Topics · Signal

Dwell time

Contents
  1. What dwell time measures and why LinkedIn uses it
  2. What dwell time means for creators
  3. How it has changed
  4. How long dwell is modeled in ranking
  5. What you can do
  6. See also
  7. References

Dwell time is how long a member spends viewing a post in the Feed, which LinkedIn uses as a clue about what the member finds worth reading, including when they do not like, comment or share1. LinkedIn's research paper on (February 2024) describes a yes-or-no prediction of whether a member spends longer on a post than is usual for similar posts2. For creators, reading time counts even when nobody reacts, and long dwell is judged against what is normal for each content type1. LinkedIn Engineering said in October 2024 that its team was working on spotting "long dwell" posts, which are posts people spend real time on, as a positive sign1.

What dwell time measures and why LinkedIn uses it

LinkedIn counts dwell time as one input to how the Feed decides what to show, and its Help Center lists content you spend the most time viewing among the items in its list on relevance3. In 2020, LinkedIn's engineering team described two kinds of dwell time: time while at least half of a feed update is visible as a member scrolls, and time spent on content after clicking on an update4. LinkedIn Engineering says likes, comments and shares are explicit actions, while dwell time signals, meaning the clues the Feed gets from behaviour, are more abundant and offer broader coverage1. A significant portion of LinkedIn's weekly active users are passive consumers, meaning they read and view without reacting, and for them dwell time is the primary sign of engagement1. LinkedIn Engineering also says dwell time helps judge content from outside a member's network, where interactions are less common1.

What dwell time means for creators

Dwell time bears on creators by format, because LinkedIn Engineering says long dwell is judged against what is normal for each content type1. A video is not compared with an image using one fixed number of seconds, and LinkedIn says it wants to avoid favouring particular formats1. LinkedIn Engineering says that for videos dwell time becomes even more significant1. LinkedIn Engineering also says it predicts when members will have a short dwell time on a Feed post and uses that prediction as a negative sign in ranking1.

How it has changed

  1. 2020

    May

    LinkedIn's engineering team set out why dwell time beats clicks and viral actions alone: it is always measurable, a real-valued measure and not a yes-or-no one, and does not run short of signals4.

  2. 2024

    February

    LinkedIn's research paper on introduced a "long dwell" signal to detect passive content consumption on the Feed2.

  3. October

    LinkedIn Engineering said it had used predicted short dwell as a negative signal for 1.

  4. October

    LinkedIn Engineering said it was evolving to treat passive consumption as a positive signal to learn from and optimize1.

  5. 2026

    March

    LinkedIn Engineering described a Feed model whose actions include long dwells, with long-dwell as a passive task beside click and skip5.

How long dwell is modeled in ranking

From LinkedIn's research and patents, which may not describe how LinkedIn works today

LinkedIn's research paper on describes a model for long dwell in the Feed2. The paper says noisy dwell time data made direct prediction unsuitable, and that a fixed cutoff for long dwell could not adapt to changing preferences and could favour content with naturally longer dwell times2. Its answer was to compare time spent with a percentile, for example the 90th, meaning longer than 90% of comparable posts2. The paper says the cutoffs are set for groups of posts using ranking position, content type and platform and are re-measured daily, and it reports relative gains of 0.8% in overall time spent, 1% in time spent per post and 0.2% in member sessions2. LinkedIn Engineering's October 2024 post says LinkedIn has updated its passive consumption models to better capture member preferences1. LinkedIn Engineering said in March 2026 that the Feed's GR model, which reads a member's posts and actions in time order, includes long dwells among those actions and treats long-dwell as a passive task beside click and skip, with like, comment and share as active tasks5.

What you can do

  • Avoid writing posts only to hold attention, because LinkedIn says it wants to limit "dwell bait" and "click bait" content1.

See also

References

LinkedIn research and patents

  1. 1Leveraging Dwell Time to Improve Member Experiences on the LinkedIn FeedLinkedIn Engineering · Oct 1, 2024
  2. 2LiRank: Industrial Large Scale Ranking Models at LinkedInLinkedIn research paper · Feb 10, 2024
  3. 4Understanding dwell time to improve LinkedIn feed rankingLinkedIn Engineering · May 12, 2020
  4. 5Engineering the next generation of LinkedIn’s FeedLinkedIn Engineering · Mar 12, 2026

Follow changes to Dwell time

Unofficial. Not affiliated with LinkedIn or Microsoft. AI answers from the sources shown.