Special thanks to Gav Knafo for his quantitative analysis work which made this article possible.

We reverse engineered Instagram’s algorithm to find the magic number of likes it takes to viral.

I commissioned a 3-day study that followed 31 posts, tracking likes and comments at different time intervals to see if there’s a linear correlation between engagement and views.

There is.

Here’s how we did it.

The Setup

We followed 31 different posts from popular Instagram accounts over the course of 3 days, tracking likes, comments, and views as our metrics.

The accounts we studied had follower counts ranging from 168,000 to 2.4 million.

Our goal was to evaluate the relationship between the number of likes and comments on a post and the number of views it gathered.

To get more actionable data, we broke down our independent variables (likes and comments) by time.

The results shed an interesting light on what it takes to go viral on Instagram.

The Results

First things first: likes and comments on a post trump follower count.

Likes within an hour and total number of comments were the most significant factors in predicting the virality of a post.

We also did an analysis to see if these two variables operate independently or if they compound each other.

Our analysis found that the two act independently. A post can go viral based on likes OR comments; it doesn’t necessarily need both.

Here’s where things get interesting.

A Game of Likes

Take a look at the relationship between total views and number of likes over the first hour after posting.

Likes in the first 20 minutes give us a trend for total views, but only a loose one:

There’s a rough slope to the data, but also clear outliers.

One post only received 1200 likes in the first 20 minutes, but the second highest total view count in the data set.

Another received over 5000 likes in the first 20 minutes, but didn’t get more than 200,000 views.

The picture gets a little clearer when we expand the timeline to 30 minutes.

instagram-likes-30-minutes-views

Now, we see a clearer trend line. Outliers are fewer and less extreme. Things are coming into focus.

But when we look at likes in the first hour, we get an important takeaway.

instagram-likes-1-hour-views

Here, we see almost no outliers. There’s a very clear relationship between the X and Y axes, and we can outline a clear S curve and inflection point around 8,000 likes:

instagram-likes-vs-views

In this data, the first 6,000 or so likes in an hour generate no more than 300,000 views, and it’s closer to 100,000 for most of the points in that section of the data.

But the next 4,000 likes – the difference between 6,000 and 10,000 likes in the first 60 minutes – generate a boost of close to a million additional views.

This points to the conclusion that Instagram’s algorithm has a clear inflection point around 8,000 likes in the first hour, at least for the type of accounts we tracked.

This variable is even more important than number of followers – and keep in mind, we tracked accounts with follower numbers varying by more than 2 million.

It’s also significant to note that immediate likes and comments aren’t the most significant factor. Based on our data, there’s an hour-long window during which the algorithm evaluates a post.

Open for Comment

While not as predictive, comments do matter for virality. The data suggests an inflection point for comments similar to the one we see for likes:

instagram-comments-vs-views

Interestingly, the magic number seems to be 8,000 in this case, as well.

Admittedly, our data for comments isn’t as useful as our data for likes because we didn’t have the same time intervals. Breaking down comments at 20 minutes, 30 minutes, and an hour is something we’re looking to study in the future.

So what does all this mean?

The Takeaways

1. Evaluate at 60 Minutes

The first, most obvious thing to note here is that the performance of a post in the first hour is a strong predictor of its long-term success.

If you see a post light up within 60 minutes of posting, you can be pretty sure it will go viral. If it’s still dead after 60 minutes, its chances are basically nil.

In terms of actionable strategies, this means there’s a lot to be said for deleting content.

If a post hasn’t gained traction within an hour, you’re unlikely to lose anything by trashing it.

Adopting this strategy gives you the freedom to test content without necessarily damaging your engagement metrics.

All the pros delete low-performance posts. The data says you should too.

2. Comment Baiting Can Work

The next takeaway is that likes aren’t the only way to the top.

Our analysis indicates that likes and comments function independently as variables affecting total views. A post can have one or the other; it doesn’t necessarily need both.

To that effect, including prompts for commenters in your posts can be an effective strategy.

While our data isn’t as concrete here, if you can generate between 6,000 and 8,000 comments on a post, it’s likely to garner more than 1,000,000 views.

3. It’s Not a Perfect Science

As always, the big disclaimer is there are no foolproof rules.

We still saw variability in our data set, and it could have come from a number of factors.

One potential source is crowding in the feeds of followers.

If a given account has followers who tend to follow a larger number of accounts, the individual feed of each follower will be more crowded, creating more competition for likes from that follower.

This could introduce noise into the data and cloud the pattern.

While we’re confident in our analysis, you should take it with a grain of salt (like everything you read on the internet). We’ll continue to update this post with new conclusions as we gather more data.

We strove for accuracy and a reliable experimental design, but also acknowledge that the study is neither perfect nor exhaustive. There could be other things at play.

For example, we may have found comments in the first hour to be a highly predictive variable. Likes after 60 minutes – say, during the second hour online – may be important as well.

And finally, there’s the subjective factor: content.

No matter how much we crunch the numbers, the basic fact is that content is still king.

The accounts we tracked tend to produce content their followers enjoy – but without a doubt, high-quality content is still one of the most important factors for virality.

Test for Yourself

At the end of the day, all of this data is circumstantial. What works for one account may not work for yours.

But data like this can give you guidelines for what to test and how to shape your social strategy.

Use this to inform your own Instagram strategy. But do your own testing, and continuously refine your content based on the results.

If you want to learn some battle-tested techniques for building a business outside of Instagram, I recently released a guide for non-technical founders based on my experience in app development.

It’s full of actionable strategies based on my 10+ years in the field and 250+ product launches. Check it out:


And whether you’re building a viral Instagram empire or a mobile app, keep testing and keep iterating.

That’s the strategy I used to go from 0 to 140,000 followers on Instagram and millions of views on LinkedIn.

There’s nothing to stop you from doing the same.

Author

CEO and Co-founder of Neon Roots Ben Lee is the co-founder and CEO of Neon Roots, a digital development agency with a mission to destroy the development model and rebuild it from the ground up. After a brief correspondence with Fidel Castro at age nine, Ben decided to start doing things his own way, going from busboy to club manager at a world-class nightclub before he turned 18. Since then, Ben has founded or taken a leading role in 5 businesses in everything from software development to food and entertainment.