WordPress Relaunches Plugin Directory with New Design and Improved Search Algorithm

The new WordPress Plugin Directory went live today. Contributors have been working for the past year on rebuilding the directory with a new design and better capabilities for searching through the 45K+ available plugins.

The new landing page displays a search bar prominently at the top, followed by sections for featured plugins, popular plugins, and beta plugins.

The design changes to the individual plugin pages have received mixed feedback from members of Advanced WordPress Facebook. The screenshots slider and FAQ accordion UI are welcome improvements, but many are not fond of how the new single plugin display ditches the previous tabbed interface in favor of truncated sections with numerous “read more” links. This requires the visitor to click and scroll repeatedly through a massive wall of text. The experience of doing this to find information is much less efficient than the previous tabbed UI.

Many of the plugin header images appear to be stretched now with the new wider content area, but this may encourage developers to update their listings. Several users have commented that they are disappointed with the removal of stats, which are now viewable for admins only. Plugin team member Mika Epstein said they are waiting for more feedback from end users (not plugin developers) on whether to make them publicly viewable again.

The new design matches WordPress.org’s homepage redesign, which went from sketches to prototype to live during the last few weeks of December 2016. The most exciting improvements, however, are found under the hood.

The plugin directory is now powered by WordPress, instead of bbPress, making it easier for plugin developers and reviewers to manage plugins, tags/categories, and committers. The submission and review process has also been streamlined so the plugin team will no longer reject plugins after 7 days. They also have to ability to rename plugin slugs prior to approval for easier handling of typos.

New Plugin Search Algorithm Delivers More Relevant Results

The most significant user-facing improvement in the redesign is better search relevance. Analysis of plugin search logs in August 2016 showed the directory handles 100,000 searches per day and receives more than 500k unique search queries per month.

Greg Ichneumon Brown, a Data Wrangler at Automattic, published an in-depth article on how he and contributors improved search fidelity for the new plugin directory. The eye-opening account explains just how opinionated search algorithms need to be in order to give users the best experience at this scale.

“In improving the fidelity of search results, it’s not just a question of how we satisfy a single user’s search query, but how we satisfy thousands of users for each unique search term: which plugins will support that volume of users and their requests for support?” Brown said. “Which are most likely to give all of these users a great WordPress experience?”

The new search algorithm delivers results based on this premise. Brown added a calculation for the lowest/highest ratio of active_installs to the number of times the search was performed. However, this isn’t the only factor that influences whether or not a users will have a good experience with the top results. The algorithm also includes signals that Brown said plugin authors have a lot more control over:

  • Resolving support threads
  • Keeping the plugin update to date
  • Testing the plugin on the latest versions of WordPress

Brown’s post breaks down the Elasticsearch query he developed for the directory and explains how text matching works to find relevant results for author searches, queries that include operators, and partial-word boosting.

The plugin directory relaunch has several known issues that the plugin team noted in the announcement today. Users are noticing many broken links, problems with SVN, and other bugs when navigating around the new directory. The team has a list of tickets on Meta Trac that they are working through and feedback is welcome. Plugin submissions are temporarily disabled while the transition is completed.


69 responses to “WordPress Relaunches Plugin Directory with New Design and Improved Search Algorithm”

  1. Love the clean, new look! The SERP pagination also feels a bit snappier. I’m guessing that maybe the design decisions were mobile-driven, hence the scrolling. I wish there were some anchor links up top to skip to the sections down below just for quicker access, but otherwise it’s a nice-looking upgrade =)

    • I agree with this, and I think the new search algorithm will give a much better experience for users.

      One idea I have is that for the plugins with a video in the readme, perhaps that video could be embedded in the screenshots slider? E.g. if the readme parser saw a link to youtube/vimeo under == Screenshots == it would show a video instead of an image from /assets.

  2. I hate to say this but, at least as far as usability and layout are concerned, the previous version was far better. And this one doesn’t even display the “last updated” field which was one of the first criteria I was paying attention to. Sad.

  3. In my opinion this new outfit of the Plugin Directory is more than one step backwards. And for many, me included, it is in an example of constantly ignoring feedback from plugin developers and users alike.

    The one pager design is user-unfriendly as one has to click the numerous “Read more” links/areas — which are way to small to capture.

    Especially for long plugin descriptions – that could be necessary for various reasons – the one page design is worse than before.

    Also direct/deep linking to various sections like before (changelog, FAQ etc.) is no longer possible.

    One improvement though, is the flipping of screenshots in the little gallery. However, screenshots still open directly the file, no lightbox etc. So it’s permanently clicking forth and back…

    I remember, Matt once saying to the new directory (when in beta), the team “could do better” — sadly, this statement is still true.

    Overall it’s a great disappointment.

    • What bugs me is that everyone is now complaining. My only question, not just to you @David, but to anyone lodging complaints here is “Did you click that link shown on every page under the old directory for the last ~6 months and submit user feedback on all these issues?”

      If you didn’t then you shouldn’t really be complaining now. I personally have been testing and giving feedback for almost a year since I first saw it as a WIP. But they asked for public feedback for half a year. People are so quick now to say its terrible but there was a process.

      As for the feedback I provided, some was taken into consideration, others added to future lists, and some discarded, but they listened none the less.

      • Yeah, well people should complain if something is crappy. The idea that everyone was responsible for this crap is, well I guess empowering, but frankly people are paid to do this as a job. One would think that a successful company (at least financially, and with good PR), would have people who could do a better job.

        Asking for feedback, ignoring half of it, and then complaining that there was not enough input from the community, lame excuse for incompetence.

      • Daniel, I get you!
        However, I gave feedback on various channels early on, Meta, Twitter, Blogs etc., took the survey(s) etc. One feedback/bug report that was totally solved, was the search functionality that was non-working turing some time period in the beta/preview version. For the other parts criticized I didn’t noticed any general changes in the last weeks and months. And I checked regularly, on a weekly, even daily basis.

        At one time however, I did gave up to give more feedback, as I felt – like many others – that they decided to not change the general direction any more. So why give more feedback, if nothing won’t change, as it seems decided long time before?

        Were I “only” a user myself, I may have arranged with this new “thing” somehow. Well, I still have lot of plugins as a developer/author in the directory, therefore I am directly effected and not only dislike the changes but even worse, see those new outfit as not helpful for users/developers alike in range of UI and UX. And it seems I am not the only one.

        It feels a bit like in the past with WordPress 2.5+2.6 admin redesign: it was not good, or a “fail”. With WP 2.7, 2.8 and onwards this was “corrected” and then it fitted again with the MAJORITY of the userbase. I hope for some such corrections here also. I still have some hope! :)

  4. And yet still , unless I am missing something, when I get search results
    I cannot sort them by user rating, number of installs, number of actual raters etc.
    Why isn’t this basic functionality still not in the directory?

    • @Patrick – Its coming. They spent so much time on the new search algorithms they wanted to get them public and scaled before adding more layers of complexity. Once elasticsearch kinks have been smoothed out adding sorting by any of the weighted criteria will be cake.

      • Thank you for your reply, Daniel. It just seems barmy that basic sorting of search results has taken so long, I don’t know how possible or not it would have been in the old version.
        eg in the new, a search for “shortcodes” currently provides 715 pages of results. No one is going to look at more than the first few pages and there is no way to sort eg via popularity of number of installs or star rating/users etc. This is long overdue.

        • @Patrick – Maybe, but if you search for just “shortcodes” you are asking for a lot of results. IE probably 25% of the plugins in the repo add shortcodes of some kind and a good portion of those likely mention it somewhere in their readmes.

          Now if you added a modifier keyword like “widget shortcodes” you cut that in half and the top results are most likely to solve your needs. If not adding additional keywords will further narrow.

          IE the search wasn’t wrong in returning 700+ results, your search just wasn’t specific enough to return what you really wanted.

          If you simply wanted a lot of useful shortcodes in a single package then “shortcode pack” or similar.

          Also of note I believe the +”keyword” modifier works, but -“keyword” does not.

    • @Peter – Yea I want this too. I have seen talk of it in the #meta slack channels, so its not off the table, but someone has to code it, and unfortunately only so many people volunteering to do stuff, most with their own agenda (specific functionalities they want to contribute to), so maybe this is something you would like to contribute? Its all open source and they love people to submit patches :)

  5. I have mixed feelings about the redesign. The improved search is great. The crucial social proof is prominently placed. It’s visually appealing. And the new description setup — short blurb followed by “read more” — is workable, but requires developers to rethink how they present their plugins.

    On the other hand, the reduced information density means a lot more scrolling, and the first page of results feels more “exclusive” now. (Kinda like being on the first page of Google SERPs.)

    Hopefully we’ll see some quick iterations in the coming weeks.

  6. New algorithm doesn’t seem to factor time into the ranking and focusing heavily on the active installs.
    It should factor the time it took the plugin to accumulate those numbers or use a time decay upto a certain time, else new plugins will never be found by the users, unless the plugin developers put a lot of effort and time in marketing them.

    ” Rich gets richer, and the poor get poorer :( “

    • That’s an interesting idea.

      I don’t completely agree for a couple of reasons:
      – The (primary) goal is for end users to have a great user experience and find plugins that do a great job at solving their particular problem.
      – When an end user installs a plugin it will probably be on their site for years. We should think about what their experience will be a year from now.
      – New plugins are great and important to the ecosystem, but marketing them is not the goal of the plugin repo.
      – A new plugin has no history of support and no indication that it will still exist in a year.
      – A plugin that has built up thousands or millions of users indicates that the developer(s) has been working, contributing, and supporting it for years. That is something that should be encouraged rather than discouraged.

      There probably is a higher barrier to getting a new plugin noticed though. Which I agree isn’t ideal, but for a new user we should use all the data we have to give them the best WordPress experience possible and by definition we just don’t have much data on completely new plugins.

      I experimented a little bit with things like boosting plugins that are authored by developers who have other plugins with lots of other installs (basically a sum of a developers active installs), but wasn’t very happy with the results at the time. I think something like this could work to counteract the problems for new plugins while also improving the search experience overall.

      It is an interesting idea to also try using the rate of change in installs, but it would be tricky to balance against other parameters. Also, a plugin with 10k installs for years that recently updated and suddenly has gone to 100k would be an interesting thing to boost as well.

  7. I’m not happy with the new design because users can not see the screenshots at a glance and compare easily . Plugin’s banner has been blurry . User could get specific info for example FAQ, Changelog easily clicking on the specific Tabs in the previous version .

  8. https://wordpress.org/plugins/search/Easy+AdSense+Ads+%26+Scripts+Manager/

    A 35 character exact match search doesn’t even put my plugin in first page, let alone the top 3. Also note the irrelevant plugins in the results.

    Search is advertised as the major improvement, it’s the most broken one. Even if I spend a huge amount marketing my plugin, it will still go nowhere because most users use search to install the plugin ultimately.

    If this is how it is gonna be, I have to take my work elsewhere :(
    Disappointing !!

    • Satish: Your plugin is structured incorrectly.


      You don’t have a readme.txt file in /trunk, and your readme.md is in the wrong format.

      You do have a readme.txt file in /assets, but that’s not the right place for it. Move it to /trunk, and correct the Stable Tag: in it to “trunk” as well, since you’re not using the /tags directory.

      Remove /trunk/assets/ entirely, since that’s the wrong place for it, and anyway you have those files in /assets correctly.

      Basically, GIGO applies. You need to get your plugin in proper order for the directory to be able to see and index it correctly.

      You can always email the plugins team if you’re uncertain why your plugin is behaving incorrectly. Your plugin is not showing up properly because of these, not because of “the new search”. It wouldn’t have shown up very well on the old search either.

      • Thanks for the clarification Otto.
        Earlier the plugin used to show up in top 10 for search term “adsense”, and I did not make any changes between the switch to new plugin repository. (I may have accidentally removed readme.txt from trunk.)

        So I thought it was the update that broke things. Will make the necessary changes and report back.

      • This is more bureaucratic nonsense. Obviously the old search engine algorithm worked well. Blame the developer for not having something formatted properly, when the old version worked fine, is not a great response, unless somewhere in that algorithm are penalties for malformatting or something like that?

        It is the removal of *exact match* that has broken search in a fundamental way.

  9. First, I will say that the new design is cleaner and workable for mobile, so congrats on those who did the design work.

    Now for second impressions and my preliminary exploring of this new directory…

    1. Featured list gets 4 items displayed, but if you click See All, it adds just two more items which seems kind of useless.

    2. How does the Popular list get populated…is it based on active installs? If so, this means the older plugins that have been around longest and high download/active installs will always be there. New plugins that could be better will never get a fair chance.

    3. I would like to see a categorized list instead of having to now use just the search. People may not know what keywords to enter in, especially for people who are new to WordPress and even to running their own website. Keep the search, but provide a categorized list or select list. I tried different keyword searches and had items show up that don’t even relate. Example: Widget Publishing….I get plugins like Facebook Feed, Events Calendar, Photo Gallery, Akismet….seriously? Now, if you change the keywords to “Show Widgets”, then you get some relevant results. But it still means that many people who many not know how to get accurate enough for finding a plugin with the right keywords or keyword structure. Just something to think about.

    4. A one-page concept is actually bad for users. Considering what used to be tabs, are now accordions and sections within the same page, a user could be facing a mile-long page…which I’ve already tested this and I had a super long page. Long for a desktop monitor, but imagine this on a phone or small tablet. Each section is not easily identifiable either; where one ends and another begins because each is provided by a heading only.

    5. The screenshots slider is too integrated into the page without anything to make it standout more. White or light backgrounds of screenshots and thumbnails are hard to see. Perhaps encase the slider in a darker container to have it standout more as a slider. Also, there’s no way to escape the window if one clicks on the slide image.

    6. Text links for See All, Read More, and even Demo links are easily missed and should stand out more…perhaps as buttons.

    I think it’s a good start, but end-user usability is more complicated IMO. A few more tweaks and adjustments are needed here.

    • Hi Georgio,

      Can you describe what you are trying to find with “widget publishing” and “show widgets” and what a good result for those searches would be?

      That would make it easier to understand how the search might be improved. Looking at it and looking at the results, I really don’t know what a good result is. Almost 30% of all plugins mention having a widget.

  10. Please bring back the Stats page for individual plugins.

    Also, if possible, it would be nice if there was a way to switch to the old tabbed layout or something similar, it’s a better UX for many users as you don’t have to mindlessly scroll up and down between sections.

    One positive point though, is that clicking on a screenshot no longer downloads it. Yay! :)

  11. Mobile is great, but this new design looks like it was designed for an Ipad as the biggest form factor. That is simply not accurate with regards to visitor window size statistics. Two column fixed width, lots of space, and 14 results per page? Please, that is just asking for more requests which will hammer the site harder. Search results can be cached for the top searches, so there is no reason not to deliver more results, for a better usability/user interface/user experience.

    The pagination at the bottom is equally ludicrous. If someone knows their result is on page 10, that takes 10 actual clicks to get there.

  12. Congrats to the team that worked on this on the launch! I know it was a long time in the making, and that it is a somewhat thankless job.

    Just to add feedback for the record:

    The Good
    – Overall design. Pretty, clean, and simple, and looks good on mobile.
    – Search. Search algorithm is a dramatic improvement. Prior version was basically broken, so it’s great to see that resolved.

    Constructive feedback
    – Stats. As with some others here, I’d still love to see stats brought back
    – Individual Plugin Page Design. As with some others here, I’ve never been a huge fan of of the single-page accordion design for individual plugin pages. Just too much info to try to condense it into a single page. I’d still rather see some kind of tabbed interface.
    – Sorting. As with some others here, I’d agree some sorting functionality would be great. But sounds like that’s being discussed, and I’m fine with iterative improvements.

    This is all in line with feedback I’ve offered on past betas and public discussions.

    All in all, I’m excited to have this out. Thanks again.

  13. I have to agree, new wordpress developers are being killed by this new alghoritm…
    Even if I understand that a plugin with 5k installation is something that the visitor must notice, there are obviously more recent plugins that are doing a better job (better UI for example, same functionalities). You run the risk to drive new WP users to “old” plugins, while the modern ones will stay in the shadow forever, because this is what the new search does.
    If you search for the EXACT name of some plugins and/or the EXACT list of their tags, they end up in 5th page or even worse. Are we sure this makes much sense?

    • I’m not sure that sends the right message though either. I mean do you really want to punish developers who have not only put 1000’s of hours into developing popular plugins, but the countless more hours spent supporting it for free.

      Based on your view, why bother to try and build plugins as a business at all, when the new guys can just come in and take what you spent all that time building. Its actually backward logic when you consider the commercial nature of the WordPress ecosystem as a whole. IE You want plugins to have long term support, you want developers to make enough to maintain and grow plugins, but you don’t want them to show up in search results?

      That said I do think they could put some effort into building tools similar to managewp.org such as “Up & Coming”, “Highest Rated”, “Newest” etc and still allow you to use keywords & tags inside those searches as well. They could be doing more with the ample data they have in terms of interconnecting things as well, such as being able to show related plugins or “other users who also looked at”.

      There is room for improvement, but I assume you don’t really want to push LTS out the door for shiny fly by night plugins by authors who may or may not continue development & support them.

      Just for reference, as you may notice from my obvious bias, I am a full time plugin dev who makes 100% of my income and several of my support teams incomes from plugins on the repo. Before the new search we had spent 2 years earning ~600 5 star reviews (mainly from excellent support) and 90k installs. We have kept our teams bills paid and such, but were not really doing great. In the old search we barely ranked at #18 for our main keyword, how is that fair? Now obviously we are #1 in everything and I have definitely seen a noticeable bump, feels like vindication in some sense, that we grinded it out for so long, hoping to be able to hire a few more people by years end if growth continues. So yea I’m a bit biased, and understand your points, I mean at one point (2 years ago) my plugin had no installs and no popularity, It was in that group of new plugins you are talking about. Pure drive & grinding it out is why we are now ranked so high, I imagine the same is true for any of the other plugin authors with popular plugins now.

      Look forward to see how it plays out either way. Love that this many people are voicing their opinions as well in a constructive way. Kinda nice to see.


Subscribe Via Email

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

%d bloggers like this: