How TikTok & Instagram actually rank your content

The algorithm isn't a mystery and it isn't out to get you. Both TikTok and Instagram have publicly described how their ranking systems work. Once you understand the signals they reward, you can stop guessing and start engineering content that the feed wants to distribute.

The TikTok For You feed: a watch-time machine

TikTok's own newsroom has explained that the For You feed is a recommendation system that ranks videos using a blend of signals, weighted by how strongly they predict your interest. The biggest levers are:

• User interactions, videos you like, comment on, share, finish, and re-watch.
• Video information, captions, sounds, and hashtags.
• Device & account settings, language and country (a much weaker signal).

Critically, TikTok has stated that watch time, completion rate, and re-watches are among the strongest indicators, a small account can reach millions if a video holds attention. That's why a strong account following is not a prerequisite for reach on TikTok the way it is elsewhere.

The single most controllable ranking signal on short-form video is whether people finish watching. Design for completion.

Source: TikTok Newsroom, "How TikTok recommends content."

Instagram: relationship, interest, and the rise of "sends"

Instagram's head, Adam Mosseri, has repeatedly explained that Instagram isn't one algorithm but several, each surface (Feed, Reels, Explore, Stories) ranks differently. Across them, the recurring signals are:

• Relationship, how often you interact with an account (DMs, comments, profile visits).
• Interest, your past behavior with similar content.
• Relevance / recency, how timely and topical the post is.
• Activity signals, watch time, likes, comments, saves, and especially sends per reach (how often a post is shared in DMs).

Mosseri has been explicit that for growing reach with people who don't follow you yet, sends in DMs have become one of the most important signals, content people share privately with friends gets pushed harder than content that just gets likes.

Source: Instagram's official "How Instagram Feed works" guidance and Mosseri's public Q&As.

The behavioral science underneath the signals

Here's where psychology and platform mechanics converge. The signals the platforms reward, completion, shares, sends, are exactly the behaviors that high-arousal emotion produces. A field study across platforms found that emotional video is shared more than informational video, and that moderate-length videos (~1.2–1.7 minutes) perform best, which maps neatly onto "long enough to land the feeling, short enough to finish."

How to engineer for the signals

• Optimize for completion, not length. Open a loop in the first 2 seconds and close it at the end. Use pattern interrupts to reset attention mid-video.
• Write for the "send." Ask: "Who would someone DM this to, and why?" Relatable, useful, or jaw-dropping content gets sent.
• Trigger comments deliberately. End with a question or a mild, defensible take, comments are a strong interaction signal and they buy you re-distribution.
• Use sound and captions intentionally. They're literal ranking inputs on TikTok and improve accessibility (and watch-time) everywhere.
• Don't over-brand early. Heavy branding suppresses sharing, earn the engagement, then attribute.

You don't beat the algorithm by tricking it. You beat it by giving it exactly the human behaviors it's built to measure, and behavioral science tells you precisely how to produce those behaviors.

Sources: TikTok Newsroom; Instagram official guidance & A. Mosseri; Tellis et al. (2019); Hendra et al. (2025); Peruta & Shields (2017). Research surfaced via Consensus. Create or connect a free Consensus account to return more than 3 results per search in Claude Code.