LinkedIn analytics, surface by surface

LinkedIn has three separate analytics surfaces that share almost no vocabulary: one for a personal profile, one for a company page, and one for whatever the network is calling creator tools this quarter. Most confusion about LinkedIn analytics comes from reading a number on one surface and assuming it means the same thing on another.

This is the complete guide to LinkedIn analytics: what each surface actually gives you, what the words mean, what disappears over time, what anybody can see from outside a profile, and how to measure people who will never connect their account to your tool.

TL;DR

The three surfaces, and why they disagree

A personal profile and a company page are different products inside one network. They were built at different times for different customers, and their analytics reflect that rather than any coherent design.

Personal profile analytics are post centred. You get detail about individual posts and about who saw them, with a strong bias toward recent activity. Company page analytics are audience centred. You get followers, visitors and demographics over long windows, with an export button, because pages are used by people who have to report.

The practical consequence is that a question easy to answer for a page is often impossible for a profile and the reverse. Somebody who ran a company page for two years and then starts posting personally will find the reporting they relied on is simply not there.

Almost every argument about LinkedIn numbers is two people reading two different surfaces and using the same word for both.

The vocabulary, in plain terms

Five words carry most of the confusion. They are worth pinning down once, in writing, in the place the numbers are kept, because a definition that drifts halfway through a year quietly destroys the comparability of the whole history and nobody notices until they try to explain a chart.

1. Impressions. A post appeared on a screen. It says nothing about whether it was read, and it is counted per appearance rather than per person.
2. Members reached. Distinct people, which is a different and smaller number than impressions. Mixing the two in one chart is the most common reporting error on this network.
3. Engagements. Reactions, comments and reposts, and depending on the surface, clicks. Whether clicks are included is the detail that makes two tools disagree by a factor of two.
4. Followers. People who see your posts by default. On a personal profile this is a different set from connections, and the difference grows once anything travels.
5. Profile views. Who looked at you rather than at a post. It is the one metric on this network with no equivalent anywhere else, and it is a genuine leading indicator.

The arithmetic that turns these into a rate, and the three denominators people use for the word engagement, is in how to calculate engagement rate. Pick one, write it in the sheet header, and never change it mid year.

What expires, and why it matters more here

Analytics on this network are a window rather than an archive. Post level detail thins out as posts age, and the further back you look the less is available, particularly on a personal profile. The exact window moves, and any guide that quotes one is wrong within two quarters, which is why it is not quoted here.

The consequence is structural and it catches people out once, expensively. If nothing is capturing the numbers as they happen, the history for the interesting month may simply not exist by the time somebody asks about it. There is no support ticket that recovers it.

• Export what a page gives you, on a schedule. A monthly export costs two minutes and is the cheapest insurance in this whole guide.
• Record post level numbers within a few weeks. Counts keep moving for a while after publication, then the detail starts thinning.
• Never rely on the network as your archive. It is a live product, not a records system, and it has no obligation to either.
• Write down the definition you exported. A column labelled engagements is meaningless in eighteen months without it.

What anybody can see from outside

This is the part that decides most tooling questions, and it is much simpler than the marketing around it suggests. A public profile shows what was published and when, the reactions and comments on each post, and the follower count. That is a real and useful data set.

What it does not show is anything the network computes for the owner. Impressions, members reached, demographics and profile views are owner only, and no product has them for an account it is not connected to. A tool that displays them for a third party account is modelling them, and a model presented as a measurement is worse than an empty column.

The full version of that table across every network is in public social media metrics, and the mechanics of reading a specific person's history are in how to see someone's LinkedIn post history.

Company page analytics, in practice

A company page is the surface built for people who report, and it shows. There is a followers view, a visitors view, a content view and an export, and between them they answer the questions a marketing team is asked in a monthly meeting.

What a page cannot do is generate the reach a personal profile does, and no amount of analytics fixes that. Teams often respond to a flat page by looking harder at the page numbers, which is the one activity guaranteed not to help. The page is a filing cabinet with a logo, and the reach lives on the profiles.

• Followers over time is the page's honest metric. It moves slowly, it is hard to game, and it accumulates.
• Visitors tell you about search and links, not the feed. A spike usually means somebody linked to you somewhere else.
• Content performance is worth reading for format, not for volume. Which shapes travel is transferable to the profiles.
• Export monthly. The page is where an export exists at all, so use it.

Used properly, a page is a research instrument for a personal profile programme rather than a channel in its own right. That is an unpopular conclusion in a lot of organisations and it survives most attempts to argue with it.

What a third party tool can and cannot add

Once the table above is clear, the tooling question mostly answers itself. There are exactly three things a product can do here, and everything on a pricing page is one of them.

1. Keep history the network does not. The most valuable of the three, the least demonstrable in a demo, and the reason month nine is different from month one.
2. Collect several accounts into one view. Real work, especially across networks, and the whole of it is collection rather than measurement.
3. Compute something on top. Rates, trends and comparisons. Useful, and always reproducible in a spreadsheet if you have the underlying numbers.

What no product can do is see private metrics for an account it is not connected to. That is a network boundary rather than a feature gap, which means it will not be closed by a competitor next quarter either. The shortlist for teams, sorted by what each product can genuinely read, is in best LinkedIn analytics tools for teams.

The personal profile problem

Most serious LinkedIn programmes are not run from a company page. They are run from the personal profiles of a founder, a few salespeople and whoever else writes well, and that is where the reach is. It is also where reporting collapses, and the collapse is usually discovered a quarter in, when somebody asks for a number that nobody has been collecting.

The reason is not technical. It is that a personal profile belongs to a person, and people do not connect their personal account to a work tool. They are right not to, the refusal is permanent, and any plan that depends on twelve colleagues changing their minds is not a plan.

• Asking people to send screenshots. Works for two weeks, then decays, and the gap always lands in a busy month.
• A shared login. Against the terms of every network, and it ends badly for the person whose account it is.
• Waiting for the network to add team reporting for profiles. It has been coming for years.
• Reading the public side of each profile. Smaller data set, no permission needed, works the same on day one and day four hundred.

Only the last one survives contact with reality, and it is why the columns in the table above are worth knowing precisely. Posts published per person per week is available from outside, and it happens to be the number that predicts whether the programme is alive, as employee advocacy sets out.

What to actually report, monthly

A LinkedIn report that lists every available metric is a data dump, and the person receiving it will invent their own interpretation, usually a harsher one than the numbers deserve. Six lines, in this order, cover almost every audience.

1. Posts published, against posts planned. The only line entirely under the team's control, and the one that explains most of the rest.
2. Followers now against ninety days ago. Never week on week on an account under a few thousand, because that is noise.
3. Comments received. The most expensive reaction a reader can give and the hardest to buy.
4. The three posts that travelled, with what they were about. This is the only line anybody reads twice.
5. One thing that changed, in a sentence. Format, cadence, subject, whatever it was.
6. What happens next month. A report with no next step is a receipt.

Impressions deliberately do not appear on that list. They are the largest number available, which is why they end up in reports, and they move for reasons nobody in the room controls. The full argument, and the format, is in how to build a social media report.

Benchmarks, and why most of them are useless

Published LinkedIn benchmarks average across account sizes, industries and posting styles that have nothing in common. An engagement rate quoted for the network as a whole is arithmetic performed on unlike things and it will mislead a small account badly.

The benchmark worth having is five to ten real accounts doing the same job as yours, read every month. That set is public, it is specific, and it moves for the same reasons your account moves, which is exactly what a comparison needs. Building it takes an afternoon and it replaces every published industry average you will ever be shown.

Two rules make it honest. Compare accounts of roughly the same size, because reach does not scale linearly with followers and a comparison against an account ten times larger tells you nothing you can act on. And compare over a quarter rather than a month, because a single post that travels will otherwise dominate the whole comparison and send you chasing a format that worked once.

There is one more comparison worth running and almost nobody does it: your account against itself a year ago, at the same size. Growth flatters everything, and a set of numbers that looks strong today can hide the fact that each individual post is travelling less far than it did when the audience was half as big. Reach per follower falling while total reach rises is the single most useful early warning this network offers.

It is also the reason a benchmark set has to be re read rather than chosen once. Accounts stop posting, change subject or hire somebody who changes the voice, and a comparison set assembled eighteen months ago is quietly measuring a different thing. Half an hour once a quarter keeps it honest.

How Groowth does this

Groowth reads the public side of a LinkedIn profile. You paste public handles, yours or anybody else's, and it records what each account published and when, the follower count, and the public reactions on each post, once a day. Nothing is connected and no one signs in, which is why it works on colleagues and competitors alike.

It never shows impressions, members reached or profile views for an account it is not connected to, because those do not exist outside the owner's own view and displaying an estimate would be inventing a number. It also builds the history the network does not keep, which is the part that matters in month nine. Three accounts are free with no card, and a free audit reads one public profile with no account at all.

Frequently asked questions