Instagram algorithms: official updates from 2024 to 2026

Review date: 19 September 2026. This guide covers selected official announcements between 19 September 2024 and 19 September 2026. It is a source-based reading guide, not a complete release log of every experiment. The linked documents do not provide one numbered algorithm version or a universal set of ranking weights.

First, distinguish the types of update

A ranking-system explanation describes how recommendations are produced. A creator tool changes what you can do. A preference control changes what a viewer asks to see. An educational resource explains existing practices. Treating every new feature as a ranking change can lead to unnecessary changes in your content plan.

October 2024: official guidance inside the dashboard

On 1 October 2024, Instagram announced Best Practices in the professional dashboard, covering creation, engagement, reach, monetisation and guidelines, including personalised tips. This was an educational launch, not a published ranking formula. Meta’s announcement.

Practical interpretation: check the guidance available in your own account before following a universal posting rule from an unrelated account. Availability and the interface may differ.

November–December 2024: recommendations reset and Trial Reels

The 19 November 2024 announcement described testing a reset for recommendations across Explore, Reels and Feed, followed by renewed personalisation through interactions. The page also carries a July 2025 update date; that is not a separate algorithm version. Recommendations reset.

On 10 December 2024, Instagram introduced Trial Reels, initially showing a trial to non-followers. Its documentation describes insights and the option to share more broadly afterward. Eligibility matters; do not assume every account has the same controls. Trial Reels announcement.

Practical interpretation: a trial is a way to examine an idea, not a promise of distribution. Record what you changed and compare relevant examples rather than treating one trial as a verdict on your entire business.

May 2025: an explanation of a system with many models

Meta Engineering’s 21 May 2025 article describes more than 1,000 machine-learning models supporting Instagram’s recommendation systems across surfaces. This is an engineering explanation of scale and experimentation, not a list of creator-facing ranking weights. Engineering at Meta.

Practical interpretation: a claim such as “one save always equals five likes” needs evidence the source actually supplies. This document supplies no such universal conversion rule.

December 2025–July 2026: more control over interests

Meta’s 12 May 2026 announcement dates the launch of Your Algorithm to December 2025. At the time of that May article, it described the feature on Reels and Explore in English-speaking countries and Feed as upcoming. May 2026 official context.

A 17 July 2026 Meta report describes the tool across home Feed, Reels and Explore in a US context. This documents a later state; it does not establish an exact worldwide rollout date. July 2026 report.

Practical interpretation: audience interests are something viewers can actively shape. Keep your subject understandable instead of trying to attach unrelated trending topics to every post.

January 2026: a scoped report about original content

On 28 January 2026, Meta reported that original posts represented 75% of Instagram recommendations in the US, following a ten-percentage-point increase in Q4 2025. This is a regional historical metric, not a global quota or an account-level guarantee. Meta’s performance report.

Practical interpretation: make your own demonstrations, examples and explanations. The report does not establish that using a particular editing application, effect or publishing schedule guarantees a result.

How to evaluate the next algorithm claim

• Open the original announcement and record its publication and revision dates.
• Confirm that it concerns Instagram, not a Facebook-only change.
• Identify the surface: Feed, Reels, Stories or Explore.
• Check country, language, age and account eligibility where stated.
• Separate a measured past result from a future promise.
• Mark your own practical conclusion as an interpretation, not an official rule.

Keep a small content log with the question answered, format, opening and audience feedback. Change one creative decision at a time where practical and review several comparable posts. For planning and quality checks, use our business video content guide. No verified source here supplies a universal number of hashtags, a guaranteed posting hour or a guaranteed follower outcome.