The WordPress Theme Directory No Longer Counts Automated Downloads

The popular themes page on the WordPress theme directory usually displays a mix of default themes and those that meet a certain criteria. While the criteria to reach the page is not publicly known, download counts are one form of measurement.

Since September, the Swift theme has routinely shown up on the popular themes page. According to download stats, Swift was downloaded between, 2,000 to 3,500 times a day in September and early October.

Swift Download Count
Swift Download Count

However, the stats drop off abruptly on October 15th and have yet to reach previous levels. On October 18th, a few members inside the Theme Review Team Slack channel shared suspicions that something was not right with Swift’s download stats.

Ionut Neagu discovered that Swift does not appear on the GoDaddy Hot 100 while other themes on the popular page do. While not conclusive, the evidence points towards some sort of anomaly.

Samuel ‘Otto’ Wood, who helps maintain WordPress.org looked at the raw data for Swift and discovered that a lot of the downloads appeared to be automated.

The swift theme is indeed getting an exceptional number of downloads, so I took a closer look at the raw data. It appears that a lot of these downloads are automated in some way. However, there’s no evidence that it is malicious, nor does the pattern apply only to the swift theme. Perhaps it is included in some installer system that auto-downloads themes for newly installed sites. It’s hard to say.

In any case, I made a way to detect and block these types of automated downloads from appearing in the stats, and retroactively applied it for the last three days of data (which is all we retain). Going forward, automated downloads will not appear as downloads and will not affect the popular listings.

Once the changes were implemented, Swift’s download stats significantly decreased. The changes also apply to all other themes hosted on the directory. Otto declined to comment on the specific changes made but tells the Tavern, “If we can detect a case where downloads aren’t being made by users, then we will try not to count it in the results.”

Swift’s Side of the Story

Satish Gandham, founder of Swiftthemes, explains what Swift has gone through the last few weeks which may account for the high download numbers, “Swift was featured for a couple of days and we spent money on Facebook ads to promote the free version. Also, there were a couple of hosted WordPress providers who asked permission to include Swift on their platform.

“The premium version of Swift also had the same slug URL as the free version. The WordPress theme directory saw v7.2.48 to be higher than 7.2.6 and premium users downgraded to the free version thinking it was an upgrade. It was a support nightmare,” Gandham tells the Tavern.

Swift’s support forums confirm that some customers of the commercial version were automatically updated to the free version in early August.

Download Counts are Almost Irrelevant

When the plugin directory was redesigned earlier this year, download counts were removed in favor of active installs. Active installs are more accurate and difficult to manipulate. On the other hand, when the theme directory was redesigned, download counts remained.

It’s possible that the theme directory will one day show active installs but the data gathering aspect is not ready yet. Also, the numbers are lower for themes than for plugins, which may lead to incorrect conclusions in regards to relative popularity.

It Happens More Than You Might Think

In this particular instance, there’s no evidence that points to a malicious intent to inflate stats to manipulate the popular themes page. In the years I’ve written about WordPress, I’ve heard rumors that certain plugins and themes automate their download counts to appear more popular than they really are. However, I’ve yet to write about a theme or plugin caught in the act.

According to Otto, it happens more often than you might think. When he or other volunteer moderators for the plugin and theme directories notice fishy download counts, they contact the author and the process usually stops. If it continues, the author risks having their themes and plugins suspended or banned from the directory.

If you notice suspicious activity with a theme hosted on WordPress.org, you can voice your concern in the Theme Review Slack channel during non-meeting times. For suspicious activity related to plugins hosted on the directory, email plugins at WordPress.org.

11

11 responses to “The WordPress Theme Directory No Longer Counts Automated Downloads”

  1. Thanks for writing post, Jeff.

    We know how valuable the top spots are. Upwards of $50k a month (in pro upsells) using data some authors make available in income reports. Are people automating downloads to snag those spots? Absolutely.

    This is at least a small step in the right direction for authors that play by the rules.

  2. Its good that the counts are being noticed…one thing that has always bugged the hell out of me is the fact that “default” themes are on that list by default and takes away any chances for theme authors to reach the top spots. I don’t think the default themes should be allowed on there…it’s like WP saying it’s our list and directory, so we get to have our cake and you get the crumbs. Eventually, as more default themes are added to WP, the list will be all default themes! Perhaps a bit exaggerated, but this is how it appears.

  3. I noticed a few more who are doing similar thing.

    You can easily check the popular ones who are getting upwards of 1800 downloads per day.

    We have been featured on 1st place in 1st spot for a day sometimes and yet haven’t gone more than 1300 downloads in a day for any theme ever.

    So yes if active installs are counted then the game will surely change as we feel there are many who are still manipulating these things.

  4. We’ve had suspicions about download manipulation in the plugins directory for a few years now. We’ve welcomed the active installs metric as it definitely shows how many folks are using the plugin.

    One way I’ve seen plugin developers manipulate the download statistics was by releasing new versions every 1-3 days. Most cases at least they add something or fix a bug, but sometimes I suspect this trickle was on purpose in order to game downloads. The active installs metric allows plugins to maintain a normal release schedule without being penalized download wise. Thank you WordPress team!

    Shame on Swift theme, I would take a hard line and ban the developers for a month (or longer) from the WordPress theme/plugin directory. If cheaters knew there was a consequence then they would not cheat.

  5. For those asking, the various default themes are indeed the most popular themes. However, to answer a couple points:

    1. The “downloads” that come from updating core and thus getting those themes installed at that point in time do *not* count towards the popular list. Only downloads of the theme ZIP file from WordPress.org count.

    2. There’s no special treatment for the popular list. The various default themes do not have an automatic entry in that list. It’s based on the download numbers alone.

    Now, I suspect that simply having that large pre-installed base does indeed cause those themes to be downloaded more often, especially when updates occur. So, they’re at the top of that list because of that, but there’s no guarantees there.

    Also, as I told Jeff, I do not believe that the inflated download numbers for the Swift theme was their fault, nor does it appear to be intentional. No blame falls on the authors for the changes made, even though they were probably most impacted by it.

    Note to anybody making an automated installer system: If you’re going to pre-download a bunch of themes from WordPress.org for users, so they have a selection available to them immediately, please do us a favor and append ?nostats=1 to the end of the ZIP file download URL. Makes things nicer for everybody involved. Same goes for plugins or any other sort of downloads where the download is not actually directly user-initiated.

  6. I guess this is something that should happened a long time ago, so no one can manipulate stats and ratings. Even though I am still a bit concerned about the criteria will it be good enough to stop such tactics in future? Lets hope for further betterment and finally thanks to the WordPress community for giving solution to this problem.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Newsletter

Subscribe Via Email

Enter your email address to subscribe to this blog and receive notifications of new posts by email.

Discover more from WP Tavern

Subscribe now to keep reading and get access to the full archive.

Continue reading