Public social media metrics: what any tool can see without your login

A tool promises you reach for every account on your team. Nobody connected anything. No password was shared, no admin access was granted. Either the number is real and someone handed over credentials you do not know about, or the number is an estimate wearing a metric's clothes. There is no third option.

Public social media metrics are the numbers anyone can see on a profile without logging in. This guide draws the line between those and the private ones, explains what each side is good for, and shows how to run a team program on the public side alone. For founders, marketing leads and agencies.

TL;DR

What public actually means

A public metric is one a logged out visitor can read on the profile page. It is published by the account itself, to everyone, as part of how the network works. A private metric is one the platform only shows to the account owner, behind their login.

The distinction is not about ethics or scraping. It is about who the platform decided to show a number to. Follower counts are public because networks use them as social proof. Reach is private because it describes the platform's distribution decisions, not the creator's work.

If a metric describes what the algorithm did, it is private. If it describes what the account did, it is usually public.

The proof is in the platform documentation

This is not an opinion about what tools can do. Meta's own developer documentation is explicit: to read insights such as impressions and reach for an Instagram professional account, an app needs an access token from that Instagram business user. No token, no insights.

LinkedIn is the same shape. Its help pages define impressions as the number of times a post was displayed on screen, and state plainly that the number is an estimate and may not be precise. That figure is shown to the author in their own analytics, not to a visitor.

So when a tool offers you reach across a team of ten people without anyone connecting an account, ask which token it is using. The honest answers are that someone did connect, or that the number is modelled. Both are fine if they are stated. Neither is fine when it is not.

What you can actually run on public data

More than people expect, and specifically the whole consistency problem. Publishing is a public act. Whether someone posted twice this week is knowable from outside the account, which means a whole category of team programs needs no access at all.

• Publishing cadence and streaks. Counting posts per person, per period, against a target. See social media posting consistency.
• Audience growth. Follower counts sampled daily give a clean growth curve, and the formula is in follower growth rate.
• Public engagement. Likes and comments per post, which is a rough but honest signal of how a post landed.
• Competitor and benchmark work. Any profile, without asking anyone. See track competitor social media.

What you cannot do on public data is optimisation work that depends on distribution: testing hooks against reach, reading audience demographics, or attributing clicks. Those need the account, and there is no clever way around it.

Why this decides whether a team program starts

A tool that needs account access needs it from every person you track. That means ten people connecting accounts, or sharing credentials, or installing something. In practice it means a Slack thread that runs for three weeks and a program that starts with four of the ten people set up.

Public data removes the negotiation entirely. You add a handle, and the tracking starts. Nobody grants anything, nobody installs anything, and a person who leaves does not take their numbers with them.

The one real limit: private accounts

A private profile cannot be read at all, and that is the correct behaviour. If someone on the team sets their account to private, tracking stops adding new readings. The history already collected stays, nothing new arrives, and no tool should pretend otherwise.

For personal branding programs this is rarely a problem, because the point of the account is to be seen. It is worth knowing before you promise a dashboard to a leadership team.

How to choose, by situation

• One account you own. Use the native analytics. They are free, they include the private metrics, and no third party is needed.
• A team publishing under their own names. Public data, because the setup cost per person is the thing that actually kills these programs.
• An agency reporting to clients. Public data, plus a client view. Client accounts are not yours to connect. See client reporting for personal branding.
• Paid media or funnel work. Account access, because attribution and reach are the whole job there.

Network by network

The line sits in roughly the same place everywhere, with small differences worth knowing before you promise a dashboard.

Video view counts are the interesting case. They are public on YouTube and TikTok, which means a video program can be measured from outside in a way a written post program cannot.

What about the official APIs

The official APIs do not move the line, they enforce it. Every one of them requires the account owner to grant access before it returns anything private, which is the same permission problem with an OAuth screen in front of it.

They also come with review processes, app approval and rate limits. For a team program tracking ten people across four networks, the cost of that path is measured in weeks. For a single brand account you own, it is often worth it.

Three mistakes when building on public data

Public data is honest and limited, and both halves of that matter. These are the ways teams get surprised in month two.

• Promising a metric before checking who can see it. Say what will be in the dashboard after you have confirmed it is visible, not during the kickoff.
• Treating a modelled number as a measured one. If a figure was estimated from public signals, label it. Once, in writing, where the report is read.
• Forgetting that a profile can go private. Build the report so a missing account reads as unavailable rather than as a zero.

The last one matters more than it sounds. A zero and an unknown look identical on a chart, and only one of them is a reason to have a conversation with someone.

What about resold data and scraped datasets

There is a market for social datasets, and some of it promises numbers that the line above says are impossible. It is worth knowing which of the three things you are being sold before you build a report on it.

• Aggregated public data. Someone else read the same public profiles and is selling you the reading. Legitimate, and no better than doing it yourself.
• Modelled estimates. Reach or demographics inferred from public signals. Useful as a directional guess, dangerous the moment it is presented as measured.
• Data obtained with credentials. Somebody connected an account somewhere. Ask whose, and whether they knew, before this ends up in a client report.

The question that separates them is short: which account was logged in when this number was produced. A vendor that answers it clearly is fine to work with. A vendor that changes the subject has told you the answer.

Where this table sits in the wider practice, from capture through to what a monthly report should say, is in social media analytics.

How Groowth handles this

Groowth reads public profiles only. It never logs in to a network, never asks for a password, and never requests admin access. It shows what a visitor can see: posts published, follower counts, and the public like and comment counts on each post.

It never shows reach or impressions, because it cannot see them, and a product that displays a number it cannot source is a product you cannot trust with the numbers it can. The free plan tracks three accounts with no card.

Frequently asked questions