Instagram users spent more time on the platform in the third quarter of 2026, with Meta Platforms reporting record-high global time spent per user and double-digit year-over-year growth. Improvements to its artificial intelligence-powered content recommendation systems helped drive the increase, particularly on Instagram Reels, where new recommendation updates introduced during the quarter increased time spent by more than 4%. The results highlight how Meta is using AI to personalize content and keep users engaged for longer.
The gains were disclosed during Meta’s third-quarter 2026 earnings call. The company said its upgraded recommendation architecture can evaluate sequences of Reels together instead of assessing each video independently, while incorporating richer watch-time signals to predict what users are likely to enjoy. Meta is also improving how quickly it recommends new content, with nearly 45% of recommended Instagram Reels now less than one day old. These changes could support advertising performance by helping Meta deliver more relevant content and commercial messages.
Key Takeaways
- Record engagement: Instagram global time spent per user reached an all-time high in Q3 2026.
- More than 4% lift: Instagram Reels recommendation improvements launched during the quarter increased Reels time spent by over 4%.
- Double-digit growth: Overall Instagram time spent per user grew by double digits year over year.
- AI-powered ranking: The upgraded system evaluates sequences of Reels together and uses richer watch-time signals.
- Fresher recommendations: Nearly 45% of recommended Instagram Reels were less than one day old.
- Business impact: Higher engagement can create more opportunities for advertising, although it does not automatically translate into proportional revenue growth.
- Wider Meta gains: Facebook Reels also reached record time-spent levels globally and in the United States.
Instagram Reels Engagement Reaches New High
Meta reported that global time spent per user on Instagram reached an all-time high in the third quarter, growing at a double-digit rate compared with the same period last year. The company attributed the performance to improvements across both Instagram Feed and Reels.
Reels has become a central part of Instagram’s content strategy, allowing users to discover short-form videos from creators and accounts they may not follow. The platform competes for attention with other video and social media services, making recommendation quality an important factor in user retention.
Rather than relying only on the accounts a person follows, Instagram uses machine-learning systems to rank content based on predicted relevance. The systems can consider signals such as viewing behaviour, interactions, content characteristics and other indicators of user interest.
Meta’s latest results suggest that changes to these systems are increasing the amount of time users spend watching Reels. However, the company’s reported engagement figures describe platform-wide performance and should not be interpreted as a guarantee that every user is spending more time on Instagram.
How AI Recommendations Increased Reels Time Spent by More Than 4%
Meta said recommendation improvements launched on Instagram Reels during Q3 increased time spent by more than 4%. The company identified two important elements behind the gains: an upgraded model architecture and richer watch-time signals.
1. Ranking sequences of Reels together
Older recommendation approaches can evaluate individual videos when deciding what to show a user next. Meta’s upgraded architecture can rank sequences of Reels together, helping the system account for how multiple videos may fit into a viewing session.
This approach can improve the flow of recommendations. For example, if a user watches several videos about technology, the system may identify a relevant sequence of additional videos rather than making each recommendation in isolation.
The objective is to produce a more coherent viewing experience and reduce the likelihood that the next video feels unrelated to the user’s interests.
2. Using richer watch-time signals
Meta has also improved the signals its recommendation models use to understand viewing behaviour. Watch time can provide information about whether a person continues watching a video, skips it quickly or remains engaged with a particular topic.
These signals are not perfect indicators of satisfaction, but they can help the system make better predictions when combined with other interactions and content information.
By using more detailed signals, the recommendation system can adjust the videos it surfaces and improve the match between content and individual preferences.
Instagram recommendation performance
| Metric | Q3 2026 result |
|---|---|
| Global Instagram time spent per user | Record high |
| Year-over-year Instagram time-spent growth | Double digits |
| Reels time-spent improvement from Q3 launches | More than 4% |
| Recommended Reels less than one day old | Nearly 45% |
| Main technology changes | Sequence ranking and richer watch-time signals |
Source: Meta’s Q3 2026 earnings call. Figures are company-reported.
AI Helps Instagram Recommend Fresher Content
Personalization is only one part of Meta’s recommendation strategy. The company is also working to show users relevant content sooner after it is published.
Meta said nearly 45% of recommended Instagram Reels were less than one day old in Q3, an increase of more than 11 percentage points from the previous quarter.
Freshness matters because users often want to discover new developments, trending topics and recently published creator videos. If recommendation systems depend too heavily on historical engagement, newer content may struggle to reach an audience before accumulating enough views and interactions.
Meta is investing in models and infrastructure that can identify promising new videos more quickly. Better content understanding helps the system evaluate what a video is about and determine which audiences may find it relevant.
The approach could benefit smaller creators by improving the chances that new content reaches interested viewers. However, a higher share of fresh recommendations does not mean every new video will receive broad distribution. Ranking systems must still balance freshness with relevance, quality, user preferences and safety requirements.
Meta Is Expanding AI Across Facebook and Instagram
The Instagram improvements form part of a broader investment in recommendation technology across Meta’s family of applications.
During the same earnings call, the company said Facebook Reels time spent reached all-time highs globally and in the United States, with double-digit year-over-year growth. A new generative retrieval architecture also increased Facebook Reels watch time by 1%.
The architecture uses semantic representations of content to generate potential recommendations. This helps the system surface relevant posts before they have accumulated substantial engagement.
Meta is also using large language models to understand public content more deeply, including identifying topics and summarizing videos. It said it had begun moving content-understanding models to its own Muse model family, with early offline evaluations showing improvements over previous production models and reductions in per-unit processing costs.
These investments indicate that Meta is developing a more integrated AI system for understanding content, predicting user interests and deciding what to recommend.
Why Higher Engagement Matters for Meta’s Advertising Business
More time spent on Instagram can create additional opportunities to display advertisements, but the relationship between engagement and revenue is not automatic.
Meta’s business depends on both attracting user attention and monetizing that attention effectively. If users spend more time watching Reels, the platform may have more opportunities to show advertising, subject to its ad-load strategy and user experience.
More relevant recommendations can also help users discover creators, products and services that match their interests. For advertisers, improved relevance may increase the likelihood that an ad reaches an interested audience.
The company has separately invested in AI systems designed to improve ad selection and predict which advertisements are most relevant to individual users. Better content recommendations and better advertising systems can therefore support different parts of Meta’s business.
Nevertheless, the 4% increase in Reels time spent is an engagement metric, not a 4% increase in advertising revenue. Revenue also depends on advertising demand, pricing, ad impressions, conversion performance, competition and broader economic conditions.
The Bigger Picture
Instagram’s latest engagement figures show how AI recommendation systems have become central to the competition among social media platforms. Meta is moving beyond basic ranking by combining richer viewing signals, sequence-based recommendations and deeper content understanding. These capabilities help it surface relevant videos, introduce newer content and potentially improve the overall viewing experience.
The next challenge is to sustain engagement while balancing user satisfaction, creator visibility and advertising growth. More sophisticated recommendations can increase the relevance of a feed, but they also require substantial computing resources and careful management of content quality. Meta’s results demonstrate the impact of its latest changes, although long-term success will depend on whether it can maintain those gains without undermining trust or the quality of the Instagram experience.
FAQs
1. Why has Instagram time spent reached a record high?
Meta attributed the increase to AI-powered recommendation improvements across Instagram Feed and Reels. The company reported double-digit year-over-year growth in global time spent per user during Q3 2026.
2. How much did AI recommendations increase Instagram Reels time spent?
Meta said recommendation improvements launched during Q3 increased Reels time spent by more than 4%.
3. How does Instagram’s new recommendation system work?
The upgraded architecture can rank sequences of Reels together rather than evaluating every video independently. It also uses richer watch-time signals to improve predictions about which videos users may find relevant.
4. What percentage of recommended Instagram Reels are less than a day old?
Nearly 45% of recommended Reels were less than one day old in Q3 2026, according to Meta. The company said this represented an increase of more than 11 percentage points from Q2.
5. Does higher Instagram engagement automatically increase Meta’s revenue?
No. Higher engagement can create additional advertising opportunities, but revenue also depends on ad demand, pricing, the number of ads shown, advertiser performance and other business factors.
Looking Ahead
Meta is likely to continue investing in AI systems that understand video content, predict individual interests and recommend fresh material more quickly. The company’s ability to sustain engagement gains will depend on whether its models can improve relevance while adapting to changing user behaviour and the growing volume of content published on Instagram.
For creators and advertisers, these changes make recommendation quality increasingly important. Creators may benefit when relevant new videos are surfaced faster, while advertisers could benefit from better audience matching. However, the long-term impact will depend on how Meta balances personalization, content diversity, monetization and user control as its AI recommendation systems evolve.
Get the day’s top stories in your inbox
One concise email. No spam, unsubscribe anytime.
