Threads video in 2026: what the ranker actually rewards

Threads sits on Meta's interest-graph infrastructure — the same ranker family that powers Reels — but the signal weights are different because the surface is different. On Reels, saves are the top intent signal. On Threads, watch-time and replies carry the most weight, followed closely by reposts. Saves matter far less. The platform is built around conversation, and the ranker reflects that: a video that sparks a reply thread gets meaningfully more distribution than a video that gets silently bookmarked.

There are no hashtags in the Threads ranking algorithm. Interest matching is topic-based — the ranker infers what a video is about from the caption text, on-screen text, and audio classification, then matches it to users who have engaged with similar content. Adding hashtags to a Threads post does nothing for reach. What matters is writing a caption that clearly signals the topic, because that caption is the primary text input to the classifier.

Completion rate still functions as a floor signal, as it does on every short-form platform. If viewers bail before the midpoint, distribution stalls regardless of reply volume. The difference from TikTok's cold-start mechanic is that Threads gives a slower decay — the ranker will re-test a post over 24–48 hours rather than making a binary expand-or-stall decision in the first hour.

Threads users arrive expecting text threads. That's the contract the platform made with its audience, and it affects how video lands. A video that opens with a strong text hook on-screen — a tweet-style statement, a counter-intuitive claim, a number that demands explanation — outperforms a talking-head opener on Threads in a way it wouldn't on TikTok. The text hook signals to the user that this video respects the surface they're on.

The caption is not an afterthought. On TikTok, a caption is a label. On Threads, the caption is part of the post — users read it before they tap play, and the ranker reads it to classify the content. A one-line caption on a Threads video is a missed opportunity twice over: you lose the viewer context that drives play-through, and you give the classifier less to work with for interest matching.

The practical standard: write a caption long enough to stand alone as a text post. Two to four sentences that state the claim, add context, and invite a response. Then let the video support the claim rather than carry all the weight itself. Accounts that treat the caption as a headline and the video as supporting evidence are consistently outperforming accounts that post video with a two-word caption.

Threads displays video inline in the feed at 4:5 ratio by default. If you upload a 9:16 vertical video, it gets cropped to 4:5 in the feed — the top and bottom are cut. This is not a bug; it's the intended behavior for the surface. The implication is straightforward: your first frame must be composed to work at 4:5 crop, not 9:16. Any text overlays or faces positioned near the top or bottom of a 9:16 frame will be cropped out in the feed view.

1:1 square format also performs well on Threads and avoids the crop problem entirely. For repurposed content from TikTok or Reels, the fastest workflow is to add a small letterbox or reframe the shot to 4:5 before uploading, rather than relying on the platform crop. The platform crop is center-weighted and won't always keep what you want in frame.

Optimal length sits between 15 and 60 seconds for most content categories. Under 15 seconds, there's not enough watch-time signal to drive meaningful distribution. Over 90 seconds, drop-off rates climb fast enough to suppress the completion floor. The sweet spot for brand content and ads is 20–45 seconds — long enough to make a case, short enough to hold through.

Threads skews sound-off at roughly 65–70% of views, but that's meaningfully lower than TikTok, which runs closer to 80% muted. The gap matters because Threads' text-native audience is more likely to be in a reading context — sitting, scrolling deliberately — rather than the passive lean-back context of TikTok. More intent in the viewing context means slightly more willingness to turn sound on, particularly when the caption sets up audio as part of the payoff.

Captions are still mandatory. Designing for sound-off first is still the right default. But music and audio choices have a secondary function on Threads that's worth understanding: even when viewers are muted, the audio classification layer reads the track and uses it as a topic signal. A track associated with a specific niche or aesthetic sends a category signal to the ranker that helps with interest matching. Music that doesn't require hearing to be effective still contributes to how the video gets classified and distributed.

The dominant creative strategy on Threads today is repurposed Reels content. Most brands and creators are simply cross-posting from Instagram Reels, and the format compatibility (same Meta infrastructure, same aspect ratio norms) means this works reasonably well as a starting point. The problem is that content built for Reels — aesthetic-first, save-optimized, lower caption investment — isn't built for Threads' ranker. Repurposed Reels content gets baseline distribution on Threads, not expanded distribution.

The accounts pulling outsized organic reach on Threads video right now are running a different playbook: brand-voice threads where a founder or operator talks through a real decision, a result, or a mistake. Not a polished brand video — a direct address, honest about the specifics, with a caption that adds context and invites pushback. This format works because it generates replies, and replies are the highest-weight signal in the Threads ranker.

Threads penalizes overtly promotional content in organic ranking — the ranker is calibrated to keep the feed feeling like a conversation, not an ad break. Content that reads as an ad (product shots, pricing, “buy now” language in caption) gets suppressed in organic distribution. The same content running as a paid Threads ad is not affected by this penalty, but organic posts that look like ads will underperform peers with identical watch-time and reply rates. The operator move is to lead with the story or the result, and let the product follow from context.