Instagram Reels posted at 3 a.m. gain 2.1 times the views of 9 p.m. posts

Reels published at 3 a.m. Eastern Time averaged 22.2 million views, 2.1 times the 10.6 million recorded at 9 p.m., according to an analysis of 20,247 Instagram Reels released by Adobe Express on June 24, 2026. The full hourly breakdown behind that figure complicates the reading: the platform's most crowded publishing hour, noon, ranked second for views across the entire dataset.

The analysis, titled The best times to post Instagram Reels in 2026, examined when a set of high-ranking accounts published short-form video and how that content performed. It arrives at a moment when short-form video carries an increasing share of both attention and advertising inventory on Meta properties, and when the question of whether publishing time still influences distribution has become commercially consequential for brands buying creator output.

A study of 149 accounts, not of Instagram as a whole

According to Adobe Express, the sample covered 20,247 Instagram Reels uploaded by 149 top content creators, with the day and time of each post recorded in Eastern Time alongside likes, views and comments. The creators were identified by Social Blade ranking. Data collection took place in April 2026, roughly two months before publication.

The arithmetic implies an average of about 136 Reels per account. That density matters for interpretation: the dataset describes the publishing behaviour of a small group of very large accounts, not a cross-section of the platform. Adobe Express does not disclose follower counts, geographic distribution, content language or account category weighting within the sample, nor does it state the period over which the 20,247 Reels were originally published.

Comments were collected as part of the methodology but appear nowhere in the published findings. Views and likes carry the entire analysis.

Where the posting volume actually sits

Publishing activity concentrated sharply around the middle of the day. Noon Eastern Time was the single busiest hour at 9.0 percent of all Reels, against a flat distribution of roughly 4.2 percent per hour. The 11 a.m. slot followed at 7.0 percent, then 1 p.m. at 6.6 percent. Taken together, the five hours from 11 a.m. to 3 p.m. absorbed 33.4 percent of the sample.

The tail runs in the opposite direction. The 10 p.m. hour was the quietest at 2.1 percent, followed by 11 p.m. at 2.3 percent and 9 p.m. at 2.4 percent. The five evening hours from 7 p.m. to 11 p.m. accounted for 12.5 percent of posts between them, roughly a third of the midday block. Overnight hours were not as deserted as the study's framing implies: the seven hours from midnight to 6 a.m. carried 22.9 percent of the sample, with 2 a.m. at 3.8 percent sitting above 6 a.m. at 3.3 percent.

The day-level picture is far flatter. Friday led at 16.0 percent, followed by Thursday at 15.8 percent, Wednesday at 15.6 percent, Tuesday at 15.2 percent and Monday at 14.5 percent. Sunday recorded 11.6 percent and Saturday 11.3 percent. An even split would place 14.3 percent on each day. Friday's lead over Thursday is 0.2 percentage points, and its lead over the weekday average of 15.4 percent is smaller still. The real division in the data is weekday against weekend, not Friday against everything else.

The hours that produced the most likes

Reels posted at 6 a.m. Eastern Time averaged 1,237,000 likes, the highest of any hour, and 1.7 times the 746,000 recorded at the 4 p.m. low. Seven of the ten highest like counts by hour fell between midnight and 8 a.m.

The full hourly series produces an unusually clean division. Every hour from midnight to 9 a.m. averaged at least 1,001,000 likes. Every hour from noon to 9 p.m. averaged 1,000,000 or fewer. The pattern reasserts itself late: 10 p.m. returned 1,025,000 and 11 p.m. 1,095,000, both above the entire afternoon block. Midnight itself ranked fourth at 1,180,000, behind only 6 a.m., 7 a.m. and 5 a.m.

Day-level variation is negligible by comparison. Friday led at 1,030,000, followed by Saturday at 1,020,000 and Monday at 1,002,000, with Wednesday lowest at 962,000. The gap between the best and worst day is 7.1 percent. The gap between the best and worst hour is 65.8 percent. Adobe Express attributes Friday's like total to the higher volume of content published that day, a reading the hourly evidence does not support, since the busiest hour of the day sat in the lower half of the like rankings.

Views, and the shape of the 3 a.m. spike

The views series carries the study's headline. Reels shared at 3 a.m. Eastern Time averaged 22.2 million views, the highest of any hour, against 10.6 million at 9 p.m. and 11.1 million at 4 p.m., the two weakest slots.

The neighbouring hours qualify that result. The 2 a.m. slot averaged 17.0 million and the 4 a.m. slot 14.6 million. The 3 a.m. figure therefore stands 30.6 percent above the hour before it and 52.1 percent above the hour after it, with 1 a.m. lower still at 12.7 million and 5 a.m. at 14.1 million. Rather than sitting on a broad overnight plateau, the peak is a single discontinuous spike surrounded by hours in the mid-teens.

Sample size compounds the point. The 3 a.m. hour held 3.0 percent of the 20,247 Reels, roughly 600 posts, one of the thinnest cells in the dataset. Averages on cells that size are highly sensitive to a handful of Reels reaching exceptional distribution, and Adobe Express publishes no median, dispersion measure or significance test that would separate a stable effect from an outlier artefact.

By day, Monday led with 16.7 million average views, ahead of Tuesday at 16.0 million, Friday at 15.7 million, Wednesday at 15.6 million and Thursday at 15.5 million. Sunday recorded 15.0 million and Saturday 14.2 million. The best-to-worst spread across days is 17.6 percent, against 109 percent across hours. On both metrics, the hour of publication varies far more than the day.

The crowded hour is also the second best for views

The finding that sits least comfortably with the study's own framing concerns noon. The most congested publishing hour in the entire dataset, holding 9.0 percent of uploads, also produced the second-highest average view count at 20.7 million, within 7 percent of the 3 a.m. peak and above every other hour of the day.

That result runs directly against the argument that crowded windows suppress reach. It also carries more statistical weight than the peak it trails: noon covers roughly 1,800 Reels, about three times the volume behind the 3 a.m. figure. Two other well-populated daytime hours performed strongly as well, with 3 p.m. at 17.5 million on 5.6 percent of posts and 10 a.m. at 17.2 million on 5.9 percent.

Adobe Express does not address the noon result in its written commentary, which characterises midday as the crowded window and overnight as the high-impact one. The charts support a weaker claim: evening hours from 4 p.m. to 11 p.m. underperformed on both metrics, while everything from midnight through mid-afternoon performed at broadly similar levels, with two spikes.

The relationship between the two metrics adds a further layer. Dividing average likes by average views produces a like rate of roughly 5.1 percent at 3 a.m. and 4.7 percent at noon, the two highest-view hours, against 8.9 percent at 10 p.m. and 9.1 percent at 11 p.m., two of the lowest. The hours generating the largest view totals converted those views into likes at close to half the rate of the quietest hours. That gap is consistent with high-view hours drawing distribution from recommendation surfaces populated by non-followers, and low-view hours drawing from a creator's existing audience, though the study reports no follower and non-follower split that would confirm it.

A day-level number the charts correct

The written portion of the study states that Mondays generated 22.2 million average views. The published chart shows Monday at 16.7 million and assigns 22.2 million to the 3 a.m. hour, matching the study's own key takeaways and the summary circulated to media. The chart data resolves the conflict: the 22.2 million figure belongs to the hourly series, and the sentence in the body text conflates the two. Adobe Express has not issued a correction, and the erroneous figure remains in the published article text.

The time zone question the analysis leaves open

Every timestamp is expressed in Eastern Time. In April 2026 that meant Eastern Daylight Time, four hours behind Coordinated Universal Time. Converting the reported peaks changes their character considerably.

The 3 a.m. Eastern peak for views corresponds to 8 a.m. in London and 9 a.m. across most of continental Europe. The 6 a.m. peak for likes falls at 11 a.m. in London and noon in Berlin, and noon Eastern, second-ranked for views, at 5 p.m. and 6 p.m. respectively. The weakest windows invert that pattern precisely: 9 p.m. Eastern is 2 a.m. in London, and 4 p.m. Eastern, the low point for likes, is 9 p.m. in London.

Accounts ranked among the world's largest by Social Blade typically hold audiences distributed well beyond North America. A pattern in which the strongest hours in Eastern Time coincide with European daytime, and the weakest with European night, is consistent with an audience geography effect rather than with any property of overnight publishing. The study reports no audience location data and does not test that explanation. For agencies operating outside the United States, the distinction determines whether the figures transfer at all.

Category-level divergence

The category breakdown produces the most operationally interesting result in the dataset. Across all 13 categories reported, the most popular posting time never coincided with the hour that earned the most likes or the most views.

Noon Eastern was the most common publishing slot in eight of the 13 categories, confirming that the midday concentration is a habit across niches rather than an artefact of one large category.

The direction of the best-performing window splits the categories almost evenly. Six peaked for likes between midnight and 5 a.m.: clothing at midnight on Monday, comedy at 2 a.m. on Saturday, finance and fitness at 3 a.m. on Monday, lifestyle at 3 a.m. on Friday and music at 5 a.m. on Friday. Six peaked in the evening: art at 7 p.m., photography at 7 p.m., beauty at 9 p.m., fashion at 10 p.m., tech at 10 p.m. and sports at 11 p.m. Cinema alone peaked in the morning, at 8 a.m. on Saturday.

Likes and views peaked in the same slot for five categories and diverged for eight. Fashion showed the widest split, with likes peaking at 10 p.m. on Saturday and views at 3 a.m. on Tuesday. Sports diverged similarly, at 11 p.m. on Friday for likes against 3 a.m. on Thursday for views.

Adobe Express does not report the number of Reels in each category. With 20,247 posts spread across 13 categories and 168 possible day-hour combinations, the cells behind each category-level peak are necessarily small, and no confidence basis is published for any of them.

What the analysis does not establish

The study is observational. It records when content was published and what it subsequently earned, without controlling for follower count, content quality, audio selection, caption length, hashtag use, paid promotion or whether a Reel was picked up by Instagram's recommendation systems. No causal claim is supported by that design. None of the reported differences carry significance testing or dispersion measures, and averages across cells this thin are heavily exposed to outliers. Collection took place in a single month, which removes any control for seasonality or for platform-side ranking changes.

The definition of the central metric also carries history. Instagram moved to views as its primary performance metric across formats, and Meta subsequently replaced the impressions metric with views in the Page Insights API from November 15, 2025. Views count repeat playback by the same account, unlike reach. A view total measures playback volume rather than audience size, which affects how the 22.2 million figure reads.

Adobe Express holds a commercial position in the outcome. The company sells content creation and scheduling tools, and the study page pairs each finding with a corresponding product, including a Reel maker, a content scheduler and a video editor. Adobe added a conversational AI assistant to Express in October 2025 as part of a broader push into consumer creative tooling. The company's concluding language is more cautious than the framing distributed to media, describing the findings as "a helpful starting point, but consistency and experimentation remain key."

The publication carries a fair use statement restricting the material to noncommercial use and requesting attribution to Adobe Express with a link.

Why timing arguments carry weight for marketers now

Publishing time has become a live commercial question because creator output is increasingly bought as media rather than treated as organic marketing. Creator content is now purchased and measured as a media asset, with United States creator economy ad spend reaching 37 billion dollars in 2025 and projected at 43.9 billion dollars in 2026. Social media accounted for 117.7 billion dollars of the 294.6 billion dollars in total US digital ad revenue recorded for 2025. Scheduling decisions affecting organic reach on a creator's own channel now sit upstream of paid amplification budgets.

The demand for timing data also reflects how little creators can see of the systems ranking them. France's competition authority found in February 2026 that 80 percent of creators hold weak or very weak bargaining power relative to the platforms distributing their work, and issued recommendations on algorithm transparency. In the absence of platform disclosure, third-party datasets fill the gap, and vendors with products to sell are the parties most willing to fund them. YouTube published its own creator marketing research in March 2026, reporting a 79 percent Gen Z trust rate and a 2.3 times long-term return on ad spend advantage over paid social, and Spotter sized a category it calls Creator TV at 6,600 channels in the same month.

Whether publishing time influences distribution at all varies by platform architecture. When X released its recommendation source code in January 2026, analysis indicated that posts are scored in isolation from surrounding feed content, removing the mechanism by which avoiding a congested window would confer an advantage. Meta has published no equivalent detail for Instagram ranking. Shelf life complicates the premise further: public content from professional Instagram accounts became indexable by Google and other search engines from July 10, 2025, extending discovery beyond the hours immediately after publication. Adobe Express states no collection window relative to each post's publication date, leaving open how much of the recorded difference reflects initial distribution rather than accumulated lifetime performance.

For media buyers, the value of the dataset rests less in its specific hours than in what the full charts show once the summary framing is set aside. Creator schedules cluster hard at midday Eastern Time. Evening publication underperformed on both metrics. And the crowded midday slot, contrary to the study's own commentary, was among the strongest hours for views.

Timeline

• July 4, 2025 - Meta's aspect ratio requirements across 25 ad placements are documented, including the multiple ratios accepted for Instagram Reels
• July 10, 2025 - Public content from professional Instagram accounts becomes indexable by Google and other search engines
• August 17, 2025 - Meta confirms deprecation of the impressions metric in the Page Insights API, replacing it with views from November 15, 2025
• October 28, 2025 - Adobe adds a conversational AI assistant to Adobe Express at Adobe MAX in Los Angeles
• November 17, 2025 - Meta details Dolby Vision HDR on Instagram, disclosing a Reels A/B test in which metadata overhead reduced watch time
• January 20, 2026 - X publishes its recommendation algorithm source code; analysis indicates posts are scored in isolation from surrounding feed content
• February 18, 2026 - France's Autorité de la concurrence publishes opinion 26-A-02, finding 80 percent of creators hold weak or very weak bargaining power
• March 2026 - Spotter publishes a 33-page report sizing Creator TV at 6,600 channels
• March 28, 2026 - YouTube publishes creator marketing research reporting a 79 percent Gen Z trust rate and a 2.3 times long-term ROAS advantage over paid social
• April 2026 - Adobe Express collects the day, time, likes, views and comments for 20,247 Instagram Reels from 149 creators
• April 16, 2026 - IAB reports total US digital ad revenue of 294.6 billion dollars for 2025, with social media at 117.7 billion dollars
• May 2026 - Creator content is documented being bought and measured as a media asset, with US creator ad spend at 37 billion dollars in 2025
• June 24, 2026 - Adobe Express publishes The best times to post Instagram Reels in 2026, including hourly and category-level charts

Related PPC Land coverage

• Adobe Express introduces AI assistant for conversational content creation - Covers Adobe's October 2025 addition of a natural language assistant to Express, establishing the product line the timing study is attached to.
• Instagram content becomes searchable on Google starting July 10 - Details the opt-in indexing change that extended the discoverability of Reels beyond Instagram's own recommendation surfaces.
• Meta deprecates additional Page Insights API metrics from November 15 - Explains the replacement of impressions with views in Meta's reporting stack, the metric on which the Adobe Express findings rest.
• Meta enables Dolby Vision HDR on Instagram with compressed metadata - Documents a Reels A/B test in which a 100 kbps metadata overhead measurably reduced watch time before compression cut it fourfold.
• Meta's complex aspect ratio requirements span 25 ad placements - Sets out the format specifications governing Reels and Reels ads across Meta's placement inventory.
• X's algorithm source code drops: what it reveals about the platform's feed mechanics - Analyses the seven-stage ranking pipeline X published in January 2026, including the candidate isolation behaviour that neutralises timing tactics.
• France's watchdog finds YouTube and TikTok trap creators in a broken power game - Reports the French competition authority's finding that 80 percent of creators hold weak bargaining power, with seven recommendations on transparency.
• Creator content is now a media asset - and brands are paying to prove it - Tracks the shift of creator output into paid media planning, with US spend at 37 billion dollars in 2025.
• US digital ad revenue hits $294.6B in 2025 - social and video lead the surge - Provides the IAB revenue baseline for social media and creator advertising against which timing decisions are valued.
• YouTube creator marketing study: 79% Gen Z trust rate and 2.3X ROAS gap with social - An example of platform-funded creator research published in the same period, useful for comparing methodological disclosure.
• Spotter's data reveals 6,600 YouTube creators now rival prime-time TV in reach - Covers a vendor-produced report defining and sizing long-form creator programming against broadcast benchmarks.

Summary

Who: Adobe Express, the content creation division of Adobe, produced the analysis. The dataset covers 149 top Instagram creators identified by Social Blade ranking. The findings bear on marketers, media buyers and agencies managing organic Instagram output alongside paid social budgets.

What: A study of 20,247 Instagram Reels comparing publishing time with recorded likes and views, published with full hourly, daily and category-level charts. Reels posted at 3 a.m. Eastern Time averaged 22.2 million views against 10.6 million at 9 p.m. Noon, the busiest hour at 9.0 percent of uploads, ranked second on views at 20.7 million. Reels posted at 6 a.m. averaged 1,237,000 likes against 746,000 at 4 p.m. Monday led on views at 16.7 million and Friday on likes at 1,030,000. In all 13 categories reported, the most popular posting time differed from the best-performing one.

When: Data was collected in April 2026. The study was published on June 24, 2026.

Where: All timestamps are recorded in Eastern Time. The sample covers Instagram Reels published by globally ranked accounts, with no audience geography disclosed.

Why: The analysis is positioned against the volume of conflicting guidance on publishing times and documents a gap between when creators publish and when the sampled content performed. It is observational, uncontrolled and produced by a company selling scheduling and creation tools. The published charts also contradict part of the study's own written framing, showing the most crowded hour among the strongest for views.