Why did my LinkedIn company page reach drop, and is it you or the platform?

Total impressions is the wrong number to panic about

Total impressions is cadence multiplied by reach per post, so it can halve without anything happening to your distribution. Work an example with numbers you can check against your own export. A page publishes twelve posts in one month with a median of eight hundred impressions each, which reports as 9,600 impressions. The next month it publishes seven posts with a median of 760, which reports as 5,320.

That is a forty five percent fall in the headline number and a five percent fall in the thing that actually matters. The page did not get suppressed. Somebody went on holiday. Every dashboard in this category reports the first number by default and almost nobody recalculates the second, which is why the panic usually arrives before the diagnosis.

So the first move is arithmetic, not investigation. Divide impressions by the number of posts in each window, use the median rather than the mean so that one post that travelled does not rewrite your history, and only then ask whether you have a problem worth solving.

Why every page you just read quotes the same drop percentage

The identical reach collapse figure appears across the pages ranking for this question because they are quoting each other, not measuring anything. Follow any of them back and the trail ends at one analysis with no stated sample, no stated period and no stated measurement method. This page will not repeat the number, because a figure you cannot audit is folklore with a percent sign attached.

Even if it were sound, a platform wide average cannot diagnose your page. Averages hide variance, and variance is the entire story here. In any month some pages fall hard, some hold, and a few climb, and knowing the mean tells you nothing about which group you are in. You would not accept that standard from a media buyer reporting on a campaign.

It is also worth being precise about what did not happen. The audience did not leave. LinkedIn reports 1.3 billion members, and Semrush recorded 1.4 billion monthly visits in February 2026. The people are still arriving. What changed is how much of that arriving attention gets routed to a company page rather than to a person, which is a distribution question, not an audience question.

The Reach Autopsy: four numbers across two matched windows

Four numbers, pulled twice, will tell you what happened. Use two 28 day windows rather than calendar months so that each window contains the same number of weekdays, because a month with five Mondays and a month with four are not comparable and posting is a weekday activity.

Numbers one to three describe the size of the fall. Number four describes the cause, and it does so because of a sequencing property that dashboards never surface. Distribution is downstream of response. A model decides how far to push a post partly by how the first people shown it behave, so response rate moves before reach does when the problem is the content, and reach moves without response when the problem is upstream.

Read the order of the fall

Match your two columns against the patterns below and you get a cause, a test and a next action. The pattern is the diagnosis. The size of the drop is only the symptom, and treating the symptom is how pages end up posting more of the thing that stopped working.

The order of the fall is the diagnosis. Read left to right and stop at the first row that matches both of your columns.

Six self-inflicted causes that look like a platform change

Most reach falls are self-inflicted and arrive with an innocent explanation attached. Each of these has a mechanism you can reason about, and each can be tested on your own page in a fortnight rather than argued about.

1. Topic drift. A page that posted about one subject for a year and now posts about hiring, culture, product and events is harder to match to an interested reader, because the signal that told the system who to show it to has been diluted by the page itself.
2. A format switch. Moving from the format your audience responded to into a new one resets the response history the system is working from, so give any format change at least six posts before you judge it.
3. Outbound links in the post body. The mechanism is that a platform prefers content that keeps the session on the platform. The published tests disagree wildly on the size of the penalty, so run a paired test on your own page with the link in the body against the link in the first comment.
4. A change of writer. When the person who wrote the posts leaves, the voice, the specificity and the willingness to take a position usually leave with them, and response rate falls before reach does.
5. Cadence bunching. Three posts on Tuesday and nothing for ten days performs worse than three posts spread across ten days, because your own posts compete with each other for the same audience.
6. Deleting and reposting. Republishing the same content resets its history and gives the system two weak signals instead of one strong one.

There is a seventh that is not self-inflicted and is worth naming so you can rule it out. Pages get quieter when the topic itself gets quieter, and a seasonal industry has a seasonal feed. Compare the same window against last year rather than against last month before concluding anything about August.

The follower quality trap nobody checks

Adding the wrong followers lowers your reach, and this is the cause most teams never test. Your follower base is the audience the system matches your post against first. Fill it with people who have no interest in your subject and the early response rate on every future post falls, which is exactly the signal that decides whether the post gets pushed further.

Two events cause this. A hiring post that travels brings in candidates who wanted a job rather than your subject matter. An invitation run brings in whoever was on the list. Both look like growth on the follower chart and both raise the denominator of your follower reach ratio while lowering the numerator.

The check takes ten minutes. Open follower demographics, compare the company, function and seniority mix across your two windows, and look for a new group that arrived at the same time the reach fell. If you have been sending invitations, read how page invite credits actually work before the next batch, because the triage that decides who gets invited is the same triage that protects your reach.

Use your own ad cost per thousand as the market control

If you buy media on the same platform, you already own a control group and almost nobody uses it. Your cost per thousand impressions is a market price for attention in your audience. When it rises, more advertisers are bidding for the same people, which usually means those people are being shown more paid content and have less room for organic.

Pull the cost per thousand trend for the same dates as your two windows, targeting held constant. If it climbed while your organic reach fell, you are looking at demand for your audience rather than at a fault in your page. If it was flat while your reach halved, the market did not change and your page did, which sends you straight back to the four numbers.

This cross read has a second use. It prices what recovering the lost reach would cost. Multiply the impressions you lost by your own cost per thousand and you have the media budget that would replace them, which is the number that turns a reach conversation into a budget conversation. If the paid side is also underperforming, that is a separate diagnosis and it starts with why LinkedIn ads stop converting.

Paid data as a control for organic. Keep targeting constant across the windows or the comparison means nothing.

The 28 day plan that produces an answer instead of a theory

Do the diagnosis in one afternoon, then change exactly one variable for twenty eight days. The reason to change one thing is that everything on this page interacts, and a page that changes topic, cadence and format in the same month has destroyed its own ability to learn anything.

If the four numbers say the fall was distribution rather than content, and the audience mix is clean, then the honest answer is that the page is doing its job and the job has got smaller. That is a strategy conversation about where the company publishes, and for a team without the headcount to spread the load it starts at growing a page with a small team.