Everything We Learned on Maximizing Engagement from 554,492 Instagram Posts

You’re probably doing it wrong — by Edgar Fortunati

Published in

Startup Grind

5 min readNov 9, 2016

This article describes our huge quest to find the Holy Grail of Instagram: the optimal posting time to maximize engagement and shares

If there is a best time to post, how should we find it? Which data do we need to available to make our analysis? Here are all the numbers we crunched — and the long sought after answer immediately below it.

Crunching the Data

We’ve analyzed 554,492 posts on Instagram, all posted in 2016 by a group of 7,855 Instagram accounts. For each post we’ve been given the engagement (that is, likes + comments), filter used, the exact date and time (GMT) it was posted, the caption, tags and type of post.

We do not know the exact geographic distribution of the accounts, but we do see that all continents are represented. 11.8% of all posts contain location information.

Average post distribution per hour (GMT)

Is that enough to uncover the great mystery of posting time? We wish it were that easy. Unfortunately, we can’t find out the exact time each account’s followers are viewing or engaging with each individual post.

We can, however, see how many likes and comments each post has and we can see its exact posting time. Good start.

Total engagement per hour

Based on this information — looking at the graph below, showing the sum of all engagements per hour of posting — we see a similar pattern.

Does this mean that the best time to post would be around 19:00 GMT?

Let’s look closer.

The sum of engagement on each point in time could be calculated by…

Total number of posts x Average engagement per post

Some fundamentals: the total sum of engagements for any hour is proportional to the number of posts during this hour. Therefore, it is not necessarily correct to say that posting around 19:00 would give my post a higher engagement than posting around 9:00 GMT, for example. We should compensate for the fact that many postings were done around 19:00.

Furthermore, we should exclude the effect of each user’s location and timezone. Let’s go deeper down the rabbit hole.

Standardize

Our next step was to analyze the standardized* engagement (likes + comments) for each individual post.

Why standardize? If we both want to compare the engagement of my posts (I have 46 followers and I live in The Netherlands) with those of Selena Gomez (101 million followers and I assume she lives in USA), and we would like to find out what’s the relative best time to post a photo or video on Instagram across each of our (Selena’s and mine) audiences, we need to standardize all of our posts.

* Standardization of engagement: per user we’ve calculated the average number of engagements and the standard deviation. To obtain the standardized number we subtracted the average from each observation and divided the result by the standard deviation:
Standardised Engagement = (Engagement per post — average engagement per user) / (user’s engagement standard deviation)
With standardized data, the mean for each individual user is zero and its standard deviation is one. As a result, Selena’s engagement and mine can be compared — apples to apples!

Distribution of standardized engagements

The graph above shows the distribution of the standardized engagement for 550,201 posts, after we removed 4,291 (0,77%) outliers from the dataset.

For the statisticians among us..
Excluding the outliers, all observations are within the interval between -3 and 3.5.
Even though each individual’s average engagement is zero, we see that the overall average standardized engagement is -0.036. The overall standard deviation of our standardized engagement observations is 0,889 and — as the figure shows — the distribution has some skewness amounting to 0.936.

So… what’s the best time to post?

Let’s have a look at the relative performance of average standardized posting times during the day. We considered all standardized observations and plotted the average engagement per hour of posting. This graph is telling a different story.

Average standardized engagement per post per hour of posting

As compared to the absolute engagement numbers, the average standard engagement per posting time shows an inverse picture.

Whereas 19:00 (GMT) is an optimal time in absolute numbers, after standardizing the engagement numbers we see that between 4:00 and 6:00 (GMT) seems to be a an optimal time interval.

In San Francisco & Los Angeles, that’s 8:00 PM — 11:00 PM PST
In New York City & Boston, that’s 11:00 PM — 2:00 AM EST
In London, that’s 4:00 AM — 7:00 AM GMT

Does this make any sense?

One could say that at the optimal posting time based on standardized engagement, competition is lower. After all, the absolute number of posts around 5:00 (GMT) is lower, so each individual post has a higher probability of being visible in a user’s feed.

If you think our analysis doesn’t make any sense or we should take a different approach, feel free to comment and discuss!

Summary

To find the best time to post on Instagram to get maximal visibility and engagement we performed an analysis on over 550,000 Instagram posts, all posted in 2016.

The most posting activity is around 18:00 (GMT) and posts from around 19:00 (GMT) generate the highest absolute numbers of engagement. We should however make use of standardization techniques to correct for differences between account profiles, thus making all posts comparable.

After standardizing the engagement per post: