The top 10% of reels by saves capture 78% of all reach.

How we got this data. Creators connect their Instagram to Dally via OAuth. This gives us access to saves, shares, and reach — metrics private to the account owner. All data is anonymized and pooled. No individual creator is identifiable. Total dataset: 6,197 posts (3,162 video, 798 carousel, 524 image) from creators with 300 to 527,000 followers. We filtered to 4,484 posts with both saves and reach data for rate analysis.

1. Reach follows an extreme power law

We sorted all 3,162 reels by save count and split them into ten equal groups (deciles). The concentration is more extreme than Pareto: not 80/20, but closer to 78/10.

The bottom 80% of reels (by saves) account for just 13.4% of all reach and 2.9% of all shares. One in ten reels generates nearly all the distribution. The other nine are, algorithmically, invisible.

2. Shares explode around 1,000 saves

We split the data into 20 equal groups (ventiles) to find where the curve bends. The answer: between the 90th and 95th percentile of saves.

The jump from the 90th to 95th percentile: saves increase 2.4x, but shares jump 4.4x. Reach doubles. Something changes in how the algorithm distributes content once saves cross roughly 1,000. Shares compound disproportionately from that point.

We observe a correlation, not a proven cause. The algorithm is opaque. But the pattern is consistent across our dataset: the share-to-save ratio inflects around this threshold.

3. The share-to-save ratio is a U-curve

We expected shares to track saves linearly. They don't. The ratio of shares per save follows a U-shaped curve across save buckets.

Low-save posts (1–4) get shared nearly as much as they're saved — small audiences where saving and sharing overlap. The valley is at 100–499 saves: content that gets bookmarked for personal reference but not passed along. At 5K+ saves, the ratio climbs back to 0.63 — viral content that people both save and share.

The 100–499 range appears to be "reference content" — useful enough to bookmark, not compelling enough to distribute. The extremes are social acts. The middle is a private one.

4. Carousels are for saving. Reels are for sharing.

We computed pool-level rates (total saves divided by total reach, avoiding the "average of ratios" distortion) across all 4,484 posts with engagement data.

Carousels and images have dramatically higher save rates (13–26%) but almost zero shares. Reels have a lower save rate (1.94%) but their share rate (1.14%) is 8x higher than carousels.

Shares drive algorithmic distribution. Reels average 706 shares per post vs. 11 for carousels and 1 for images. This is why reels dominate reach (61,923 avg) despite a lower save rate: they get distributed through shares.

Note: save rates above 100% are possible for older posts because saves accumulate indefinitely while reach measures unique viewers during initial distribution. Image and carousel save rates are inflated by this effect. Rate comparisons are most reliable within the same format.

5. Small creators' posts reach far beyond their followers

We grouped creators by follower count and measured how far their content travels relative to audience size.

Creators with fewer than 5,000 followers reach, on average, 5x their follower count per post. Their content is overwhelmingly distributed to non-followers. Creators with 200K+ followers reach only 21% of their audience — their content stays mostly within their existing community.

Caveat: this is 13 creators across a wide range. The 50K–200K bucket has a single creator. These are directional observations, not benchmarks. The pattern is consistent with Instagram's known bias toward distributing content from smaller accounts to the Explore and Reels feeds.

What this means

The data points to one thing: saves are the leading indicator. They're the earliest private signal of content quality — they precede shares, which drive reach, which compound distribution. Saves are also the metric Instagram keeps hidden from everyone except the creator.

Most analytics tools work with public data: likes, comments, follower counts. Those metrics are visible to anyone. Saves and shares require the creator's own API access. Without OAuth connection to the creator's account, the data simply doesn't exist for outside observation.

This is why we built Dally. When creators connect their Instagram, their private metrics become visible — not just to them, but pooled anonymously across the network. The more creators who connect, the sharper these benchmarks become.

About this analysis

Data from 13 creators who connected their Instagram to Dally between March and April 2026. Follower counts: 300 to 527,000. Total posts: 6,197. Posts with both saves and reach data: 4,484 (3,162 video, 798 carousel, 524 image). All data via Instagram's official API (OAuth). No scraping, no estimation.

Limitations: 13 creators is directional, not statistically conclusive. Creator-size analysis is especially limited (1 creator in the 50K–200K bucket). Saves are cumulative while reach is point-in-time, creating measurement asymmetry for older posts. All observations are correlational — Instagram's algorithm is opaque and we cannot establish causation from observational data.

This analysis will be updated as the creator pool grows. If you're a creator and want your data included, connect your Instagram at dally.ai.

See your own saves, shares, and reach

Connect your Instagram. Dally shows you the metrics that predict performance. Free.