WordPress 4.1 To Introduce Plugin Recommendations

On the Add New plugin page in WordPress 4.0, there are two different tabs to help users discover new plugins: Featured and Popular. What if WordPress could recommend plugins based on the ones you have installed? Thanks to a recent update to the API, users will be able to access a list of recommended plugins in WordPress 4.1.

Plugins Recommended To Me Based on Data of Sites with Similar Plugins Installed
Plugins Recommended To Me Based on Data of Sites with Similar Plugins Installed

The recommendations take into account the plugins you have installed and suggests plugins based on which ones are commonly used together. It’s similar to an e-commerce shopping cart that recommends products other customers have purchased based on what’s in the cart.

Recommended Plugins To Get its Own Tab

In today’s WordPress 4.1 developer meeting, the group reached a consensus to create a new tab for Recommended plugins instead of replacing the Popular tab. I’m in favor of this decision because they each have a different definition. Also, the Popular tab shows which plugins are doing well across the directory. Another benefit to keeping the Popular plugins tab is for fresh installs of WordPress, as it gives users a good foundation to start looking for plugins.

In my tests, I noticed that some of the plugins recommended to me are 5-8 years old. I’ve reported the behavior to Alex Shiels and it will be addressed before 4.1 ships. If you are testing WordPress 4.1 and come across any oddities with plugin recommendations, please report them in the ticket.

The Recommended plugins tab exposes new and useful plugins to users instead of limiting them to the Featured and Popular categories. Since recommendations are based on actual data, they should get better as time goes on. Do you think a recommended plugins page is a useful addition to the plugin discovery process? Let us know in the comments.


15 responses to “WordPress 4.1 To Introduce Plugin Recommendations”

  1. Recommended plugins is based on a good algorithm could be great, but I’m not a fan of adding another menu item for it in the Dashboard menu vs. simply incorporating it on the Add New page.

    Now there will be Featured Plugings, Popular Plugings and Recommended Plugins? Seems like 2 too many IMO.

    • Agreed. As well as that list, there’s also ‘bundled’ – you get Akismet with your WP download.

      I think both the featured and bundled plugin(s) should just be dropped – what is the featured list really of? These two mechanisms push paid Automattic products (e.g. Akismet, VaultPress via JetPack), and presumably is commercially valuable for them. Do they pay the WordPress foundation for that advertising space? I think that some transparency about how the list is put together would be valuable.

  2. I see a risk for new plugins being less recognized when people only use the recommended plugins. When people are looking for a plugin now, they search the web or the plugin repo and might even bother reading the description of a plugin that is new and give it a chance. I see many great plugins rise, but if people only react on specific recommendations we might cut something important.

  3. It seems to me that as soon as the powers that be start highlighting certain plugins over others, there are going to be serious concerns about bias. Popularity is a user driven metric. “Featured” and “Recommended” are not really user driven. They’re driven by the preferences of a small few – either in the sense that plugins are chosen directly, or in the sense that those who write any algorithm decide which factors should be included and weighted.

    If they do roll out a recommended plugins page, I think it would be critical that the algorithm be absolutely transparent. Everyone should know clearly why certain plugins are being recommended over others. And, of course, as soon as they do that, there’s likely to be controversy and debate what factors are included and how they’re weighted.

    Honestly, I’d rather see them deploy a “filter” interface that allows users to filter plugins based on certain criteria like downloads, ratings, forum resolved rate.

    For example, if I could select any social sharing plugins with at least X downloads, X out of 5.0 on ratings, and a forum resolved rate of at least X%, then I’d be looking at plugins that are popular, well rated, and well supported.

    You could come up with a lot of other useful filtering criteria like that. And that approach seems much more valuable to me than some black box spitting out a list of plugin without knowing why.

    • Most core development “resources” are volunteers giving their time to the features they’re interested in improving. If more people are interested in contributing (not just with code) to a particular area, improvements can be made there. The post editor experience is getting another big improvement in 4.1, after a lot of attention in 4.0, fyi.

  4. Ok, I’ve updated the post with a screenshot of what the Recommended plugins tab looks like and the plugins it recommended to me. I believe the recommendations will get better over time as more data is acquired on similar groups of plugins installed on other WordPress sites.

    • I know, as Nick noted above, that contributors put a lot of work and heart into this stuff, but I still just don’t really understand this idea. If we do want to improve the plugin interface, It seems like there are so many better ways to help people find plugins.

      The main challenge right now is that there are *so many* plugins to sift through for any given type of functionality, and there is no efficient way to do so, either via the dashboard plugin interface or on WordPress.org. I think that’s the real crux of the problem. To solve it, you need to give users the power to filter plugins based on the criteria they care about.

      Instead, this mostly offers up a random collection of plugins, that may or may not be relevant to what the user wants/needs, and which is based on criteria/biases the algorithm developers – rather than any given user – care about.

      Does anyone have idea what specific criteria are being used to generate these lists?

  5. Hey Jeff – Thanks for checking that. That’s what I was figuring. But I guess I’d just say that someone is still deciding which data factors should be included in the algorithm, and how they should be weighted. In another words, the bias is just worked permanently into the code.

    That’s one reason I think it’s critical the algorithm be very transparent.

  6. By the way, apart from bias, there’s also just the question of how useful this is.

    For example, in the thread you posted above, nacin mentioned the “Customers Who Bought This Item Also Bought” approach from Amazon.

    But I don’t I think plugins really work the same way. If get a book on javascript from Amazon, I’m obviously interested in javascript and related programming languages. So you can show books on that subject and it makes sense. The fact that I install, say, XYZ SEO plugin doesn’t really indicate that I want to see other SEO plugins, which are redundant, or that I want to see other plugins by the same author, which may be totally unrelated, etc. It doesn’t really tell you much at all about what plugins I might want in the future.

    I think for me, this approach just doesn’t reflect how people look for plugins. I suspect (I could be wrong) that most people start out wanting specific functionality and they go to the plugins page and search for it. They don’t usually go looking for a list of random but potentially interesting plugins they might try out.


Subscribe Via Email

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

%d bloggers like this: