Renting your audience — what a social profile gives you, and what it never will

A company profile on a social platform has one property that stays invisible until it starts to hurt: it is not yours. You use it on terms somebody else sets, changes without asking, and need not explain.

This text is not an argument for leaving social media. It sets out what you lose if all of your company's presence stands on somebody else's platform — and it does so using the platform's own statements and our own measurement, rather than the percentages that circulate on industry blogs.

The sentence the platform wrote itself

The most repeated claim about social media is that organic reach collapsed. It usually arrives with a number — seventy per cent, eighty per cent — and never with a source. The previous version of this article carried both and had no source either.

The number is not needed, because Meta announced the direction itself. On 11 January 2018 Adam Mosseri, then head of News Feed, published the change. Two sentences are enough:

> We will also prioritize posts from friends and family over public content.

and, quite plainly:

> Pages may see their reach, video watch time and referral traffic decrease.

Business pages may see their reach, watch time and referral traffic to their own sites fall. That is not a leak or a conspiracy theory — it is a notice from the maker, published in advance, telling the companies using the platform that they would get less.

What matters is what the announcement proves: not that the platform works against you, but that the rules are its own, and changing them requires neither your consent nor your knowledge. Anyone who had built a business on organic reach had a different business the next morning without changing a thing.

How many people actually arrive from social media

Instead of somebody else's percentages, here is our own traffic split over six months, counted in people, not sessions, and given in shares. The measurement is from our Polish domain — the mechanism carries over, the exact proportions are ours.

Three things on that chart are worth stopping at.

Unpaid social is four per cent of our visitors. Not zero — but a fraction. A profile, maintained steadily, brought that many people in half a year.

Organic search brought nearly three times more, and we paid nothing for it. The difference between those two rows is, in practice, the difference between a channel where content keeps working after publication and one where it stops working after two days.

Paid social is four times larger than unpaid. That is the most honest sentence anyone can say about social media in a B2B setting: it works when you pay for it. This is not an accusation against the platform; it is a description of its business model, which was never hidden.

The caveats without which these numbers mean less than they appear to. Analytics sees only people who accepted cookies, so every share is a floor. We are a B2B services company — a restaurant, a clothing label or a tattoo studio will see an entirely different split, and for them social may well be the first channel. The mechanism of depending on somebody else's algorithm is identical regardless of sector, and that mechanism, not the particular percentage, is the subject here.

"So should I pay?" — the honest answer

The first chart raises a question that cannot be dodged: if paid social is four times larger than unpaid for us, is the answer simply a budget?

The answer is: yes, but what you are buying is not what it looks like. Advertising on social platforms buys reach to people who were not looking for you — and it does that better than any other channel. What it does not buy is anything that remains. On the day you switch the budget off, the channel returns to the size of the unpaid one, which for us is a few per cent.

That is precisely the difference between the two charts above. Content on your own site has a one-off cost and keeps working; advertising has a running cost and works exactly as long as you pay. Neither sentence is a criticism — they are two different tools, and confusing them is expensive in both directions. A company treating advertising as a foundation pays for presence indefinitely. A company relying on content alone waits months before anything happens.

The sensible division, which is what these shares show us doing: advertising for what must happen now, content for what should be working a year from now. The problem starts when three years of funding go entirely to the first and, three years on, there is nothing of the second.

Sessions are not people — and this is the easiest way to fool yourself

Before drawing conclusions from your own report, check one thing: are you looking at sessions or at people? This is not a nicety for analysts; it is a distinction that can reverse the conclusion.

Our largest discrepancy fell on referrals from other sites: fourteen per cent of sessions and one point four per cent of people — a tenfold gap. Looking at sessions alone, we would have called it the third most important channel on the site and started investing in it.

We checked before doing anything. One visitor accounted for about seven per cent of all sessions on the site — and it was our own tooling traffic, returning repeatedly through developer tools. A second similar source turned out to be an advertising panel we use ourselves. The "third most important channel" was entirely us.

The conclusion transfers to any report, including one from an agency: sessions describe frequency, people describe reach. A channel where one person returns fifteen times and a channel that brings fifteen different people look identical in a session report and mean completely different things. If your report gives only sessions, or only "visits", you do not know which case you have.

The number everyone quotes, and what it actually is

The previous version of this text gave platform shares of the Polish internet to a tenth of a percentage point and did not say where they came from. We replace them with figures anyone can check — together with the caveat their publisher prints, because that caveat matters more here than the numbers.

Digital 2026: Poland, produced by We Are Social and Meltwater and published in November 2025, gives Poland 27.1 million social media user identities, or 71.3% of the population. Advertising reach by platform: Facebook 18.7 million, Instagram 12.4 million, TikTok 12.4 million among adults. We quote the Polish row because it is the market we measured ourselves; the report breaks out twenty-four countries and your own will be in there.

And now the part that usually goes missing in citation. The publisher writes plainly that these values "may not represent unique individuals", and that estimated advertising reach "is not a proxy for monthly or daily active users". It also warns against comparing the values with previous editions of the report, because platform methodology changes and the comparison will produce a false result.

Three sentences of caveat change the meaning of all three numbers. "Eighteen point seven million people on Facebook" is a sentence the report does not make — it says that is the advertising reach estimate published by the platform itself in its ad-buying tool. One person with accounts on three platforms appears three times in that data.

It is exactly the same error as confusing sessions with people in the section above, committed on somebody else's data instead of our own. Before you base a decision on a number from a report, check what the number measures — often the publisher wrote it a paragraph below, and whoever quoted it stopped reading.

What happens when the channel simply disappears

On 4 October 2021 Meta's services became unreachable — not slower, not partially available, but absent from the internet. Company profiles, customer conversations and sales conducted through messages stopped existing at the same moment for everyone.

Meta described the cause in an engineering post published the next day. A faulty command issued during routine maintenance of the backbone network:

> unintentionally took down all the connections in our backbone network, effectively disconnecting Facebook data centers globally

The audit tool that should have caught such a command had a bug and did not stop it. How long the outage lasted is not stated in that account — and so we give no number of hours here. The mechanism matters more than the duration.

The thing to take from this is not "platforms are unreliable", because every system is sometimes unreliable, including your hosting. It is this: when it happens on somebody else's platform, you have no influence, no information and nobody to ask. With your own site an outage will also happen — but you know whose fault it is, you have a contract, and you can move the site elsewhere. Here you can only wait and refresh.

What a profile alone never gives you

None of the following depends on the platform, the sector, or how well you run the profile. It follows from the fact that an account is granted, not owned.

• A list of contacts. You have followers, but no way to reach them off the platform. If the account disappears, they haven't — you have simply lost the route to them.
• An address that belongs to you. A domain is your asset and stays with you when you change supplier, hosting or technology. A profile stays where it was.
• Control over what the audience sees. Order, format and visibility are decided by an algorithm whose changes are announced when they are announced — see the 2018 notice above.
• Full measurement. You see what the platform's panel shows, in the breakdown it chose, for the period it set.
• A counterparty. What binds you to the platform is terms accepted with one click, not a contract negotiated by two sides. Meta's terms of service, in the version effective from 1 January 2025, say plainly: "we can suspend or permanently disable your access", and may "permanently disable or delete your account". They promise a warning as a rule — and reserve the right not to warn when a breach is judged severe enough. An appeals route exists; a counterparty to discuss it with does not.

It is worth seeing how these five gaps look in practice, because separately they sound theoretical and together they form one situation. A company runs a profile for three years and gathers a few thousand followers. Sales go through messages, orders are written down by hand, contacts stay in the platform's inbox. Everything works — until the day the account is blocked or reach falls after an algorithm change.

That day it turns out there is no customer list, because the list was a feature of the platform rather than an asset of the business. There is no way to tell anyone about the change, because the only channel of contact was in the same place. And there is nobody to appeal to, because the relationship was never a contract — it was terms accepted with a single click.

This is not a disaster scenario. It is the ordinary consequence of customer contact being stored on somebody else's site and never copied to one of your own.

Three of the platform's own documents, quoted in this text, line up on a single timeline — and at no point on it was the company a party.

What the channel genuinely does better

An article listing only one side's faults is an advertisement for the other. Three things social media does better than any website, worth naming before the reckoning above is taken as a verdict.

It reaches people who were not looking for you. Search serves demand that already exists — somebody has to type a query. A social platform can put you in front of someone who did not know you existed and was not searching for anything. That is a real advantage and no amount of content on your own site reproduces it, because there you have to be found first.

It shows the company is alive. A profile with activity from the last week says something no website can: that there is somebody present on the other side. For services where a client hands over money or data, that is a credibility signal that works immediately.

It allows a conversation. A comment, a message, a question under a post — a lower threshold than a contact form, and sometimes the only contact somebody is ready for at that stage.

The reckoning above cancels none of the three. It concerns something else: what remains when the platform changes the rules or stops working — because at that moment all three advantages vanish at once and leave no trace.

How to move an audience to a channel that is yours

This is the only part of this text that requires work, and all of it can be done without a budget.

A link that leads to the specific thing, not to the home page. Sending somebody from a profile to your home page leaves them where they must find what they came for by themselves. The link should lead to the one thing the post was about: the calculator, the particular service, the article.

A reason to leave a contact. Someone who follows you will not leave an email address because you asked. They will leave one if they get something with standalone value in return — a calculation, a template, a report, a price comparison.

What specifically to offer. The rule is one: what you give must have value in itself, including to somebody who will never buy from you. A calculation for a specific situation works, so does a price comparison that exists nowhere else, a document template, or a checklist to tick off. "Subscribe to our newsletter" without saying what will be in it does not work, and neither does a discount for somebody who does not yet know whether they want to buy.

What not to do. Do not move the whole conversation to your own site overnight — people stayed on the platform for the platform. Do not switch the profile off after building a list; these are two channels, not one instead of the other. And do not measure the move by follower count, which usually grows more slowly afterwards — measure it by the number of contacts you hold yourself.

Tagging links so it can be measured. Without parameters in the address, traffic from platforms falls into a bucket marked "direct" or "referral" and a month later you cannot tell the post that worked from the one that did not. How to set that up is covered in the article on analytics tools.

Content that works longer than two days. A post has a life measured in hours; an article answering a question people put to a search engine works for years. That is the difference visible in the first chart — and how to write for questions people actually ask is a separate text on search.

What this text does not claim

It does not claim social media fails. For us, paid social is the second largest channel. It claims social works differently from the way it is described: as a channel of reach that must be fed, not as a place where a company's assets accumulate.

Nor is this an article about why you need a website. That is a separate and broader question. Here the subject is one specific difference — between a rented channel and an owned one — and what that difference means on the day the platform changes the rules or stops working.

We give you no threshold. The previous version of this text advised that "if more than 60% of traffic comes from social media, that is a warning sign". That threshold was invented. The honest question is different and contains no percentages: if your profile disappeared tonight, would you have any way of contacting your customers tomorrow morning? If the answer is no, that is precisely the dependency this article is about — whatever your percentage happens to be.

What we removed from the previous versions

This text was merged from two articles, and more numbers came out of it than stayed in. We list them, because an article about the reliability of channels cannot stay silent about its own:

• Two reports that do not exist. The previous version cited a "Social Media Marketing Report 2024" and "research by Social Media Today from 2024", each with specific percentages attached. Neither could be traced, because there is nothing to trace. That is a worse category of error than a number without a source — here the source was invented along with the number.
• Reach declines of 70% and 80%, "around 60%", "reach fell to 2% of the community", "lost 70% of customers", "revenue falls by 65%" — none of these could be confirmed. One 2018 announcement from Meta replaced them all, saying the same thing truthfully.
• A case study of "+340% organic traffic and −60% cost of acquisition" with no company, no period and no before-and-after data.
• A promise of a 30% traffic increase placed in a heading, along with "retargeting converts at 10–15% against 2–3%", "five minutes of configuration will raise traffic by 20–30%" and "a customer from Facebook stays 40% longer than one from Google Ads".
• Platform shares of the Polish internet given to a tenth of a percentage point, with no indication of the study they came from.

The sources that remain, all checked directly with their publishers in September 2026:

• The News Feed change announcement — Adam Mosseri, Meta, 11 January 2018.
• The engineering account of the outage of 4 October 2021 — Santosh Janardhan, Meta Engineering, 5 October 2021. That account gives no duration and neither do we.
• Meta's terms of service, in the version effective from 1 January 2025.
• DataReportal, "Digital 2026: Poland" — We Are Social and Meltwater, 8 November 2025, data for October 2025, quoted together with the publisher's caveats.

Figures about our own traffic come from GA4 for the period 20 March – 15 September 2026 on the domain digitalvantage.pl, cover only users who accepted cookies, and are given exclusively in shares.