Improvements Wishlist for Magnatune

Posted by Konstantinos Christidis on March 07, 2005

I like Magnatune a lot; the honest business model, the diversity of genres, and the occasionally interesting music.

It could use a couple of improvements (most of them minor) when it comes to how the site works and looks, though.

For example, the site’s all about selling music. How do you know what music to buy? Unless you’re a fan of randomness and serendipity, logic says you’ll want to see some sales/downloads charts and some reviews.

Charts and Stats

Magnatune offers plenty of stats of your pleasure (best-selling by genre this month, and more), but how you get to those pages? The only links I could find where on the bottom of the artists‘ and genres‘ pages. And with the artists’ page being quite long, your best to find it is via the genres’ page.

So, suggested improvement #1: more prominent placement of charts — why not on top of the artists/genres listing? And why not a direct link from the top of each page on the site, or from the left sidebar?

Customer Reviews

Charts are very good, but they’re not a proper, all-around indicator of the album’s identity. Sure the album’s probably good because it sells a lot, but why is it good? What’s its strong point? Do only a few tracks standout, or is it good as a whole? Is it good because of the great lyrics, the inspiring music, the captivating choruses, the unresistable beats, or you-name-it? (The information on each artist’s page doesn’t cut it, and it’s probably not meant for this either.)

If you’re looking for such information before getting an album, Magnatune offers you the chance to stream it in its entirety, which is undoubtedly a great option.

But it’s an option only for those with fast connections; and there’s lots of us living with slo-o-ow connections who still want to be a part of Magnatune. Listening in low-fi is a somewhat viable option, but you can’t count on it to draw conclusions on some aspects of the music (such as the quality of the production, the various layers, etc.).

Also, listening to a stream takes time (equal to the album’s duration ~40-60 minutes). And this is not time where music acts as a background activity; if you wish to draw conclusions on the music, you have to focus on it, making the listening session a foreground activity.

So while streaming rocks, it has a few weak points; keep that point, I’ll be back at it, in a second.

Another way to find out more information about the good albums, is by visiting the Magnatune Forums. Good luck discovering the threads talking about good picks; you’re on your own when it comes to finding them, and when you come to think of it, it’s just too much trouble for something that should be so simple.

Conclusion (and suggested improvement #2): let customers add their reviews on the albums’ pages, Amazon-style. This is a vast improvement over the searching-the-forums-for-good-picks method, and a nice compliment to the streaming (perfect for those who don’t want to spend 60 minutes to listen to the full stream).

Why do listeners think this is a recording worthy of my time? What’s so special about it? That’s the special insight customer reviews give, and that’s what’s missing from the picture in Magnatune.

And it’s information I can skim in a few minutes, closing the gap in the preview-buy window.

[Note: only let those who have paid (i.e. not licenced it) to review it; you value/judge something far more seriously when you've paid for it.]

Minor Rants

Mostly a personal wish, but I’m sure someone out there shares it too: the grey-text-on-white-background on the site’s pages (example) should go. #666666? What’s wrong with dear ol’ black (#000000)? A bit darker, but insanely more readable and makes me feel like I’m not sitting in front a bleached screen.

And while we’re at it. A quick look at the source code showed bad signs of HTML 4-itis (probably because John Buckman, Magnatune’s founder, isn’t primarily a designer?):

<FONT FACE=”Verdana, Arial, utopia, sans-serif” SIZE=”2″ COLOR=”#666666″>

This is obviously trivial since the site’s about music, and the redundant markups in the code don’t add remarkably to the pages’ loading time, but still; such a hip idea deserves to be XHTML/CSS (and it’ll raise John’s geek quotient by at least 10 points).

In Closing

John is a smart guy and I’m sure he’s either working on some of my suggestions as we speak, or he has dismissed some of them for reasons I haven’t thought of yet (I’m not claiming these are false reasons; just that they haven’t crossed my mind yet).

For the slim chance he has missed some of my suggestions, I hope he considers them and makes Magnatune even better.

4 Comments to Improvements Wishlist for Magnatune

John Buckman
Jul 06, 2005

Hey, thanks for the suggestions re: Magnatune, very timely!

Per your thoughts:

- RE: top charts — yes, I have a brand new layout for the artist pages almost ready to go, with where each album is in the charts prominently in the left hand column, as in “this album is 4th this week in the Jazz charts, 12th overall in all genres” so you can easily explore the charts. Also, the “also sounds like” links will be longer, with 20 recommendations.
- I’m adding a wiki to Magnatune, so that users can not only add comments, but go nuts with concert photos, comments, related ideas, etc…, as well as bands being able to chime in.
- HTML color 666666 — that’s probably caused by me using a really nice, bright LCD monitor on my Mac, which are notoriously bright and beautiful. I’ll give black text on white a try.
- I don’t like CSS because I can do all I visually need to do with HTML 4 and it works on all browsers — my geek code is to use the oldest, most stable technologies I can get away with. Yes, that means Magnatune is a no Java zone. I actually test against Netscape 3, and everything I do works on every browser, from Opera, to IE3, Safari, Firefox, etc. I have yet to see a CSS based web site that works everywhere. My blog uses CSS, and I hired a CSS guru to make it work, and it still does strange little things on some browsers (Safari, for instance, wraps one of the navigation buttons)

More suggestions on “how to find music I’d want to buy” are VERY much desired, so fire away with ideas…

Konstantinos Christidis
Jul 09, 2005

John, you’re welcome, and thank you for taking the time to expand on my points. I’m quite happy to hear that some of my recommendations are going to be realized, in one form or another.

As for your asking for suggestions on “how to find music I’d want to buy”, here’s another one: it’s CD Baby’s “”Sound Like“” search feature, but with a twist.

From their website: “Name a famous artist, and we’ll find you some artists we think you’ll like!”.

This helps folks who come from mainstream-land to easily find a nice landing spot in indie-land. (I know because I’m one of them.)

As for the twist, let me say first that, of course, it’s up to the artists themselves to name who they sound like, and that sounds fair.

But that breaks things a bit — look for CD Baby artists who “sound like” your favorite artist. Sample the first 20 results, and tell me if you came up with more than 5 artists who beared even the smallest similarity to your favorite mainstream artist.

What I’m trying to say is: I guess some times the artists exaggerate, or simply don’t see things the way the customers do, thus breaking the potential of the “Sound Like” feature.

So, the twist is to let the users decide on whether an artist in Magnatune indeed sounds like the mainstream artist he claims to. When the specific connection is voted badly and reaches a certain point (e.g. less than 50% of the voters think there is one), it gets dropped.

Two points I’d like to make:

1. Unlike the case of “customer reviews” (read my note in the corresponding section in my post above), where I believe you should only let paying customers review the albums (you value/judge something far more seriously when you’ve paid for it), anyone should be able to vote on the artist connections. If I bought it, chances are that I believe in the connection, so voting results from paying customers will be probably predisposed towards endorsing the connections. And you’re letting out all those who were about to buy it, but rejected it after sampling it and failing to see the claimed connection. So, all users get to vote on the artist connections.

2. Allow the Magnatune artist to claim a set number of mainstream artist connections. I believe that 5 is a fair number, though I may be wrong. It should definitely be larger than 1 (well, duh), but not large enough because the efficiency of the “Sound Like” feature will decrease. Effective targeting is the goal here.

To sum up:

- The artist gets to make up to 5 connections to mainstream artists.
- These connections are visible in all of his albums pages (and his bio/profile). Have radio buttons (”yes, I agree”, “no, I don’t think so”) next to each claimed connection and a “submit” button at the end. Don’t force the user to vote necessarily on all connections, just the ones he has an opinion about.
- The users vote on the connections as they see fit. Wait till you get 500 votes on a connection, and then start rating it; if it’s approved by less than 50%, drop it. Also, the connection that gets the most approvals gets moved to the top in the connections list (i.e. order them, apply the data in as many ways you can). Note: the “500 votes” threshold should exist for the obvious reason: assuming there isn’t, if, say, two of the first three raters disapprove the connection, it’ll get dropped since it’s approval percentage is smaller than 50%. But that’s unfair since the number of votes isn’t big enough. (Of course, the number of votes that makes up the threshold is up to you.)

And something else: when a connection is dropped, a slot in the artist’s “5 mainstream connections to claim” form opens. Let him submit another one. (Or let the users, suggest possible connections.)

Before ending this seemingly endless comment, I’d like to make two points against it!

1. Isn’t it kinda of belittling for your artists to have to define themselves through mainstream artists? Just trying to think it from the artist’s view, I may be exagerrating here.

2. Obviously, the approach I’m suggesting hands the control to the user and kinda takes it away from the artists. Are you sure you want to do that? I, speaking from the customer’s side, am in favor of doing this. But the artists may see it differently: “What if the lame users don’t think I sound like ‘X’ mainstream artist? I know better, and I think the connection stands.” So, think about that too.

John Buckman
Jul 09, 2005

I like the idea, but I’d be more inclined to let users define what bands an artist sounds like. I’ve found artists to be *way* off in who they think they sound like, and usually name major influences, rather than other bands they actually compete with (no band wants to say they are a “sound-alike”, they’re too brilliant and original for that).

Kind of related to this idea, is one I am already set on doing, which is “make me a playlist of things that sound like this song”, so that when you find one song you like, Magnatune will make a mix of songs that are similar, and you can let them play for a while. There’s some academic research going on in Spain that I’m cooperating with, that can do this, and is eerily accurate at finding things with a similar mood.

I’m gonna check out CDBABY’s implementation, to see how Derek did it, as I generally admire his work. The big thing people like about Magnatune is the web site’s simplicity, so if/when I do your idea, I’d want to make it really simple to use.

Konstantinos Christidis
Jul 09, 2005

Great. I’d be interested to see how you implement the “sounds like” feature if/when you decide to do so.

Also? That second paragraph of yours has got me excited.