← Insights

The Instagram Algorithm in 2026: Three Signals Instagram’s Head Named, and What Brand Accounts Should Do

Oct 30, 2026 · 13 min read

The Instagram Algorithm in 2026: Three Signals Instagram’s Head Named, and What Brand Accounts Should Do

Last week your brand account posted a Reel, and it got several times more views than usual. This week you shot another one the same way, and the numbers fell right back to where they started. A lot of people blame “the Instagram algorithm” and decide that running a brand account on Instagram comes down to luck.

In fact, Instagram head Adam Mosseri has been explaining ranking in public all along. In September 2026, he launched a four-part series on ranking, covering everything from why Instagram ranks posts to how creators can earn more reach. As for the three signals that matter most, he named them back in a January 2025 post: watch time, likes and sends. In May 2026, he reminded creators to keep an eye on engagement rates when they read their Insights (Social Media Today).

By the end of this article, you’ll know which three numbers to check each week, what to fix first when they’re low, and which long-standing tips you can let go of.

The three signals of the Instagram algorithm: divide likes and sends by reach

In that January 2025 post, Mosseri suggested creators open their Insights and watch three numbers: average watch time, likes per reach and sends per reach. He said these three signals matter most for ranking. “Sends” here means viewers sending the post to their friends (Social Media Today).

“Per reach” is the key to reading these numbers. Say Post A was seen by 2,000 people and 60 of them liked it: a like rate of 3%. Post B was seen by 20,000 people and 200 liked it: a like rate of only 1%. Post B has more likes, but a much smaller share of the people who saw it chose to like it.

Post A was seen by 2,000 people and got 60 likes, a 3% like rate. Post B was seen by 20,000 people and got 200 likes, a like rate of only 1%

According to a May 2026 report by Social Media Today, Mosseri explained in a post that reach matters, of course. But if you want to know why your reach went up or down, look at engagement rates: of all the people who saw the post, how many actually engaged with it.

He also splits viewers into two groups, followers and non-followers. Likes have more effect on what your followers see, while sends have more effect on what people who don’t follow you see. So when you read your Insights, look at the two groups separately.

The three signals Mosseri named: watch time, like rate (matters more for followers) and send rate (matters more for non-followers)

Instagram’s creator blog also lists retention and comments among the signals it uses to recommend Reels, but these are the three Mosseri singled out.

Signal 1: Watch time. Make people want to watch to the end

In February 2025, Mosseri explained that Instagram looks at both what percentage of a video was watched and how many seconds were actually watched, because they don’t want longer videos to lose out. So you don’t need to cut every video under ten seconds just to raise your completion rate. Do cut the shots that add nothing new, though, such as an overlong logo intro or repetitive filler shots.

On its creator blog, Instagram suggests that every Reel should make sense on its own, hook viewers in the first three seconds and make good use of music and sound. One practical approach is to open with a question viewers want answered.

Beauty brands: Film a wear test of a foundation. The first shot is the makeup at 8pm as you leave work, with a caption asking, “Makeup on at 8am. How much is left now?” Then rewind to putting it on in the morning, with clips from lunch, an air-conditioned room and outdoors cut in. Viewers want to know the result, so they keep watching, and along the way you’ve shown how much to use and how to apply it.

F&B brands can keep it even simpler. A Cantonese roast meat shop films its chef chopping a roast goose in one unbroken shot, from the first cut to the plate, about forty seconds long. The caption is a single line: “The first goose of the day, out at 11.” The sound of the knife, the crisp skin and the glistening fat are enough to hold viewers, and that line tells them the best time to come.

Signal 2: Like rate. Look after the people who already follow you

People who follow you already know your brand. Content made for them can carry more of the details only regulars care about, like new stock arriving, limited-time items, or news about opening hours such as “Closed on Mondays from next week”.

Regulars often see your latest news while scrolling in a spare moment (Photo: Joyce Romero / Unsplash)

Timing matters too. According to Social Media Today’s summary of the September series, Mosseri said recency is one of the main ranking factors, and the feed is roughly in chronological order. Time-sensitive news, like new stock or a limited item, is best posted on the day. When a cha chaan teng (a Hong Kong-style café) launches a limited soup noodle for autumn and winter, it can post that same morning: “Available from today, 50 bowls a day.” Regulars will know to come early.

Lifestyle brands: A homeware shop gets a new batch of handmade ceramic cups and posts a carousel that afternoon: the cups fresh out of the box, the one the shop manager uses herself along with her thoughts on it, and a final slide with how many there are and how to reserve one. For regulars, that post is far more useful than product photos posted at the end of the month.

Signal 3: Send rate. Give strangers a reason to send it to a friend

To reach people who don’t follow you, sends have the bigger effect. Back in May 2024, Mosseri said that in his experience, send rate is more closely tied to overall reach than any other metric.

Instagram’s creator blog also explains how Reels reach non-followers. A Reel is first shown to a small group of people Instagram thinks will enjoy it. If they respond well, it goes to a slightly bigger group, and then wider still.

Content that gets sent usually makes people think of one particular friend the moment they finish watching.

F&B brands: A hot pot restaurant films “Birthday dinner for eight: the best-value way to order”, listing soup base combinations, portions and the price per person. Anyone who sees it may well send it to the group chat that’s planning a birthday dinner.

A beauty salon could film “Five things to know before your first facial”, from what to prepare before booking to how to care for your skin on the day. Viewers will think of the friend who keeps saying they want to try one but still hasn’t booked.

As we wrote in Chasing Views Can Cost You Money. 5 Ways to Make Customers Remember You, when people save a post or send it to a friend, that usually means they’re genuinely interested. Your send rate tells you whether your content moved people who had never heard of you.

Trial Reels: let strangers try it first

In the September series, Mosseri talked about breaking content: Instagram shows posts from small public accounts to non-followers to see how they respond. He also suggested creators try different kinds of content, because trends keep shifting, and Trial Reels are made for exactly that kind of testing (Social Media Today).

Trial Reels let you show a video to people who don’t follow you first. Any public account with 1,000 or more followers can use them (Social Media Today). It takes three steps.

1. Turn on “Trial” before you post

The video is shown only to non-followers. It won’t appear in your followers’ feeds or on your profile grid (Engadget). Since April 2026, you can also schedule Trial Reels (Social Media Today).

2. Check the numbers after 24 hours

You can see the numbers 24 hours after posting. Compare average watch time and sends first. If you want to reach people who don’t follow you, sends matter most.

3. Like the results? Upgrade it

If you’re happy with how it did, you can upgrade it so your followers see it and it shows on your profile (Social Media Today). If you don’t, it won’t clutter your grid.

Trial Reels in three steps: turn on Trial before you post, check the numbers after 24 hours, and upgrade it if you like the results. For public accounts with 1,000 or more followers

Take a scented candle brand launching a new scent. It could film two versions with different openings: one starts with the room after the candle is lit, the other with the perfumer adding essential oil drop by drop. Post both as Trial Reels, then after 24 hours upgrade the better one for your followers. Let real numbers decide the direction.

“Your Algorithm”: users can pick topics, so make your account’s topic clear at a glance

In December 2025, Instagram launched “Your Algorithm” on Reels in the US. Users can see a summary of their interests put together by Instagram, then choose the topics they want to see more or less of (Instagram’s official blog). By June 2026, the feature had expanded across all of Instagram (Feed, Reels and Explore) and rolled out globally in English. There’s no word yet on when a Chinese interface will follow.

Posts recommended in the feed based on the interests a user picked carry a topic label, such as “From Running”. Instagram hasn’t said how this affects brand accounts. Still, when an account’s topic is clear at a glance, people find it easier to remember what you do. If a beauty account posts a skincare tutorial today, a staff dinner tomorrow and a giveaway the day after, it’s hard for anyone to say what the account is about.

One simple fix is to state the topic in the first line of every caption or the first frame of every video, such as “Wear test” or “New arrivals”.

Originality rules: accounts that mainly repost others’ content lose their place in recommendations

In his post for part two of the September series, Mosseri said that beyond engagement, Instagram also values originality, recency and breaking content. Of these, originality has the most to do with how brand accounts work day to day.

In April 2026, Instagram extended its originality rules from Reels to photos and carousels. Accounts that mainly post unoriginal content will no longer appear in the places where Instagram recommends content, though how your followers see your posts isn’t affected (Instagram’s creator blog). Instagram defines original work as something you wholly created, or something that reflects your unique perspective. Re-uploading someone else’s content doesn’t count as original, even if you add a border, a watermark or subtitles, or credit the source in the caption. Adding something meaningful, such as text that gives context, a graphic with new information, or a remix, may count.

Only photos and videos you shoot yourself count as original content (Photo: wenbin sia / Unsplash)

An account found ineligible can regain eligibility once most of its posts from the last 30 days are original, and you can check this and appeal under “Account Status”. The same blog post says original posts now make up 75% of recommended content in the US.

Beauty brands: Say an influencer films an unboxing video of your product, and you download it and re-upload it to your brand account. Even if the caption names the influencer, it isn’t original under Instagram’s definition. Instagram suggests sharing others’ work with a collab post or a remix instead, and the paid partnership label or resharing to Stories works too. You could also remix the video and have a beautician add tips on how to use the product, so it carries your own point of view.

If you want more content shot by creators, you can use Try.Eat!’s influencer matching platform to invite influencers whose style suits yours, then publish with a collab post, which is one of the ways Instagram recommends sharing. F&B brands can take a look at our restaurant marketing plans, and beauty brands at beauty influencer marketing.

Hashtags, caption length, sharing to Stories: do these Instagram tips still work?

In April 2025, Mosseri said hashtags are no longer a primary way to increase reach. According to a December 2025 report by Social Media Today, Instagram will gradually limit each post to five hashtags, saying that fewer, more targeted hashtags can improve both content performance and people’s experience. The same report says Mosseri still sees hashtags as a great way to let people know what a post is about and to connect related posts.

So for a Vietnamese restaurant in Prince Edward, two or three hashtags that say what the post is about, like #princeedwardfood or #vietnamesefood, are enough. Generic tags like #food or #instagood say nothing about the post, however many you add.

Mosseri has answered questions about other small tricks too. He said caption length doesn’t affect reach much either way (Social Media Today, August 2025). And sharing a post to your Stories won’t meaningfully change your overall reach: tricks like this sometimes work, but usually don’t (Social Media Today, April 2026). He also said Instagram makes small changes every day and big overhauls are rare. When reach suddenly jumps or drops, outside factors like holidays or school breaks are the more likely cause (Social Media Today, April 2025).

Running a brand account on Instagram: ten minutes a week on these numbers

In Hong Kong, Instagram ads can reach about 4.05 million people, equal to 54.8% of the total population (DataReportal’s Digital 2026 Hong Kong report, as of October 2025). This is an estimate from Meta’s advertising tools, not a count of active users.

Pick a set time each week, open your Insights, and divide each post’s likes and sends by its reach. Then note down the average watch time. Compare against your own past posts; there’s no need to look at other accounts.

If your like rate is low, check first whether your content for regulars has anything new in it. If your send rate is low, ask whether a stranger would have a reason to send it to a friend. If average watch time is short, start by checking whether the first three seconds hook viewers.

Three numbers to check each week: if like rate is low, check your content for regulars; if send rate is low, ask whether strangers have a reason to send it to a friend; if average watch time is short, check the first three seconds

In the September series, Mosseri said ranking is about predicting how users will respond to what they see, such as whether they’ll keep watching or send it to a friend (Social Media Today). To get those responses, you can start with the questions customers ask most, like “What time does it come out of the oven?” or “Do I need to book?”, and turn the answer straight into a Reel.

Checking the numbers every week and shaping your next post around them takes time, and it takes a clear brand direction. With Try.Eat! Social’s brand social media management, we plan and deliver it all with you, from brand positioning, voice and visuals to every single post, and you approve each post before it goes out. Take a look at the brands we’ve worked with.

FAQ

What does the Instagram algorithm value most in 2026?

Watch time, likes and sends. In January 2025, Mosseri suggested creators look at average watch time, likes per reach and sends per reach (Social Media Today). In May 2026, he stressed again that like rates matter more for followers and send rates matter more for non-followers (Social Media Today).

Which numbers should a brand account check each week?

Average watch time, like rate (likes divided by reach) and send rate (sends divided by reach), with followers and non-followers looked at separately. Compare against your own past posts.

Should brand accounts still use hashtags?

Yes, but you don’t need many. Mosseri said hashtags are no longer a primary way to increase reach (Social Media Today), and Instagram will gradually cap each post at five (Social Media Today). A few that say what the post is about are enough.

Are Trial Reels a good fit for brand accounts?

They’re good for testing a new direction. Any public account with 1,000 or more followers can use them (Social Media Today). The video is shown only to non-followers, and if you’re happy with how it does, you can upgrade it for your followers to see.

Want to know what your brand account could post next? Talk to us.

Cover photo: Collabstr / Unsplash

Hand your next campaign to the right influencers.

Book a 20-minute consultation to find out which plan suits your brand best.

Try.Eat!

The KOL Matching Platform

Try.Eat! is an influencer matching platform that connects brands with creators who genuinely love them. From posting campaigns and influencer applications to post performance, everything happens on one platform.

Download Try.Eat! on the App StoreGet Try.Eat! on Google Play
Copyright © 2026 InnoLab Limited. All rights reserved.