rather than starting from zero.

The alternative, going back to the brainstorm list and picking the next theme down, throws away the only real information the test gave you. You now know which territory your audience shows up for. Go further into it.

Give the replacement a fresh eight posts before judging it, and expect the second test to run slightly better than the first, because you are now writing about a subject you have evidence for rather than a subject you assumed.

How should your content pillars be balanced?

Unevenly, and deliberately so. Equal splits are a sign nobody has made a decision.

The ratio to watch is the top one. If a single pillar goes above roughly half your output, you no longer have four pillars, you have one subject and three hobbies. That is a valid choice, but make it explicitly and cut to two pillars rather than pretending. Turning this split into an actual weekly schedule is the job of a LinkedIn content calendar, and if you need shapes to pour each pillar into, the recurring structures are collected in LinkedIn post templates.

What are the most common content pillar mistakes?

1. Pillars that are categories, not promises.“Industry news”, “tips”, and “company updates” are bins. Nobody follows a bin.
2. Format pillars. A carousel is not a subject. Formats belong underneath your pillars, not instead of them. A reader remembers what you told them, not whether you told them in five slides.
3. One pillar for the audience and three for yourself. Usually visible as three pillars nobody comments on and one that carries the account. The three are decoration.
4. Copying a competitor's pillars. Their pillars encode their evidence. You do not have their client stories, so you will write the same subjects without the proof, which reads as commentary.
5. Rebuilding every quarter. Every reset costs you the recognition the previous set was accumulating. Six months minimum before a full rebuild.
6. No pillar that risks anything. If none of your four pillars could produce a respectful argument, all four are wallpaper. That is the argument made in detail in our piece on LinkedIn thought leadership.

How do you set your LinkedIn content pillars up this week?

A three-sitting version that takes under two hours total:

1. Sitting one, forty minutes. Pull the last five long explanatory emails you sent and the last ten projects you were hired for. Write down every recurring subject without editing.
2. Sitting two, thirty minutes. Take the top six candidates and run the ten posts test on each. Most will fail at four or five ideas. Keep the three or four that survive.
3. Sitting three, twenty minutes.Push each survivor down the specificity ladder from theme to pillar. Add “why” or “when” and finish the sentence. Write the four pillar statements on one page.
4. Then wait. Eight posts per pillar before you touch any of it. Record comments and who made them. Change nothing until the test is done.

The waiting is the hard part and it is the part that separates a set of pillars from a set of guesses. Most people rebuild before they have evidence, then rebuild again, and after a year they have run four two-month experiments that each proved nothing. One honest test, finished, is worth more than four abandoned ones.

The short version on LinkedIn content pillars

Pick three or four. Extract them from your sent mail, your recurring arguments, and what you actually get hired to fix, rather than from a brainstorm. Push each one down from a theme to a specific claim that excludes most readers. Then run the part everyone skips: eight posts per pillar, format held roughly constant, scored on comments and on who made them, against your own median rather than anyone else's. Drop a pillar when the median is dead and the audience is wrong, and replace it by splitting your best pillar rather than by starting over.

None of it is complicated. It is just slower than the advice usually admits, and the evidence in our study of 34,000 LinkedIn posts keeps pointing at the same conclusion: consistency inside a narrow subject beats range, and the accounts that win are the ones that stayed still long enough to find out what was working.