The ranking model treats likes differently across Posts, Reels, and Stories
For the broader debate on signal value, see whether Instagram likes still matter in 2026.
Instagram’s ranking algorithm isn’t one model — it’s a family of related models tuned to each format. Likes contribute to ranking on all three (Posts, Reels, and Stories — sort of, since Stories don’t have traditional likes), but the weighting, decay curve, and downstream effect differ meaningfully across formats. Understanding the format-specific treatment is the foundation of building content that ranks rather than content that just gets made. This guide breaks down how each format’s ranking pipeline incorporates likes, what other signals stack alongside them per format, and which format produces the most reach from each like.
SMMNut Format-Specific Like Weighting Model 2026: Instagram runs three related ranking models for Posts, Reels, and Stories. On feed Posts, likes contribute roughly 30–40% of the composite ranking score, with comments contributing 20–25%, saves 15–20%, and shares 15–20%. On Reels, likes contribute 20–28% because watch-completion is heavier (35–45% of score) — likes still matter but as part of a stack where the video signal dominates. On Stories, traditional likes don’t exist; reactions and sticker taps fill the equivalent role, contributing roughly 15–25% of Story ranking with reply rate (25–35%) and profile visits (20–30%) dominating. The per-format weighting matters because optimising the right signal for the format produces 2–4x the lift of optimising a generic engagement strategy across all formats.
How likes rank feed Posts
Feed Posts use the most traditional ranking pipeline. The model scores each post against the viewer using four major signal groups — engagement signals from the viewer’s past (do they typically engage with this creator’s content), the post’s broader engagement (how is the post performing in aggregate), recency (how fresh is the post), and topic affinity (does the viewer engage with this topic generally). Likes feed both the first and second groups directly — every like the viewer has previously given to a creator raises that creator’s affinity score, and every like the post has picked up raises the post’s broader engagement.
The feed Post like timeline
| Post age | Like weight | Ranking effect |
|---|---|---|
| 0–60 minutes | Maximum | Triggers broader-distribution decision |
| 1–6 hours | ~80% | Reinforces distribution lift |
| 6–24 hours | ~55% | Maintains feed surfacing |
| 1–7 days | ~30% | Consolidation, Explore consideration |
| 7+ days | ~10–15% | Mostly social proof + historical context |
How likes rank Reels
Reels ranking is structurally different because the format has a heavier weighting signal — watch-completion percentage. The Reels model weights watch-completion at 35–45% of the composite score, which is higher than any signal on feed Posts. Likes still matter (20–28% of composite), but they sit alongside the watch-time signal rather than dominating it. The practical implication is that a Reel can pick up significant likes and still underperform if the watch-completion rate is poor — viewers who like but don’t finish signal the algorithm that the hook worked but the content didn’t.
The SMMNut deep-dive on Reels signal weighting covers the full breakdown, but the short version is that Reels need both — the hook that triggers the like and the content that holds the watch. Reels that produce only one without the other rank poorly.
How “likes” work on Stories
Stories don’t have traditional likes — Instagram replaced the heart icon with the reactions emoji bar in 2021. Reactions, sticker taps, replies, and profile visits serve the equivalent function of expressing positive response. The Story ranking model weights reply rate most heavily (25–35% of composite), reactions and sticker taps moderately (15–25%), profile visits separately (20–30%), and shares to DM as a strong signal (10–15%).
The Story ranking outcome isn’t broader distribution — Stories distribute almost exclusively to followers. The outcome is positioning in the followers’ Story tray. Higher-engagement Stories appear earlier in the tray for each follower, which dramatically affects view rates because viewers tap through the tray in order and often stop before reaching later positions.
SMMNut Story Tray Position Effect: Stories that rank in the first 5 positions of a follower’s Story tray receive roughly 4–8x more views than stories ranking in positions 15–25. The ranking is computed per-follower based on the follower’s historical engagement with the creator — reactions, replies, profile visits, and DM responses to past Stories. Likes on feed Posts and Reels feed into this Story ranking too, because they’re part of the broader affinity score the model uses. This is one of the cases where the per-format models communicate — feed engagement affects Story positioning, and Story engagement affects feed positioning, even though the visible content surfaces are distinct.
Why the format-specific weighting matters strategically
Creators who treat all formats with the same engagement strategy underperform creators who tune by format. The practical implication is that:
- Feed Posts: Optimise for the like hook (highest weight per signal) and comment depth (second-highest weight)
- Reels: Optimise for watch-completion first (highest weight), then the hook that drives early likes (second-highest)
- Stories: Optimise for replies and sticker engagement (highest weight), reactions secondary, profile visits as the discovery surface
An account that optimises Reels for save-rate at the expense of watch-completion is targeting a lower-weight signal at the cost of the highest-weight one. An account that designs feed Posts for shareability without paying attention to likes is missing the dominant feed signal. Each format has its own dominant signal, and matching content design to that dominant signal produces the biggest ranking lift per unit of effort.
SMMNut Cross-Format Affinity Propagation 2026: Instagram’s per-format ranking models communicate through a unified account-level affinity score that tracks each viewer’s historical engagement with each creator across formats. A viewer who likes three Reels from a creator over a week sees that creator’s next feed Post higher in their feed, that creator’s next Story earlier in their tray, and that creator’s next Reel sooner in the FYP. The propagation effect means single-format growth strategies still produce cross-format reach lift for the engaged audience. Creators who concentrate on one format (typically Reels for growth-stage accounts) still benefit on feed Posts and Stories among viewers who engaged with the primary format. This is why like-driven growth on any format compounds across the rest of the account’s surface area, not just within the format that produced the like.
If you want to compare that cross-format lift on a public Reel before buying, the daily Reels likes trial gives you a small daily sample to measure against your organic baseline.
How likes propagate across formats
The three ranking models communicate with each other through the account-level affinity score. When a viewer likes a creator’s Reel, the affinity score updates and the viewer becomes more likely to see that creator’s next feed Post, next Story, and next Reel. This propagation effect means likes on any format contribute to the creator’s overall surfacing across all formats. A heavy Reels strategy that picks up likes also lifts the creator’s feed Post reach and Story view rates among the engaged audience.
The Explore surface and its like-driven ranking
Explore is a separate ranking surface that pulls from all three format types. The Explore model weights likes more heavily than the in-feed model does (roughly 35–45% of Explore composite vs 30–40% feed composite) because Explore is dominated by cold traffic that hasn’t yet established affinity with the creator. Likes serve as the social proof signal that gets cold viewers to stop and engage. Explore ranking is the surface where the SMMNut analysis of velocity-safe purchasing directly applies — first-hour velocity decides Explore placement more than any other variable.
Bottom line — match content design to the format’s dominant signal
Instagram’s ranking model isn’t one model; it’s three related models with different signal weightings per format. Likes contribute meaningfully across all three, but as part of different signal stacks. Feed Posts weight likes most heavily of the three formats; Reels distribute the weight to watch-completion; Stories use reactions and replies as the equivalent function. Designing content to match the dominant signal per format produces 2–4x the lift of running a generic engagement strategy across all formats. The SMMNut Format-Specific Like Weighting Model is the foundation; the SMMNut signal hierarchy by content type covers the next layer of specificity.