WordPress has long had download counts for core WordPress, plugins, and themes.
But downloads counts are deceiving. They count downloads, but are not representative of actual active installs.
Recently — in my Club member newsletter — I noted that it is past time to get data for actual installs, versus download counts. There is a private beta program on WordPress.org that offers just that, and at least some of the new stats are launching very, very soon (edit: the initial changes are now live).
The new plugin stats pages will show four new charts, visible to plugin authors:
New installs per day
Updates per day
Active installs per day
Active versions
Most importantly, the “active installs per day” chart shows us — with much greater precision than we previously had — how many actual websites are running any plugin available on WordPress.org.
For the beta period, utilizing a non-public query string parameter, I was able to see these stats publicly for any plugin on the repo. That query string parameter has now changed, so the data is no longer publicly visible.
These stats are partially launching very shortly, to highlight the active installs for plugins as well as the breakdown of active versions. Active install counts will replace the download count in the right column.
You can see an example of what the new public stats from the WordPress Importer plugin look like, which will be visible to users. Plugin authors will see a slightly different view that contains the additional stats:
Click to view full size
The most popular WordPress plugins
I took some time to look through the most popular WordPress plugins I could think of, as well as other notable plugins. What follows are some of the numbers I recorded, which were noted in the charts as being recorded February 24th.
Once this data goes live, the counts will be revised to round with less detail, and for installs over 1 million, will simply say “1 Million+”.
Plugin
Downloads
Active Installs
Contact Form 7
24.304 million
6.3+ million
Akismet
28.128 million
6.2+ million
WordPress SEO
15.928 million
3.7+ million
Jetpack
14.587 million
3.7+ million
WordPress Importer
9.635 million
3.4+ million
Google Sitemap Generator
16.477 million
3.4+ million
All In One SEO
21.853 million
3.1+ million
WP Super Cache
6.869 million
1.7+ million
Tiny MCE Advanced
4.788 million
1.4+ million
NextGen Gallery
11.991 million
1.3+ million
Google Analytics for WordPress
8.159 million
1.2+ million
WooCommerce
6.445 million
1.2+ million
Hello Dolly
510,136
1.1+ million
WP Page Navi
5.432 million
1.1+ million
W3 Total Cache
4.491 million
970,000+
WordFence
4.828 million
710,000+
Better WP Security
4.259 million
600,000+
WPTouch
6.743 million
410,000+
bbPress
1.741 million
250,000+
MailPoet
3.447 million
230,000+
The Events Calendar
1.438 million
230,000+
BuddyPress
2.647 million
150,000+
WP eCommerce
3.009 million
80,000+
As you can see, the data is incredibly insightful, especially compared to the public download stats. However, beware that there is not a guarantee of accuracy here, and in fact you should know that in some instances, the team knows the data isn’t reporting properly, especially for plugins with more installs.
Contact Form 7 and Akismet are by far the most popular WordPress plugins, with over six million active installs each. I included most of the plugins from the Popular page on the repo, plus some additional notable plugins.
There are also a number of outliers, where the discrepancy between downloads and active installs is huge. I believe this is largely due to plugins with long histories that were once immensely popular but have since faded in popularity.
The download count keeps old plugins relevant far longer than if active installs were the prominent number.
Compared to BuiltWith data
How accurate is this data? Well, one smell-test we can give it is to compare it to other potentially inaccurate data. From my research, BuiltWith is doing one of the best jobs around of accurately checking site data. Go ahead and input your domain to BuiltWith and see how accurate it is for you.
So, if we compare this same plugin list to BuiltWith data for “the entire Internet” — as they segment based on site rankings — then we can get a feel for how well the WordPress.org number and theBuiltWith number get along.
In this table, I link to the BuiltWith page so you can get more information if you like. Do note that some plugins aren’t available on BuiltWith, so are in the list but don’t have information.
Plugin
BuiltWith
Active Installs
Contact Form 7
594,677
6.3+ million
Akismet
166,791
6.2+ million
WordPress SEO
2.602 million
3.7+ million
Jetpack
1.629 million
3.7+ million
WordPress Importer
n/a
3.4+ million
Google Sitemap Generator
n/a
3.4+ million
All In One SEO
n/a
3.1+ million
WP Super Cache
991,729
1.7+ million
Tiny MCE Advanced
n/a
1.4+ million
NextGen Gallery
977,343
1.3+ million
Google Analytics for WordPress
871,144
1.2+ million
WooCommerce
685,937
1.2+ million
Hello Dolly
n/a
1.1+ million
WP Page Navi
585,809
1.1+ million
W3 Total Cache
571,864
970,000+
WordFence
n/a
710,000+
Better WP Security
n/a
600,000+
WPTouch (Pro Data only)
15,696
410,000+
bbPress
126,081
250,000+
MailPoet
149,018
230,000+
The Events Calendar
56,095
230,000+
BuddyPress
76,853
150,000+
WP eCommerce
58,115
80,000+
Data Accuracy
There are zero instances from this list where BuiltWith shows more sites than WordPress.org using a plugin. That is a confidence enhancing note, as WordPress.org should be much more fully comprehensive.
There are some instances, especially with plugins that are on the backend, or have limited front-end visibility, where the numbers don’t line up too well.
However, for the highly visible plugins, the WordPress.org data feels good to me, in comparison to BuiltWith.
The team working on the new stats has less confidence in the numbers as the numbers get higher. So for plugins with well beyond a million installs, they have reason to believe adoption may be different than what is reflected in the stats.
How stats are collected
WordPress.org collects this data via the update checks that a WordPress install performs.
There are a couple of caveats to consider for the WordPress.org data:
The numbers could include (and therefore inflate numbers) non-public (dev and staging environments) domains. According to Andrew Nacin, the team does exclude sites with “localhost”, “.dev”, “.local” and “other telltale signs of local development and staging sites, like IP addresses.”
The numbers don’t include websites that have, via code, turned off update checks.
Some data is making more sense as it is collected. Now that it’s a few months old, they are able to make better determinations about what is accurate and what is not.
Keep in mind the percentage of installs that wouldn’t be included in these checks is minimal. Also, unfortunately, a large percentage of folks just don’t keep dev or local installs of their websites, so those numbers probably do not inflate the totals as much as you may think either — though I’m guessing on that front.
For the accuracy of the numbers, the team has seen some discrepancies, where one graph shows data that should also impact another, but it does not. They’re working to resolve these issues still.
This data is important
The team working on this project has been deciding on which stats pages to showcase for the general public, and which to show to plugin authors.
The active install data is important, and it needs to be public for users, in addition to plugin authors. If we can ensure that the data is even remotely accurate, it is far superior to using download counts alone.
Download counts give legacy plugins too much clout in the repo, and also make misinformation all the more likely.
A prime example is that of Slimstats, which had the recent security breach. That plugin has over 1.3 million downloads. But when I looked at the stats, there were only 100,000 or so active installs.
Yet, when tech blogs covered the security breach of the plugin, they largely cited “over a million” installs at risk. This is bad for WordPress and it’s unnecessary. We can fix it. We have the data.
Lead developer Dion Hulse also had this to say about the progress of the stats pages, when I notified him that I had access to these numbers:
The current beta stats page was put together as a trial to find what data plugin authors wanted (We’ve been in touch with a few who have provided great feedback in the past on the stats).
Based on what they have learned, they are launching the active install numbers and active version chart, and will follow up with a future launch for the plugin author only stats. The plugin author stats are going to go live with the notice that they may not be totally accurate, and they may change over time as the team continues to analyze the data.
WordPress is open source, and has absolutely massive adoption across the web, as we all know. It is therefore a source for what’s become a multi-billion dollar industry. WordPress.org and those that hold the keys to the website have great power.
Changes to the website and how themes, plugins, and any other potentially business-centric data make a difference to real businesses.
All those businesses have signed up for this (being involved with the WordPress project), and therefore any risks associated, but in my mind this data — especially of active installs — brings welcome sunlight to a murky plugin landscape.
I’m glad I was able to see this data. I’m very pleased that the active install numbers are replacing the download counts. WordPress itself will be better for it.