api.video

Use Cases

Tutorials

Video Analytics: A Primer

October 12, 2020 - Doug

You have created the perfect video, uploaded it to api.video, and embedded the player on your website. Are you curious about how many people are looking at your video? How about where they are in the world?

Maybe you want to get even more granular, and know how much of the video is being watched? Are they rewinding and watching parts again? Are they skipping ahead? With api.video’s analytics, you can find out.

In this post, we’ll take a look at how to extract this information for one video, and in future posts, we will go into deeper depth into building analysis tools.

My Video

I’ve used this video of Tuba players playing the Beatles in other tutorial posts. So, we know it will have a lot of views. But, how many? Using the List Video sessions endpoint, we can get information about each time this video was played with the api.video player.

To use this API, you need 2 items: a videoId of the video you want to measure, and an authentication token. In order to authenticate your session, you need your api.video token. Read the details in the authentication tutorial, and then come back with your token (it will be valid for one hour).

In this example, we’ll use curl to access the video session endpoint. The videoId is vi4blUQJFrYWbaG44NChkH27. The curl request looks like:

curl --request GET \
  --url 'https://ws.api.video/analytics/videos/vi4blUQJFrYWbaG44NChkH27?currentPage=5&pageSize=100' --header 'authorization: Bearer {token}'   --header 'accept: application/vnd.api.video+json'

I am requesting page 5 (currentPage in the url), with pageSize=100 (100 sessions per page, since I knew that this video had between 400 and 500 views. For simplicity, I’ll show just the 2 most recent sessions:

{
	"data": [
		{
			"session": {
				"sessionId": "ps4mPLq8bZzjXlxwJG4cg4vl",
				"loadedAt": "2020-10-12T11:21:09+00:00",
				"endedAt": "2020-10-12T11:27:20+00:00"
			},
			"location": {
				"country": "United Kingdom",
				"city": "Glasgow"
			},
			"referrer": {
				"url": null,
				"medium": "unknown",
				"source": "unknown",
				"searchTerm": "unknown"
			},
			"device": {
				"type": "computer",
				"vendor": "unknown",
				"model": "Other"
			},
			"os": {
				"name": "Mac OS X",
				"shortname": "Mac OS X",
				"version": "10.15.6"
			},
			"client": {
				"type": "unknown",
				"name": "Chrome",
				"version": "85.0.4183"
			}
		},
		{
			"session": {
				"sessionId": "ps42gcSLNv8hSQxKOWWgxeNj",
				"loadedAt": "2020-10-12T11:43:56+00:00",
				"endedAt": "2020-10-12T11:44:19+00:00"
			},
			"location": {
				"country": "Netherlands",
				"city": "Uithoorn"
			},
			"referrer": {
				"url": "https://t.co",
				"medium": "social",
				"source": "Twitter",
				"searchTerm": "unknown"
			},
			"device": {
				"type": "computer",
				"vendor": "unknown",
				"model": "Other"
			},
			"os": {
				"name": "Mac OS X",
				"shortname": "Mac OS X",
				"version": "10.15.7"
			},
			"client": {
				"type": "unknown",
				"name": "Safari",
				"version": "14.0"
			}
		}
	],
	"pagination": {
		"currentPage": 5,
		"currentPageItems": 22,
		"pageSize": 100,
		"pagesTotal": 5,
		"itemsTotal": 422,
		"links": [
			{
				"rel": "self",
				"uri": "/analytics/videos/vi4blUQJFrYWbaG44NChkH27?currentPage=5&pageSize=100"
			},
			{
				"rel": "first",
				"uri": "/analytics/videos/vi4blUQJFrYWbaG44NChkH27?currentPage=1&pageSize=100"
			},
			{
				"rel": "previous",
				"uri": "/analytics/videos/vi4blUQJFrYWbaG44NChkH27?currentPage=4&pageSize=100"
			},
			{
				"rel": "last",
				"uri": "/analytics/videos/vi4blUQJFrYWbaG44NChkH27?currentPage=5&pageSize=100"
			}
		]
	}
}

Let’s start with the information at the bottom. We can get links for each page of sessions, and itemsTotal is 422 - meaning there have been 422 sessions watching this video.

Let’s look at the 2 sessions in the logs. The first session has sessionId = ps4mPLq8bZzjXlxwJG4cg4vl. This user watched the video for about 6 minutes based on the session length: "loadedAt": "2020-10-12T11:21:09+00:00", "endedAt": "2020-10-12T11:27:20+00:00"

We also know that they are in Glasgow, UK, using Chrome 85 on a Mac.. It just so happens that my ISP enters the internet in Scotland (all my advertising is Scotland based) even though my physical location is England, so this session is from my computer.

The 2nd session is interesting:

"location": {
			"country": "Netherlands",
			"city": "Uithoorn"
		},
		"referrer": {
			"url": "https://t.co",
			"medium": "social",
			"source": "Twitter",
			"searchTerm": "unknown"
		},
		

I accidentally tweeted the url to this video, and deleted the tweet within 30 seconds, Yet, someone on Twitter in the Netherlands still got to see the video before the tweet disappeared.

Did you know that the api.video player works in Twitter? Now you do! Your analytics will show video plays from any share on Twitter!

Applications

With this data - you could build a map of where the video is being watched by country, and in some cases by city (with the caeat that this is not 100% accurate - my ISP says I am in Scotland). You can also understand the referring urls: Twitter, blogs, Linkedin, etc. Finally, you can also monitor the traffic patterns by day of week, time of day, etc.

Going even deeper

It is great to know that there was a video view at 11:21 GMT - but we can learn even more about this video view. Using the session events endpoint and the sessionId, we can actually describe the exact usage of that session. Making a request using a sessionId (note this is a different - more interesting - session):

--request GET \
  --url 'https://ws.api.video/analytics/sessions/ps42gcSLNv8hSQxKOWWgxeNj/events?currentPage=1&pageSize=100' --header 'accept: application/vnd.api.video+json' --header 'authorization: Bearer {token}'  --header 'accept: application/vnd.api.video+json'

This request will return all of the video player events in that session: so we can see any play, pause or seek events that occurred in playback:

{
	"data": [
		{
			"type": "ready",
			"emittedAt": "2020-10-11T20:43:32+00:00",
			"at": 0
		},
		{
			"type": "play",
			"emittedAt": "2020-10-11T20:43:32+00:00",
			"at": 0
		},
		{
			"type": "pause",
			"emittedAt": "2020-10-11T20:43:57+00:00",
			"at": 2
		},
		{
			"type": "resume",
			"emittedAt": "2020-10-11T20:44:00+00:00",
			"at": 18
		},
		{
			"type": "seek.forward",
			"emittedAt": "2020-10-11T20:44:00+00:00",
			"from": 2,
			"to": 18
		},
		{
			"type": "pause",
			"emittedAt": "2020-10-11T20:44:01+00:00",
			"at": 19
		},
		{
			"type": "resume",
			"emittedAt": "2020-10-11T20:44:04+00:00",
			"at": 2
		},
		{
			"type": "seek.backward",
			"emittedAt": "2020-10-11T20:44:04+00:00",
			"from": 19,
			"to": 2
		},
		{
			"type": "pause",
			"emittedAt": "2020-10-11T20:44:11+00:00",
			"at": 9
		}
	],
	"pagination": {
		"currentPage": 1,
		"currentPageItems": 9,
		"pageSize": 100,
		"pagesTotal": 1,
		"itemsTotal": 9,
		"links": [
			{
				"rel": "self",
				"uri": "/analytics/sessions/psyd0zH6RWUCxP7xnWqcnKm/events?currentPage=1&pageSize=100"
			},
			{
				"rel": "first",
				"uri": "/analytics/sessions/psyd0zH6RWUCxP7xnWqcnKm/events?currentPage=1&pageSize=100"
			},
			{
				"rel": "last",
				"uri": "/analytics/sessions/psyd0zH6RWUCxP7xnWqcnKm/events?currentPage=1&pageSize=100"
			}
		]
	}
}

The video was ready to play (entry 0) and started to play (entry 1) at 2020-10-11T20:43:32+00:00. (That is expected, as the player associated with this video is set to autoplay as soon as it is ready). The video is also set to loop, which is evident in the next entry.

The third entry shows that the video was paused at 2020-10-11T20:43:57+00:00. The video timestamp shows “2” meaning 2 seconds into the video. About 25 seconds elapsed since the video started, so how does that work? The video is about 20 seconds long, meaning that the video played to the end and started again before the pause occurred.

The next entries show that there is a seek.forward event from 2->18s, and the video resumed for one second before rewinding back to 2 and then playing to 9 seconds into the video.

Aggregating these session stats across all sessions can give you an idea of when users rewind - perhaps when an important point is raised - indicating that in future videos that data should be highlighted for your users. You can also see if there is early abandonment, or jumps ahead - as users look for the content that they are interested in.

Fast analytics

These results are available as soon as the session is completed. To see how quickly, I recorded a video where I watched the video, and then retrieved the analytics immediately afterward.

Conclusion

When you use the api.video player, there are a number of interesting analytics collected. Beyond knowing the number of views, the geo-location of the views and the time that the videos were viewed, you can also dig into the analytics of each session to better understand how your videos are being played. If you have questions, please reach out in the community forum

Doug

Doug

Developer Evangelist

Get started now

Connect your users with videos