Social media algorithms: definition, ranking signals and what they mean for your posts

What a feed algorithm evaluates, platform by platform, and the practices that work with it rather than against it

The essentials

  • Definition: a social network's algorithm is the ranking system that decides, for each user, which posts to show and in what order, by predicting the likelihood of interaction.
  • The shared signals: early interactions (comments, shares, watch time), the relationship between the author and the reader, format, freshness, and the user's history.
  • The consequence: the average organic reach of a page sits between 2 and 6% of followers on Facebook, 5 to 15% on Instagram and LinkedIn; TikTok distributes without relying on followers.
  • What does work: post in a way that prompts a response (comment, save, share) within the first hour, and adapt the format to each platform.

Every social network shows users a selection of posts, chosen by a recommendation algorithm. This system ranks thousands of candidate items according to the likelihood that the person will react to them. Understanding its signals does not let you “trick” it, but it does stop you posting against the grain. This article describes how it works in general, the specifics of each platform, the reach figures worth knowing and the practices that improve distribution in SMO.

How a feed algorithm works

  1. Inventory: the system gathers the candidate posts (accounts followed, suggested content, adverts).
  2. Signals: for each post, it reads hundreds of variables: who published it, when, in what format, how the first readers reacted, and what the user's history is with that type of content.
  3. Predictions: models estimate the probability of each action (like, comment, share, watch to the end, hide).
  4. Score and ranking: the predictions are weighted (a comment counts for more than a like, a hide penalises heavily) and the posts are then ordered.
  5. Diversification: rules prevent too much similar content, or too much from the same author, being shown in a row.

Meta published this outline in 2023 in its “system cards”, and LinkedIn and TikTok have described equivalent mechanisms. The common thread: the interactions of the first few hours determine the distribution that follows.

The signals, platform by platform

PlatformDominant signalsFavoured formatsSpecific feature
FacebookComments and shares, conversations between friends, time spentVideo (Reels), posts that spark exchangesPage posts are deprioritised against friends and groups
InstagramSaves, shares by private message, watch time, relationship with the accountReels, carouselsThree separate rankings: feed, Explore, Reels
LinkedInSubstantial comments in the first hour, dwell time, relevance to the professional networkLong text, documents (PDF carousels), native videoExternal links in the post reduce reach; the author's own comments count
TikTokCompletion rate, rewatches, shares, interactions from the first impressionsShort vertical videoDistribution by test batches, independent of follower count
YouTubeWatch time, thumbnail click-through rate, stated satisfactionLong videos and ShortsA search engine as much as a feed
XReplies, reposts, reading time, the author's Premium subscriptionShort text, images, videoAlgorithm published as open source in 2023

What that looks like in figures

Metric (2025-2026 averages, business accounts)FacebookInstagramLinkedInTikTok
Average organic reach per post (as % of followers)2 to 6%5 to 15%5 to 15% (pages); 20 to 40% (profiles)Not tied to followers
Median engagement rate per post0.1 to 0.3%0.5 to 1.5%1 to 3%2 to 5%
Decisive window1 to 2 h1 to 3 h1 to 2 hFirst few hundred views

These ranges come from the annual studies by Socialinsider, Rival IQ and Metricool. They vary by sector and by account size; measure your own with the analysis of your social performance. The Organic reach has been falling structurally for ten years, which explains the growing place of advertising (SMA).

The practices that improve distribution

  • Aim for the reply, not the like: a precise question, a stated position or a before/after prompts comments, the most heavily weighted signal.
  • Work on the first three seconds or lines: the hook decides time spent, which decides reach.
  • Post when your audience is online to maximise early interactions; the platform statistics show the time slots.
  • Reply to comments within the hour: each reply restarts the conversation and extends distribution.
  • One native format per platform: vertical video uploaded directly, PDF carousel on LinkedIn, external link in a comment rather than in the post; see adapting your tone to each platform.
  • Consistency rather than volume: 3 to 5 well-crafted posts a week beat two a day; see should you post every day.
  • Involve people: individual profiles (directors, teams) get 3 to 5 times more reach than pages on LinkedIn.

The practices that hurt

  • Explicitly asking for likes or shares (“engagement bait”): detected and downranked on Meta and LinkedIn.
  • Publishing identical content on every platform, with unsuitable links and hashtags.
  • Deleting and reposting a post that starts badly: the negative history follows the account.
  • Artificial engagement “pods”: interactions without real reading degrade perceived quality, and the accounts are identified.
  • Systematic external links: platforms keep users on their own side.

Tracking the algorithm's effect on your accounts

The metrics to watch each month: reach per post relative to followers, the share of reach coming from non-followers (a sign that the algorithm is recommending you), engagement rate by format, and how things change after each shift in practice. Platforms adjust their weightings several times a year; a consolidated dashboard makes it possible to spot a drop affecting every account (an algorithm change) as opposed to an isolated drop (a content problem). The metrics are detailed in the SMO KPIs.

Our advice: identify your five highest-reach posts of the last six months and your five weakest. What they have in common (format, hook, subject, time) will tell you more about the algorithm as it treats your account than any general study.

How GreenRed helps

Rather than juggling several tools, GreenRed's social media tracking brings these metrics together in a single dashboard, compares them over time and tells you which actions come first. You can try it free, with no card, from the Pricing.

Frequently asked questions

Why is my organic reach falling when my followers are increasing?

Because reach depends on interactions, not on follower count. Platforms first show the posts that started well, along with recommended content from outside your subscriptions. If your posts prompt few comments or shares, they are shown to a shrinking fraction of your audience.

Do algorithms penalise external links?

On LinkedIn and Facebook, a post containing an outbound link gets on average 30 to 50% less reach than a post without a link, according to measurements by Socialinsider and several agencies. The common practice is to put the link in a comment or to publish the content directly on the platform.

Should you post at the same time every day?

Consistency helps, but the time matters less than whether your audience is present when you post. Check each platform's audience statistics to find the time slots, test two or three times, and keep the one that produces the most interactions in the first hour.

Can you know a platform's algorithm precisely?

No, but the platforms publish the broad families of signals: Meta in its system cards, LinkedIn on its engineering blog, X as open source. The exact weightings change regularly. The reliable method is to measure your own posts and work out from that what works for your audience.

Grow your social presence

GreenRed tracks your performance on Facebook, Instagram, LinkedIn and TikTok in a single dashboard.

Track my social media

Related articles