TikTok explains how the For You feed recommends videos
TikTok's own account of its For You feed lists three groups of signals and says strong signals count for more than weak ones. It is a company description, not an audit, and it gives no weights.
For a psychology course
Separate the documented system or audit from an inference about human behavior. Identify the objective, observation and comparison in this record, then ask what additional evidence would be needed for a causal claim about people.
This record was transferred from Popular Psychology with its original claims, sources and reading limits. Its historical review date does not establish that the current platform works the same way.
- Case type
- Platform
- Context
- Consumer
- Organizations
- Tiktok
What happened
On 18 June 2020 TikTok published a newsroom post titled How TikTok recommends videos #ForYou. It describes how the For You feed picks videos and groups the inputs into user interactions, video information, and device and account settings. A page on TikTok's transparency site adds that the system predicts how likely a user is to like, share, comment on or skip a video and combines the predictions into an overall score for each video.
The psychological idea
TikTok describes a ranking system in which signals carry different weights. It says strong signals, such as finishing a longer video, count for more than weak ones, and that device and account settings count for less because users do not actively choose them as preferences. It says follower count and past high-performing videos are not direct factors. It also says the feed generally will not show two videos in a row with the same sound or by the same creator.
In practice
The post gives examples of each group: videos liked or shared, accounts followed, comments and content created; captions, sounds and hashtags; language, country and device type. The transparency page describes user controls named Not interested, keyword filters and a For You refresh. Neither page says how much each signal or prediction counts. The post also mentions a transparency center in Los Angeles where invited experts can learn how the algorithm works, with source code to be made available there for testing and evaluation.
Why it mattered
It is TikTok's own published account of the For You feed, and a company description of this kind gives outside testers something to compare with their results. The post states that finishing a longer video is a strong signal, which is a point that independent experiments can check.
Limits and qualifications
This is company-self-report, not an audit. The pages read give no weights, model details or full list of signals. The offer of expert access to source code is a company statement; no report of who reviewed what was opened for this record. A list of signals does not show what effect the feed has on viewers or what content any person sees.
Evidence behind the claims
On 18 June 2020 TikTok's newsroom listed three groups of recommendation inputs: user interactions, video information, and device and account settings.
TikTok says strong signals such as finishing a longer video get more weight than weak ones, and device and account settings get lower weight because users do not actively express them as preferences.
TikTok says neither follower count nor previous high-performing videos are direct factors.
TikTok says the feed generally will not show two videos in a row made with the same sound or by the same creator.
TikTok's newsroom post says invited experts can learn how the algorithm works at its Transparency Center in Los Angeles, and that source code will be made available there for testing and evaluation.
TikTok's transparency page says the system predicts how likely a user is to like, share, comment on or skip a video and combines the predictions into an overall score for each video; it does not disclose the weights.
TikTok's transparency page lists user controls: Not interested, keyword filters for the For You feed, and a For You refresh.
People involved
Course connections
- OpenStax Psychology 2e, section 2.1: Why Is Research Important?Context. The section discusses this topic.
Opened the page: it urges asking about a claimant's expertise and what they might gain, which applies to a company describing its own system.
Course relevance is not evidence for the historical or corporate claims.
Sources and inspection
Inspection statements below preserve the original reading scope. A source marked as not verified supports no claim by itself.
- [1]How TikTok recommends videos #ForYou. (2020). TikTok Newsroom. https://newsroom.tiktok.com/en-us/how-tiktok-recommends-videos-for-you. TikTok Newsroom. 2020. Company document. Company document.
Company document read. Reviewer re-opened the TikTok Newsroom post through the fetch tool and read the date, the three signal groups, the weighting passage, the follower-count passage, the diversity passage and the Transparency Center passage as returned by the tool.
- [2]Recommendation system. (n.d.). TikTok Transparency Center. https://www.tiktok.com/transparency/en/recommendation-system/. TikTok Transparency Center. Undated. Company document. Company document.
Company document read. Reviewer opened the transparency page: prediction and overall-score passage, no weights disclosed, and user controls. No publication date was shown. A footer year was previously used as a placeholder; it is now null and is not treated as a publication date.
Access note: Both sources are TikTok's own pages. No independent review of the source code or of the weighting was found or opened.
Original independent review record
Claude review agent, independent of the research agent. 2026-10-21. Verdict: pass-with-edits.
- Added a claim for the Transparency Center and source-code statement, which the prose used.
- Removed the reference to later records in the batch and softened an unsupported statement about later independent tests.
- Reworded the claim on weights to include the stated reason for lower device-setting weight, as the post says it.
- Set the s2 year to 2025 from its footer and rewrote the audit notes to describe the reviewer's own re-opening.
- Codex audit: replaced an unverified publication-year placeholder with null; the actual historical milestone date remains unchanged.