LinkedIn Growth4 min read
How LinkedIn Recommends Connections: Where They Come From
How LinkedIn recommends connections in People You May Know: the shared connections, profile overlaps, and imported contacts behind the suggestions.
By Animesh Kudake, Co-founder, Apna Project
The short version
- LinkedIn recommends connections mainly through People You May Know, which suggests members based on shared connections, similar profile information, the same company or industry, the same school, and contacts you have imported.
- The strongest single signal is mutual connections, the friends-of-friends pattern LinkedIn engineers describe as triangle closing, so who you already know shapes who you are shown.
- Imported email and phone contacts feed suggestions too, which is why syncing an address book can surface people you never searched for.
- LinkedIn states it does not scan the content of your messages to build these suggestions, a specific privacy boundary worth knowing.
- You cannot pick your own suggestions, but you shape them, since a complete profile, real connections, and consistent engagement all make the recommendations more relevant.
Why mutual connections matter most
Of those signals, shared connections do the heavy lifting. LinkedIn's engineering team has described the core pattern as "triangle closing," the friends-of-friends idea: if you know Bob and Bob knows Carol, then maybe you know Carol too. The more mutual connections you share with someone, the stronger the hint.
That is why your first-degree network shapes your suggestions so heavily. Every person you connect with brings their own connections into your 2nd-degree pool, and those are exactly the people PYMK starts recommending. If you want to understand the degree system underneath all of this, we lay it out in what 1st, 2nd, and 3rd mean on LinkedIn, and the mechanics of reaching further out in how to see 3rd-degree connections on LinkedIn.
LinkedIn engineers have also written that the system weighs finer features, such as whether two people overlapped at an organization or school and how far apart they are geographically, and that it learns from feedback: suggestions repeatedly shown without leading to a connection get pushed down. So the list you see is ranked, not just assembled.
The address-book signal people forget
The suggestion that makes people slightly uneasy is usually the imported-contacts one. If you have ever let LinkedIn sync your email or phone contacts, those people can appear in your recommendations even if you have never interacted on the platform.
This is why a person you emailed once, or exchanged numbers with at an event, can surface out of nowhere. It is not surveillance of your activity; it is the address book you handed over doing its job. If that feels like more than you want, you can review and manage synced contacts in your settings, and decline the import prompts that appear during onboarding.
What LinkedIn says it does not use
One boundary is worth stating plainly, because people assume the worst: LinkedIn explicitly says it does not scan the content of your messages to generate these suggestions. The recommendations come from network overlaps, profile fields, and imported contacts, not from reading your conversations.
That distinction matters. The inputs are structural, not conversational. Suggestions are about who sits near you in the graph, which is a different question from what your feed shows you, the topic we cover in the LinkedIn interest graph explained. One decides who to recommend; the other decides what content to surface.
How to get more relevant suggestions
You cannot hand-pick your People You May Know list, but you feed it every day, so you can improve what comes back:
- Complete your profile accurately. Your employer, school, and industry are matching fields. Vague or empty profiles get vaguer suggestions.
- Build real connections, not random ones. Since mutual connections are the strongest signal, a network of genuinely relevant people produces genuinely relevant recommendations. A pile of strangers produces noise.
- Engage where you want to belong. Being active in your field helps LinkedIn place you near the right people. Commenting thoughtfully on posts in your niche is the habit we recommend most, and we detail it in our LinkedIn commenting strategy.
We build a commenting tool, so here is the thing we would add that a settings page will not: the connections worth having usually start as recognition, not requests. When you show up usefully in the right conversations, the people PYMK suggests to you are also being suggested you, and by then your name is already familiar. That is the quiet advantage of being present where your future network already gathers.
Frequently asked questions
Where does LinkedIn get the information to recommend connections?
LinkedIn's People You May Know feature draws on shared connections, similar profile information and experiences, working at the same company or in the same industry, attending the same school, and contacts you imported from your email and mobile address books. It combines these signals to suggest members you are likely to know.
Why does LinkedIn suggest people I only met once or barely know?
Usually because of an overlap you may have forgotten, such as a mutual connection, a shared employer or school, or an imported contact. If you once synced your phone or email contacts, someone you exchanged details with can surface even without any LinkedIn activity between you. The overlap, not a memory of the meeting, drives the suggestion.
Does LinkedIn read my messages to suggest connections?
No. LinkedIn explicitly states it does not scan the content of your messages to generate People You May Know suggestions. The recommendations come from network overlaps, profile data, and imported contacts rather than the text of your conversations. That is one of the clearer privacy boundaries LinkedIn draws around this feature.
How can I get better connection suggestions?
Complete your profile with an accurate employer, school, and industry, since those fields power the matching. Build genuine first-degree connections, because mutual connections are the strongest signal. Engaging consistently in your field also helps LinkedIn understand who you belong near. You cannot hand-pick suggestions, but better inputs produce more relevant ones.
Written by the team building Commenti on real LinkedIn growth data. Found an error? Tell us and we’ll fix it. Accuracy beats winning.
Co-founder, Apna Project
Co-founder at Apna Project, where he helps founders and agencies turn ideas into products. Writes here on LinkedIn growth and what actually earns attention in a comment section. Commenti is the LinkedIn comment automation tool by Ampliflow.
If you have ever wondered how LinkedIn recommends connections, you are looking at one of the platform's oldest and most quietly influential features. The suggestions in "People You May Know" are not random, and they are not a black box either. LinkedIn has said quite a bit about what feeds them.
Knowing the inputs is useful for two reasons. It explains the occasionally uncanny suggestions ("how does it know I worked with her?"), and it tells you how to make your own recommendations more relevant.
What People You May Know is
People You May Know, often shortened to PYMK, is the feature on your My Network page that suggests LinkedIn members for you to connect with. It is the main way LinkedIn recommends connections, and for most people it is where a large share of their network actually comes from.
The goal is simple: surface people you probably already know, or would benefit from knowing, so growing your network takes a click instead of a search. The interesting part is how it decides who those people are.
The signals LinkedIn says it uses
According to LinkedIn's own help documentation on the feature, the suggestions are based on a handful of concrete inputs:
Notice what ties these together: they are all forms of overlap. LinkedIn is not guessing who you might like. It is finding the places where your world and someone else's already intersect, then betting that intersection means you know each other.