Instagram Algorithm 2026: What Mosseri Actually Says

What Did Adam Mosseri Actually Say on September 10, 2026?

Australia is debating opt-out algorithm laws as part of its Digital Duty of Care framework. In response, Instagram head Adam Mosseri gave ABC News a September 10, 2026 interview arguing against making a chronological feed the default. His central claim: forcing users off the ranked feed makes the product worse for the people it is supposed to help.

ABC quoted him directly: "If you go to a chronological feed, it ends up getting overwhelmed with content from companies, brands, and publishers," and "creators get drowned out, and then your friends get drowned out even more." He added that when Instagram measures satisfaction on both an aggregate and a post-by-post basis, in a chronological feed "those numbers fall off a cliff."

The important part for creators is what he said next about the option that already exists. Chronological Following is live globally, including Australia — you tap the Instagram logo and choose Following. But it resets to Home when you close and re-open the app. Mosseri conceded to ABC that "anything that isn't the default just doesn't get used the majority of the time" and guessed Australian uptake sits "in the low single-digit percentages." He said Instagram is not "trying to hide it" and there is "definitely room to improve the discoverability of that."

What Are the Three Ranking Signals Mosseri Keeps Naming?

Eclincher's July 24, 2026 explainer states it plainly: "In January 2025, Mosseri publicly confirmed the three ranking factors Instagram weighs most heavily, and they still define 2026." Eclincher names them as watch time, sends per reach (how often your post is shared in DMs), and likes per reach — and reports Mosseri "has said sends carry roughly three to five times more weight than likes when deciding whether to push content to non-followers." The watch-time URL is updated on the watch-time page.

Sprout Social's January 28, 2026 update reframes the same signals from Instagram's side. It says "across all surfaces, the Instagram algorithm now prioritizes these signals, according to Adam Mosseri" and puts three at the top: sends per reach as the strongest distributability signal, Views as "the primary currency" (counted for every appearance on screen, including repeat views), and originality as an aggregator penalty — reposts get replaced with the original creator's version in recommendations.

Eclincher also says the signal weighting has shifted "away from passive engagement": watch time and sends count more than likes; saves count meaningfully higher than likes because they indicate lasting value. Two implications follow. Likes-per-reach still moves connected reach (your existing followers). Sends and watch time move unconnected reach (Explore, discovery, cold audiences). The mix you optimise for depends on which audience you are trying to grow.

For a companion read on how the Views change plays out at the metric level, see our Views-first system deep dive.

How Does Each Instagram Surface Weigh Signals Differently?

Both Eclincher and Sprout Social separate the surfaces. The table below combines their summaries — each row is a paraphrase, not a Meta document, so read it as directionally correct rather than an internal leak.

The practical implication Eclincher spells out: "a strategy that wins on Reels will not automatically win in Feed." Same footage cross-posted to all four surfaces underperforms on at least three of them. You get more from tailoring the hook, pacing, and CTA per surface — even if the underlying idea is the same.

Four More 2026 Shifts to Plan Around

Four specific 2026 shifts show up in both Eclincher and Sprout Social. All are sourced to Instagram or Mosseri, not to secondary interpretation.

Longer Reels are Explore-eligible. Eclincher: "Longer Reels, up to three minutes, can now reach non-followers through recommendations, so you are no longer penalized for a longer story as long as it holds attention." Sprout: Instagram "is now actively rewarding longer storytelling by recommending them in the Explore feed." Neither source promises longer means more reach.

Aggregator accounts get suppressed. Sprout Social: "Instagram now actively penalizes 'aggregator' accounts that repost content without adding significant value. If you repost a meme or video, the algorithm will often replace your post with the original creator's version in recommendations." Eclincher goes further on volume: accounts that repost without adding value, "and especially those posting large volumes of reposts in a short window, can be excluded from recommendations entirely."

Hashtags carry less weight; captions do the search work. Eclincher: "Hashtags have lost most of their weight since late 2024; five to ten relevant, niche tags are plenty, and keyword-rich captions now do more for discovery, since Instagram Search increasingly reads captions, on-screen text, and audio like a search engine."

The polished aesthetic is dead — Mosseri said it himself. Engadget (December 31, 2025) quoted him arguing it "will be more practical to fingerprint real media than fake media" as feeds fill with "synthetic everything." The feed of "polished" square images, he says, "is dead." Engadget reports he says more "raw" and "unflattering" images will be how creators can prove they are real, not AI.

What Should Multi-Account Operators Do About the 2026 Signals?

If you run one account, "post the mix Mosseri suggests" — Eclincher reports his suggested cadence is "roughly two Reels plus three to five feed posts per week" — and design each post for either sends or watch time is the entire strategy. If you run N accounts, the math changes.

Four ranking systems × N accounts is a lot of surface-specific decisions. The failure modes we see most often across operator portfolios: identical Reels cross-posted to five accounts (aggregator penalty risk), identical caption templates (originality signal drops), or the same 30-second video re-uploaded five times to different niches (the send signal collapses because friends only send one). This is a content-planning problem before it is an infrastructure problem — but infrastructure limits what content-planning can rescue.

Our position on infrastructure has not moved: dedicated real phones, isolated environments (each account on its own device with its own IP), and proprietary sequences built on hours of experimentation. On top of that, we plan content around the four Instagram surfaces separately per account, and we keep an originality margin — no account reposts another account's Reel; each account has its own hooks, its own captions, its own cadence — so the aggregator penalty stays off the entire portfolio. See our multi-account strategy guide for the operator-level breakdown, and our Instagram growth strategy write-up for the single-account variant.