Instagram has said the signals that matter most for Reels are watch time, likes per reach and sends per reach, with sends weighing more for reaching non-followers. TikTok weighs strong engagement signals such as watching to the end and sharing, and says follower count is not a direct factor. Both now favour original content.
Algorithm advice is full of guesses. This post sticks to what the platforms have said publicly, labels everything else as an estimate, and then turns it into something practical: how to choose which competitor posts are worth your research time.
What Instagram has confirmed
The big three. In January 2025, Adam Mosseri said the signals that matter most are watch time, likes and sends, and that creators should watch average watch time, likes per reach and sends per reach in their insights. Likes are slightly more important for content shown to followers; sends are slightly more important for content shown to people who do not follow you (Social Media Today).
Note the phrase "per reach". These are rates, not totals. A Reel seen by 1,000 people and sent by 60 of them is a stronger signal than one seen by 10,000 and sent by 80.
How Reels are ranked for each viewer. Instagram's own explainer lists the main inputs for Reels as your activity (what you have liked, saved, reshared and engaged with), your history with the poster, information about the reel itself (audio, visuals, popularity) and information about the person who posted (Instagram, May 2023).
Originality. Instagram recommends originals over reposts, and in 2026 extended that to photos and carousels, removing accounts that mostly post unoriginal content from recommendations (MediaPost, May 2026). Full breakdown: Instagram's 2026 originality rules.
What TikTok has confirmed
TikTok's explainer says For You recommendations are shaped by user interactions (likes, shares, follows, comments, videos created), video information (captions, sounds, hashtags) and device and account settings, with strong signals weighted more heavily. Finishing a longer video, for example, counts for more than whether the viewer and creator are in the same country (TikTok newsroom).
The same explainer says neither follower count nor previous high-performing videos are direct factors. Big accounts still reach more people through their followers, but a small account's post can be recommended on its own merits. That is one reason competitor research should measure posts against each creator's own baseline rather than chasing the biggest accounts.
What is estimate, not fact
You will see confident numbers online: "DM shares are weighted 3 to 5 times more than likes", "aim for 7 to 15 seconds", "post at a specific hour". Some come from tracking studies, some from small samples, and none come from the platforms. Use them as hypotheses to test, not rules.
The safe reading of what the platforms have said is simpler:
- Make people keep watching.
- Make something people want to send to someone.
- Make it yours.
Turning signals into a research filter
Competitor research takes time, so spend it on the posts most likely to teach you something about these signals. A simple filter:
| Signal you can see | What it suggests | Research priority |
|---|---|---|
| High multiplier and high share ratio | The idea spread person to person | Study first |
| High multiplier, normal shares | Strong hook or format, travelled through recommendations | Study the hook and structure |
| High engagement rate, normal multiplier | Loyal audience liked it, did not spread far | Useful for followers, less for growth |
| High raw views, multiplier near 1× | Normal week for a big account | Skip |
How to calculate the pieces:
- Multiplier (outlier score): views ÷ the creator's median views on earlier posts. See how to calculate an outlier score.
- Share ratio: shares ÷ views, where the platform shows shares.
- Engagement rate: (likes + comments + shares) ÷ views.
You cannot see a competitor's watch time, so use structure as a proxy. When you study a breakout, ask: what keeps someone watching after the first second? Is there an open loop, a promised payoff, a list that is not finished yet? Our hook patterns guide covers the openings that tend to hold attention.
Applying it to your own posts
When you turn a competitor pattern into your own video, design for the same signals:
- Watch time: cut anything before the hook, deliver the promised payoff before the call to action, and keep each beat adding something.
- Sends: make the idea useful or relatable enough that someone thinks of a specific person. Practical fixes, honest tests and "this is so you" moments travel.
- Likes per reach: speak clearly to the audience you already have.
- Originality: your footage, your example, your words.
Avoid engagement bait such as "comment YES if you agree". Coverage of TikTok's 2026 changes widely reports it being discounted, and it rarely improves the signals above anyway.
A 15-minute weekly insights review
Your own insights show the signals competitors hide. Once a week, open the insights for your last five to ten Reels or TikToks and note three numbers for each:
- Average watch time (or average percentage watched). Which videos held people longest?
- Shares or sends per 1,000 views. Which videos travelled?
- Likes per 1,000 views. Which ones did your existing audience respond to?
Then compare the top performer on each column with your bottom performer and ask what differed: the hook, the length, the topic, the format? After four weeks you will have a short list of what your audience rewards, which you can set against the patterns you see breaking out for competitors.
Keep the comparison fair. A video posted yesterday has not finished collecting views, so compare posts of similar age, and judge rates rather than totals.
How ViralWatch uses these signals
ViralWatch stores each post's views, likes, comments and, where the platform provides them, shares. It calculates the multiplier against the creator's median, plus engagement rate and share ratio, and ranks breakouts by how far they beat their creator's normal. When you open one, the analysis focuses on the hook, format and structure: the parts that drive watch time and sends, and the parts you can rebuild originally.
Start with our competitor content research guide for the full weekly workflow.