How would you improve LinkedIn engagement?

The question is a trap if you treat LinkedIn like a generic social feed. The real problem in 2026 is not that users don’t want to engage; it’s that engaging publicly feels professionally risky, and the feed is thin because only ~1% of users produce content. Fix the supply problem, and consumption follows.

What “engagement” means on a professional network

Narrow the definition before proposing anything. Likes and reposts are gamed constantly (engagement pods, reaction bait). LinkedIn’s own 2025-2026 ranking shift tells you what actually matters: the platform replaced click-based signals with a Depth Score measuring actual reading time. The right north star is meaningful professional interactions per active user per week: substantive comments, DMs originating from a post, profile views converting to a connection, and application clicks triggered by content. Not likes. Not shares.

LinkedIn’s business model reinforces this. Talent Solutions, Sales Navigator, and Marketing Solutions pay for intent signals, not impression volume. A feature that inflates vanity metrics while degrading recruiter-to-candidate conversion will not survive a business review.

The root cause: creation confidence, not feed quality

LinkedIn has ~134M DAU against ~320M MAU (DAU/MAU ~0.42, Q1 2026). Most active users lurk. The feed is thin not because ranking is broken but because the supply of quality content is narrow. 450,000+ verified creators produce most of the inventory; the other 99% say nothing. The core barrier is professional identity risk: your boss, your clients, and your next employer all see your reactions and comments.

LinkedIn’s Generative Recommender model rewards topical consistency and expertise signals over recency, so a lurker’s first comment is high-stakes for their own algorithmic profile too. The algorithm already downranks generic AI content, engagement bait, and automation artifacts. AI-assisted drafting without fixing identity risk just produces more posts that get suppressed.

The prioritized bet: contextual reaction scoping

Idea: Let users react or comment with a visibility setting: visible to all / visible only to mutual connections / visible only to the poster. This directly attacks identity risk without breaking the open feed. A lurker who won’t publicly comment on a layoffs post might leave a substantive private note. The poster gets genuine signal. The algorithm registers the depth interaction without broadcasting it.

Why this over other options: Profile improvements and notification digests don’t touch the creation-confidence barrier. Creator monetization rewards the 1% already posting; it does nothing for lurker activation.

Success metrics: New commenter activation rate (first-ever comment from a previously-silent user), Depth Score lift on posts receiving scoped reactions, and recruiter-to-candidate conversion rate on posts with private comment activity.

Trade-off to name: Advertisers use public engagement as a reach signal. Private reactions reduce visible reach numbers. The counter: what they actually pay for is intent, and a scoped substantive comment is a stronger intent signal than a public like from an engagement-pod member.

What the interviewer is checking

They want to know whether you can distinguish a supply-side problem from a demand-side one, and whether you understand that LinkedIn’s engagement problem is structurally different from Meta’s because the stakes of visible engagement are professional, not social. The 2026 bar is higher: a candidate who names the Depth Score shift, the identity-risk barrier, and ties the proposed feature to Talent Solutions revenue will be in a different tier than one who proposes a notifications digest.

See also: north star metric framework, design a news feed (2026), and when two metrics conflict.