Cancel

Jeff Chandler

The WordPress Theme Directory Replaces Download Counts With the Number of Active Installs

Earlier this year, the WordPress plugin directory was redesigned. As part of the redesign, download counts were replaced with the number of Active Installs to reflect more accurate data. The WordPress theme directory has finally followed suit by replacing download counts with the number of Active Installs.

Active Installs of Twenty Fifteen
Active Installs of Twenty Fifteen

As you can see from the above screenshot, the Twenty Fifteen default theme included in WordPress 4.1, 4.2, and 4.3, is active on more than one million sites. Active themes are those that are activated and in use on a site. Themes that are installed and not activated are not counted, neither are child themes.

In the Themereview Slack Channel, Tom Usborne explains why active installs for child themes should be counted.

I think an argument for child themes to be included in the active installs count can be made. For example, we offer a completely blank child theme for our customers so they can make CSS and PHP adjustments. This means our install count isn’t accurate on w.org, even though those people are using the theme actively.

Dion Hulse, WordPress lead developer, agrees that child themes should be counted but the team doesn’t have the data yet. Some theme authors are concerned that new themes will have a tough time making it on the popular themes page.

Originally, the popular themes page was determined by the number of downloads over the previous week, which led to some authors to try to game the system. Hulse says, “The actual comparison between active installs and the previous week’s downloads were very similar, except for a handful of themes that had a lot more downloads than installs.”

Hulse plans to experiment with the algorithms to give newer themes a chance, “I’m also looking at ranking popular themes based on the age of the theme and installs, which will help promote some of the newer themes,” he said.

Thanks to active install counts for themes, we can see which default WordPress theme is the most popular.

  • Twenty Fifteen 1+ Million
  • Twenty Fourteen 800K+
  • Twenty Twelve 500K+
  • Twenty Eleven 500K+
  • Twenty Ten 300K+
  • Twenty Thirteen 300K+

Download counts are a terrible way to determine a theme or plugin’s popularity which is why I support this change. It’s more accurate and helps to further level the playing field for authors. Are you a fan of the change and if you’re a theme author, what other stats would you like to see?

Category:
Tags: , ,
52

52 responses to “The WordPress Theme Directory Replaces Download Counts With the Number of Active Installs”

  1. I do like the active installs info being displayed, but it would also be nice to keep the total downloads as well. I keep (kept) a download tally of themes for my site visitors, but now becomes obsolete…unless I change things up to be the active installs of my free themes + premium theme purchases. Still, having the total downloads + active install info visible helps give one an idea of how many are using the theme out of the total downloads.

    For the end-user, they can probably judge the quality of the theme based on active installs and compare it with the themes age. Perhaps seeing a download count can add to that determination too.

    As for the popular list, this is something that could be of huge concern for many theme authors.

  2. Regarding child themes, an argument could be made for excluding them from the number of active installs if you consider that some child themes are just the same parent theme but with a different color scheme. However, this is by no means the general case, and with this system child themes are penalized as a result.

    I have a Twenty Fifteen child theme on the .org directory, which gets completely ignored by this system. My child theme has kept the layout structure of the parent theme, but has substantially changed both the look and feel and Customizer options with respect to the parent. Also, it provides support for a number of plugins on the WP.org repository. Nonetheless, the parent is being considered as being actively installed on the websites where my child theme is also active. I guess other child theme developers find themselves in a similar situation.

    • To Clarify, my original comments & thoughts were that both the stylesheet (child theme) and template (parent) would be counted. So if you’re running a TwentyFifteen child theme, it’d count +1 as an active install for both TwentyFifteen and for the child theme.
      My reasoning for that is that they’re both active and providing some form of customisation to the appearance of the site.

      To take it from another angle, If you’re running a TwentyFifteen child theme purely to change a colour scheme right now, that’s not counted as an active install of TwentyFifteen & only the Child theme gets a +1 for Active Installs – that seems unfair to the parent theme to me, but is the decision I had to make to deploy it now rather than later. Iterations will come in due time.

        • Hmm, I’m thinking the child should just be counted as a +1 for the parent in most situations.

          For, example, I’m using X Theme by Theme.co on most of my sites (and more coming soon). They strongly encourage the use of a child them. I’m not sure what percentage of their install base actually implements them, but they make it *so* easy, it’s probably a lot.

          So, if I’m reading this correctly, they actually get penalized for users who do the right thing and install a child-them, as it gets counted separately from the parent?

          • I wonder if I’m reading something wrong here. If I download a theme, install it, run it via a child theme which all users are encouraged to do, then the active install count for the THEME (the parent, obviously), should be 1 total. Correct?

            Or are people thinking that the Parent would be 1 and the child would count as another 1, so 2 total?

            Or is the child being treated as a completely different theme altogether? 0 count for parent, 1 for child?

            charles como • ceo
            underground network
            +1. xxx.xxx.xxxx • los angeles <https://underground.net/

      • Sorry, I misunderstood how the system works. I can see your point about not counting the parent, therefore I also think, like Jeff, that counting both parent and child looks like the best option.

      • I think it would be best if you could have a “child theme” active installs count as well as the “active installs” count.

        Otherwise it gives themes which were built with the intention of being used as a parent theme, a huge advantage over others.

  4. It is not a surprise, but it does emphasize the importance of having good default themes as those are the only themes many users will use (lol including me as I am just so lazy :) ).

  6. The themes bundled with WordPress should be axed from the popular list. Perhaps themes from Automattic should get their own list.

    I like the active installs count much better than the number of downloads count. I’ve spent countless hours over the last 9+ years trying out themes that didn’t match what I’d expect by their placement on the list.

  7. Download counts are actually not a terrible way to do the popular list if those counts are done on a time frame. Previously, they were counted by a single week. This meant that during any given week a new or old theme would have a legitimate shot of moving up the ranks.

    You might fall the next week, but at least you had those glorious few days where you were in the top. For a lot of theme authors, that’s a damn good feeling.

    It wasn’t a completely level playing field (it’ll never be) because popular themes are still shown in the popular list. Being on that list gives your theme an advantage, making it easier to stay there. And, there were definitely people who would try to game that system (I’ve seen theme “updates” of popular themes that just had a readme.txt file change).

    However, any given week a new theme could potentially take a spot.

    I’m not sold on using active installs to handle the popular list because it means themes that were already popular will stay popular. A theme with 1,000 installs today doesn’t have a shot of competing with a theme that has 500,000 installs.

    With all that said, I’d like to discuss other methods of showing off themes. There’s some really, really good themes in the repository that simply don’t get exposure. I’d love to get those themes out there a bit.

    While we’re on the subject, I’d also love to separate things out like:

    * TRT picks.
    * Most downloads (week/month/year).
    * Most installs (week/month/year).
    * Highest rated.

    And, I’d like to knock all the “Twenty” themes off the current popular list. In a few years, there won’t be anything but those themes there.

    • I wanted to further clarify that I don’t necessarily think that downloads are the best measurement. I’d just like for us to lay some different ideas on the table.

    • Previously, they were counted by a single week

      Actually, previously, they were counted by downloads over the last 3 days. The default themes still topped the rest on that score too.

      • Perhaps we could come up with an algorithm which made the most active themes show up, but also the most downloaded over a short period, and the ones with biggest increase in active users.

        This might it easier for new themes to burst into the most popular list, whilst still allowing the most popular themes overall, to stay in the list.

      • Perhaps we could come up with an algorithm which made the most active themes show up, but also the most downloaded over a short period, and the ones with biggest increase in active users.

        This might it easier for new themes to burst into the most popular list, whilst still allowing the most popular themes overall, to stay in the list.

        We could also factor in age too perhaps. So newer themes get some preference over older ones.

        • I’m starting to see some great ideas about all this within the comments, but a factor of age should be in there somewhere in relation to active installs. Algorithms… gotta love them!

    • Some good ideas yes, and I would love to see the default themes removed. With regards to download totals, and correct me if I’m wrong, but updates were counted as “downloads”. This would of course inflate download totals.

      I still think the total download count should be put back in, but definitely keep the Active Installs. I remember a couple years back I was always curious how many themes were actually being used when you see the download counts. If the download total gets back in, one thing I would like to see is that theme updates are not counted as a download if the theme is being updated.

  8. This is a good development. I hope that plugin and theme authors will be given more stats to help them understand their businesses in the future.

  9. Here’s a question for anyone who knows the answer…what is the plus + range after the base number? 2000+

    2000 within what range?
    2000 to 2999 for example, is a significant range. Is this by 500’s or 1000’s?

    • Andre — This is not an official answer, but I spot-checked a bunch of themes in the 2,000-4,000 range, and found no 500s.

      Browsing “popular” themes…
      D5 CORPORATE LITE is 3,000+
      The very next theme down is Voyage with 2,000+

      Enigma has 30,000+
      Minamaze has 20,000+ (next theme down)

      So it appears that themes 10,000 and under are rounded to 1,000s, while themes over 10,000 active installs are counted in increments of 10,000.

      I would guess the same is true for the hundreds of thousands, though I didn’t check.

  10. I’m definitely in favor of including TRT picks. The TRT gets up-close-and-personal with every theme. They see things that the average user will never see (or understand). And I’d love to see theme developers who spend time on these details get some extra love.

    I’m also curious how often the “Feature Filter” gets used. I’d love to see the “Subject” taxonomy expanded to include more than just:

    * Holiday (39 themes)
    * Photoblogging (166)
    * Seasonal (19)

    Including every industry would be overkill, but at least an option for ‘business’ vs. ‘personal/blog’ would be nice.

    With Automattic’s purchase of WooThemes, and the continued rise of ecommerce sites, I think it’d also be helpful to see WooCommerce support, or even ecommerce support in any capacity. (BuddyPress is already listed, but not bbPress??)

    If there was a separate “Plugin Support” taxonomy, and it listed themes that support styles for some of the top plugins, would that encourage theme developers to stop including so much functionality in their themes?

  11. The concept is pretty good – the reality is not so good. What it tells you is that many, many users and webmasters either (a) turned on and are now ignoring a site, or (b) perhaps don’t grasp the concept of downloading or adding themes to their sites.

    Perhaps it would be a more useful metric of “active themes on sites that have updated in the last 180 days” or “active themes on sites with more than 10 posts, as an example, to eliminate all of the “hello world” sites out there.

      • Since most sites (pretty much all) get at least one visit from Googlebot, BingBot, or any one of thousands of other bots, it’s likely that pretty much every site in the universe that is accessible by the public will be hit and will make an update.

        Remember, unless you tick the box to keep a blog from being announced to the world, you will always be known by the search bots, and they do tend to show up from time to time.

        So yes, any site that is turned on is likely to report. Perhaps not every week, but often enough to skew the results. Perhaps it would be more useful to consider additional items (like say number of posts, or posts in the last 365 days) to consider if a site with the default themes is in fact a valid site to count. After all, a 2 year old site with 2012, and a single “welcome to wordpress” post isn’t exactly an active install, even if a bot does hit the page every week.

  12. People tend to leave sites full of old themes they tried once and never use.

    Isn’t possible to make it at least numerically recognizable, whether the update request comes from the active theme, rather than simply from an installed theme?

    • As far as I know, the stats are based on installed and activated themes. So themes that are installed but not activated don’t count.

    • The count says “Active Installs” and that’s exactly what it means. We do see all installs in the update requests, but that also sends whether the theme is the active one or not.

      The count it is displaying now is just the count where that theme is the currently active theme.

    • Why not?

      I also suspect a huge number of them are hacked up versions of the default themes. I often see situations where people have hacked one of the default themes to pieces, but it’s never dawned on them to change the slug. In that instance, the theme would presumably show up as one of the defaults, even though much of the code has changed.

  14. I’m not sure if I’m reading this correctly, if you are running a child theme of a main theme (I currently run Automattic’s Gazette Theme) but of course I run a child theme version of it, because nobody should ever run the Parent theme if they ever make any changes.

    Does this mean that my Gazette theme is not being counted as an active theme?

    Charles

    • Very good question. I’m quite sure the parent theme would be considered the active theme… but then again I can’t find a solid answer. Anyone?

    • Total download count is also valuable information when coupled with # of active themes because if the active # is much smaller than the downloads it tells you that a lot of people downloaded the theme tried it out and it’s not good.

  16. Great idea… love it for both the theme directory and the plugin directory. Whenever I search for plugins, it’s confusing when a certain plugin has had more downloads than its “competitors” (those that claim to do the same or similar functions) but it hasn’t been updated in several months or a year. You had no way of knowing how many people were still actually using the product, versus how many had since abandoned it for something better. There have been plenty of times I’ve downloaded a theme or plugin, but almost immediately didn’t like it, and thus, I got rid of it pretty quickly… so I can’t help but wonder how many others did the same thing. Active installs is definitely more helpful.

    WP should definitely credit child themes toward the parent theme’s count. A child theme might simply be a matter of two or three lines of code to preserve certain tweaks from being lost when the parent theme is updated. The parent theme shouldn’t lose credit. If the child theme’s style.css has to declare the parent theme’s folder, it should not be too difficult for WordPress’s theme directory to determine the parent and give credit accordingly.

  17. I don’t feel comfortable WordPress/Automattic/etc…knowing what themes/plugins I have active. I was never asked if I was ok with it.

    Not saying I don’t trust WordPress

  18. Just to chime in on the parent/child thang…

    I’ve done maybe 50 WP sites (including staging sites – so is that a +2?) and not a single one is up without a child theme.

    So pointless to count my child themes as it would only amount to 25 (or 50?) individual active installs. Much more sense to count just the parent themes (maybe just five that I’ve worked with.

    But what happens about the bare bones framework themes? Many of the themes here are children of these. But, I suppose, the vote should go to the ‘published’ theme, which would be covered anyway. And downloads of the frameworks will no longer skew the results, so that is better :)

    OK, ramble over,,,

  19. Thanks for the mention in the article!

    Great ideas above. I still think child themes should count as a +1 to the parent theme – hopefully there’s some progress there soon.

You may also like

Newsletter

Subscribe Via Email

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

%d bloggers like this: