LinkedIn’s algorithm looks at various factors to determine which pages are similar and should be suggested to users browsing the platform. The main factors include:
Industry
Pages in the same or related industries are deemed similar. For example, a page for a software company would likely be suggested to users viewing other tech company pages. Industry similarity is a strong signal for relevance.
Location
Pages for companies in the same geographic area, such as the same city or region, are considered similar. Local businesses often have overlapping target audiences and partnerships. Location is a useful signal, especially for users searching locally.
Company Size
Pages for companies of comparable size – such as in terms of employees or revenue – tend to attract similar audiences. Startups likely share interests with other startups, while enterprise corporations probably have common ground with other large businesses.
Services or Products
Pages offering similar services, products, or solutions are deemed related. For example, CRM software companies would be connected or HR consulting firms would be grouped together. Overlapping offerings indicates overlapping audiences.
Engagement
Pages that receive interactions and engagement from the same users are assumed to have related content. If the same members engage with Page A and Page B, those pages will be suggested together. User behavior provides strong signals.
Followers and Connections
Pages followed or connected to by the same users and pages are considered similar. For instance, if users following Page X also follow Page Y, those two pages will be linked. Shared followers/connections indicate shared relevance.
Keywords and Topics
Pages that contain similar keywords and discuss related topics are connected. Natural language processing identifies topical similarities between pages based on textual content. More keyword and topic overlap increases assumed relevance.
Industry Influencers
Pages that are followed or engaged with by the same industry influencers and thought leaders are treated as similar. For example, a page engaged by top marketing experts would be suggested to users who follow other marketing thought leaders.
Employee Connections
Pages whose employees have overlapping connections and networks are related. For example, if employees of Company A connect with employees of Company B, those company pages are seen as more similar. Shared employee connections signal relevance.
Advertiser Overlap
Pages targeted by the same advertiser campaigns are linked. Pages frequented by audiences from shared ad campaigns are assumed to have relevance to those audiences. Advertiser overlap provides useful signals.
Other Possible Signals
Some other factors LinkedIn may use to determine similar page suggestions include:
- School or university attended by employees
- Academic field of study of employees
- Work experience at common companies
- Participation in shared industry events or groups
- Common investor profiles
- Mentions in the same news articles
- Connections to the same public figure profiles
Conclusion
In summary, LinkedIn uses a diverse range of factors related to industry, location, company traits, user behavior, networks, keywords, topics, and more to determine which pages are most relevant to show to users browsing the platform. The algorithm aims to provide the most useful suggestions to enhance the experience. Location, industry, company size, offerings, engagement, connections, and influencer signals seem to be some of the strongest indicators of similarity.
Factor | Reasoning |
---|---|
Industry | Pages in the same industry have highly overlapping target audiences and relevance |
Location | Local companies often have partnerships and shared local audiences |
Company size | Startups relate to startups, large companies relate to large companies |
Offerings | Overlap in services, products, or solutions indicates overlap in target markets |
Engagement | User behavior signals interest and relevance |
Connections | Shared followers and connections indicate shared relevance |
Keywords and topics | NLP identifies topical similarity from page content |
Influencers | Shared engagement by industry influencers indicates relevance |
While LinkedIn has not publicly revealed the exact algorithm it uses to suggest similar pages, it likely utilizes a weighted machine learning model that considers thousands of overlapping signals – with some factors carrying more importance than others. The model is trained on billions of data points from real user activity on the platform. It gets smarter over time as more data comes in.
For any given page, the algorithm evaluates potential similarity with millions of other pages across multiple dimensions, and predicts which pages have the greatest statistical likelihood of being relevant to the same audiences. The pages with the closest calculated matches become suggested similar pages to optimize recommendations.
By periodically tweaking the machine learning model’s weights and test thresholds, LinkedIn can iterate the system to improve relevance over time. The levels of similarity are not absolute, but represent probability scores that can flex to get better at serving users the most useful recommendations in their given context and interests.
Website administrators and marketers who want to strategically increase a page’s suggested similar pages have a few options:
- Use relevant industry keywords in page content, job postings, and employee profiles
- Encourage employees to engage with desired similar pages by liking, following, commenting, etc.
- Connect with industry influencers by partnerships, content shares, event participation
- Publish blog posts and articles about topical themes related to desired pages
- Sponsor content from or mention desired related pages
- Advertise to audiences targeted by desired similar pages
- Optimize local SEO if aiming for location-based suggestions
Increasing topical, influencer, and user engagement overlap with other pages can organically shift the machine learning model’s recommendations over time. Advertising directly to a desired page’s existing audience can also convince the algorithm of meaningful similarity and shared appeal.
Key Takeaways
- LinkedIn uses sophisticated machine learning to suggest similar pages based on multi-factor statistical models of relevance.
- Industry, location, company traits, offerings, networks, topics, influencers, and user behavior provide strong signals.
- The algorithm balances many different similarity factors to calculate contextual recommendations.
- Website owners can strategically optimize content, advertising, SEO, and engagement to shift suggestions.