The best time to post on Instagram on Friday, from 850 posts

I pulled 5,734 Instagram posts from 171 public accounts, 850 of them published on a Friday, and divided each post's likes by its own account's median likes, so a two-million-follower account and a two-thousand-follower account count the same. On Friday the working day is a trough and the evening is a peak, and the gap between them is about twice the evening lift in the seven-day Instagram study that used the same method.

The five studies that rank for this question name five different Friday windows, from 3am to 10pm, and none prints a number per hour. So here is the number per hour.

At a glance

Friday after 4pm beats the Friday working day by about a third

Friday posts published between 6pm and midnight ET ran 1.30x posts published between 9am and 5pm, with a 95% interval from 1.09 to 1.60. That's 122 evening posts against 369 working-day posts, resampled by account, so one prolific account can't carry it.

Cut the day into four-hour blocks and the shape is a step. Everything from 4am to 4pm sits at or under normal. Everything from 4pm on sits well above it.

• Midnight to 4am: 1.00x, from 101 posts
• 4am to 8am: 0.95x, from 157 posts
• 8am to noon: 0.90x, from 242 posts
• Noon to 4pm: 0.89x, from 167 posts
• 4pm to 8pm: 1.24x, from 95 posts
• 8pm to midnight: 1.20x, from 88 posts

The best block against the worst is 1.40x, interval 1.06 to 1.84. The two evening blocks are the two thinnest, which is why the interval is wide, and it still clears zero with room.

What would make this wrong: accounts that save their best content for Friday evening. This is observational, and a creator who posts a throwaway at Friday lunch and the week's best reel at 6pm makes 6pm look good. The share of posts that beat their own account's median tells the same story, though: 56% in the 4pm to 8pm block and 42% from noon to 4pm. So it isn't one viral post carrying the block. Most posts in it did a little better.

The morning has no peak, and the midday dip is real

Friday posts from 5am to 9am scored 0.99x the working day, interval 0.78 to 1.25. There's no 6am or 7am lift here. The smoothed curve drifts from 1.10x at midnight down through normal by 2am and stays there until it bottoms out at noon, at 0.84x.

That noon trough is the single lowest hour in the study, and it's also the hour accounts post most. Just over a third of Friday posts went out between 8am and 1pm. About one in seven went out after 6pm. If a scheduler is choosing between those two slots for you by default, it's picking the wrong one on Friday.

Read raw, without smoothing, the individual Friday hours bounce between 0.60x and 1.44x on cells of 13 to 71 posts, which is why the chart smooths each hour over its two neighbours on either side. Every smoothed window holds at least 96 posts. The 2pm cell is the one to distrust in either direction: 26 posts at 0.60x raw, and 0.94x once its neighbours are in.

Friday is an average day, and Friday evening is an unusually good evening

Across the full week, Friday sat at 0.98x an account's normal likes. Monday was 0.95x, Tuesday 0.98x, Wednesday 0.93x, Thursday 0.99x, Saturday 1.05x, and Sunday 1.08x. Friday against the other six days is 0.99x, interval 0.91 to 1.07: nothing.

What is different about Friday is the shape of the day. Friday daytime posts (8am to 4pm) ran 0.93x other days' daytime posts, interval 0.82 to 1.05, and Friday evening posts ran 1.19x other days' evenings, interval 1.01 to 1.41. On the other six days, the 4pm to 8pm block ran 1.04x against 0.97x at noon to 4pm, a seven-point gap. On Friday the same two blocks are 35 points apart. Sprout Social's and SocialPilot's read, that people check out of the feed as Friday afternoon wears on, fits the daytime half of that. The evening half says they come back.

Two things this sample can't tell you. The 7 accounts over a million followers had 29 Friday posts between them, so there's no large-account result. And the earlier Instagram study, on a fifth of this sample, had Friday at 1.09x and Thursday to Saturday as the good block; on four times the posts from the same accounts, that block is 1.05x with an interval that includes 1.00, and the weekend is what's left standing. A day effect that small moves when the sample does. The Friday evening effect is several times its size.

How this compares to what is already published

• Buffer, from 9.6 million posts across 200,000 accounts, names 10pm, then 9pm, then 6am as the best Friday times, and rates Friday "Low" among the days.
• Sprout Social, from nearly 2 billion engagements across about 307,000 profiles, says there are no significant peak hours on Fridays because users check out for the weekend early.
• Hootsuite, from over 1 million posts in 118 countries, names around 4pm.
• Later, from over 6 million posts, names 3am to 6am.
• SocialPilot, from 7 million posts across about 50,000 accounts, names 7am to 8am.

Buffer's 9pm and 10pm and Hootsuite's 4pm sit inside the window this sample found. Later's and SocialPilot's early mornings don't; here the morning was flat against the working day. Sprout's "no peak" is the odd one out, and its data window is three months of winter, from late November to late February.

The methods land in different places because they measure different things. Every one of the five ranks engagement across accounts, which means the hour the biggest accounts prefer wins by default, and every one converts to the audience's local time, which this study doesn't. None prints a per-hour value or a sample size per hour, so there's no way to tell whether Later's 3am is a thousand posts or a hundred. Scoring each post against its own account, and showing the count under every bar, is the whole difference.

How I measured it

The crawl ran through one Instagram endpoint on AnyAPI, instagram.user_posts, in 657 calls, and cost $0.53.

1. Start from the accounts already in the sample. The 211 accounts came from the earlier Instagram study, which found them through 40 creator-niche searches and picked them round-robin across four follower tiers. No new discovery, so the two studies describe the same population.
2. Pull deeper. instagram.user_posts returns twelve posts per call with a timestamp and a like count, plus a cursor for the next page. Each account was paged up to four times, stopping early once a page reached back past the window. The median account contributed 38 posts.
3. Apply the cutoffs. Posts had to be published between 2025-09-22 and 2026-09-08, so every like count had at least 14 days to settle. Accounts needed five scorable posts. That dropped 1,969 posts as too recent or too old, 10 posts with zero likes, and 40 accounts that fell under five, leaving 171 accounts and 5,734 posts, published between 2025-09-23 and 2026-09-08.
4. Score each post against its own account. Divide each post's likes by that account's median likes over the window, and cap the result at 6x so one viral post counts once. 433 posts hit the cap.
5. Bucket and average. Convert every timestamp to US Eastern. Group by weekday, by four-hour block, and by hour, and take the geometric mean of the scores in each group. Hourly values are smoothed over a five-hour window. Every interval in this post comes from resampling accounts, not posts, 1,000 times.

Here's one page of the crawl, on one account in the sample:

curl -sS -X POST https://api.getanyapi.com/v1/run/instagram.user_posts \
  -H "Authorization: Bearer $ANYAPI_API_KEY" \
  -H 'Content-Type: application/json' \
  -d '{"handle":"_garden.ing"}'

{"output":{"data":{"posts":[{"id":"3988678296405120718","createdUtc":1789707527,"likes":804,"comments":217}],"nextCursor":"..."}},"costUsd":0.0008}

The sample: 69 accounts under 10k followers, 63 between 10k and 100k, 32 between 100k and 1M, and 7 over 1M. The typical account's typical post had 112 likes.

The limits of this data

• Likes only. Nothing here speaks to reach, saves, shares, comments, or follower growth, and Instagram's ranking leans on sends and saves more than likes.
• US Eastern for everyone. Accounts were found globally and no account was mapped to its audience's time zone. Friday 6pm in New York is Saturday morning in Sydney, and some of the evening lift is somebody else's weekend.
• Thin evening cells. The two blocks that carry the finding hold 95 and 88 posts. The interval on the headline ratio runs from 1.09 to 1.60, so the true effect could be a tenth or it could be half again.
• Observational. Content, format, and promotion all confound this. An account that posts its best work on Friday evening makes Friday evening look good. The beat-the-median share (56% against 42%) says it isn't one or two posts, and says nothing about why.
• A convenience sample. The accounts came from 40 niche searches a year ago. A different set of niches could shift a result this size.
• Current follower counts. An account that grew during the year is labeled by today's size, and the tier cuts have 7 accounts at the top.
• Same accounts, new number. The earlier study's Friday 1.09x and this study's 0.98x come from the same 211 handles. The difference is four times the posts. Treat day-of-week effects under about 10% as sample noise until two samples agree.

The hour effect on Friday is about 30%, and it's the one number in this post I'd change a schedule over.

What to do with it

1. Move your Friday post to after 4pm in your audience's time zone. Anything from 4pm to midnight scored above normal. Noon to 4pm scored the lowest of the week.
2. Don't move it earlier. The morning scored the same as the working day. There's nothing to gain before 4pm.
3. Don't move it to another day on the strength of the day effect. Friday is an average day. Saturday and Sunday ran about 10% better than the weekdays, and that clears zero, but the earlier study on the same accounts had Sunday at exactly normal, so give it a second sample before you move.
4. Check your own account once it has thirty Fridays. Pull your posts with instagram.user_posts, page past the newest twelve so you're not scoring posts that are still collecting likes, divide each post's likes by your median, and group by hour in your audience's time zone. Ignore any hour with fewer than about ten posts. Four calls is about a third of a cent.
5. Give the new slot a month before you trust it, then measure again.

Frequently asked questions

What is the best time to post on Instagram on Friday?

In this sample, between 4pm and 8pm US Eastern, at 1.24x an account's normal likes, with 8pm to midnight close behind at 1.20x. The single best smoothed hour is 6pm.

What is the worst time to post on Instagram on Friday?

Noon to 4pm, at 0.89x an account's normal likes. Noon itself is the lowest smoothed hour of the day at 0.84x, and it's also the hour accounts post in most.

Is Friday a good day to post on Instagram?

It's an ordinary one. Across 5,734 posts, Friday sat at 0.98x, level with Monday through Thursday and a few points under the weekend. What's unusual about Friday is how much the hour matters: the evening ran about 30% above the working day, against a few points on other weekdays.

Why do other studies name a different Friday time?

They rank raw engagement across accounts, so the hour the biggest accounts favour wins, and they convert to each audience's local time, which this study doesn't. Buffer and Hootsuite land inside the 4pm to 10pm window found here. Later and SocialPilot name the early morning, which was flat in this sample.

The Instagram study across all seven days uses the same method on the same accounts, the TikTok and Facebook versions run it on those platforms, and the Instagram endpoints are what you'd use to rerun it on your own account.

Use data responsibly and follow AnyAPI's Acceptable Use Policy.