I have raised this in this forum, in feedback documents, and in direct conversations with Adobe (at least) since 2023. Every time, the answer was some version of “thanks, we’ll review this internally” or “there is currently nobody available to work on this.” Nothing has changed. So here it is again, with today’s screenshots.
Rank 1 under “Most popular” has an average rating of 1.1 out of 5
Pantone Connect. 5,404 ratings. 1.1 stars. Position one. The very first thing every user sees when they open the “All plugins” tab in the Marketplace.
This is what happens when “popular” means nothing but a cumulative install counter, a number that only ever goes up, that no amount of user dissatisfaction can move, and that is self-reinforcing, because whatever sits at the top keeps collecting installs precisely for sitting at the top. Five thousand people took the time to tell Adobe this listing did not work for them, and the ranking system did not register a single one of them.
Think about what that does to a user opening the store for the first time. They install the #1 recommendation, it disappoints them, and the conclusion they draw is not “that one plugin was bad.” It is “the Adobe plugin store is junk.” Every one of us publishing there pays for that impression.
“Top rated” is not a quality ranking either
Sorting by Top rated uses the raw star average and completely ignores how many people voted. One single 5-star rating beats twenty 5-star ratings. It beats three hundred ratings averaging 4.9.
So the “best” plugins on the platform are, by construction, the ones almost nobody has rated. A developer who spent years earning hundreds of genuine reviews ranks below a listing with one vote from a friend (or from himself). The one sorting option that is supposed to reward quality is the one that systematically hides it.
And the cards do not even show the rating count. A user looking at five stars cannot tell whether that means five people or five hundred without clicking into the listing.
This is not a hard problem
That is what makes it so frustrating. This is not a redesign, not a re-architecture, not a roadmap epic. It is two sort keys.
Top rated – a Bayesian weighted average instead of the raw mean:
score = (v / (v + m)) * R + (m / (v + m)) * C
v = number of ratings for this listing
R = average rating for this listing
m = minimum-votes constant (e.g. 20)
C = mean rating across all listings
Listings with few votes get pulled toward the platform average and can no longer win on a single rating. Listings with a real track record keep their own average. IMDb has done it this way for over twenty years. A Wilson lower bound works just as well. Either one is an afternoon of work.
Most popular – factor in the average rating, the number of ratings and the date of the last update, instead of raw installs alone. At the absolute minimum, stop ranking a 1.1-star listing first.
Both views – print the rating count next to the stars.
Who this is addressed to
To be clear: I don’t think the Adobe people reading this forum disagree with me. My impression over the past three years is that you know all of this already, and that the blocker sits above you – nobody is being given the time to touch it.
So consider this post something you can point at. If it helps to have the problem stated bluntly and in public by a developer who has been building on this platform for a decade, please use it.
Adobe built all of this. Why not give it a little love?
Adobe invested a great deal into this platform: the marketplace itself, the review pipeline, the developer distribution portal, the UXP architecture, the pre-release program, the developer relations team, the events. An enormous amount of work went into getting developers to build here.
And then the two sort keys that decide what a user actually sees – the last three metres of that entire pipeline – have been left untouched for years, to the point where the store’s top recommendation is a listing rated 1.1 out of 5. All of that investment is being undercut by something that would take one developer one afternoon.
For whoever has to justify the headcount: this is not a developer-fairness request. The store’s number-one recommendation currently has an average rating of 1.1 out of 5. That is the first impression of the Creative Cloud plugin ecosystem for every new user who opens that tab. Whatever the Marketplace’s conversion and retention numbers look like, the entry point is actively working against them, and the fix is a sort key, not a project.
I understand that teams are stretched and priorities are set elsewhere. But “no resources” has now been the answer for three years running, for a change of this size. At some point that stops being a resourcing issue and starts being a decision.
Is this still being maintained?
Is the Plugins Marketplace still being maintained at all, or has it been effectively abandoned?
I would much rather be told there is hope, and that small fixes like this one can still land.
To the other developers reading this
If you have seen the same thing, please reply, with your listing, your ranking position, your numbers. One person complaining for three years is easy to file away. Fifteen developers in one thread is something that can be taken into a planning meeting. That is what would actually move this.
Previous discussion, same issue, March 2023: How the Plugins Marketplace could be improved – Detailed feedback (section 2.3).
.