Summarizing 3 Years A/R Stats

I am talking about the Anime Ratings Statistics project, of course. A project that started out as a little weekend relaxation/distraction. Nothing fancy whatsoever.
Turns out it was a miscalculation and it is now one of my biggest projects so far.

Now it's been three years since it's "birth" and it got many overhauls, environments changed and whatnot. So, it's time to draw some conclusions. Why now after three years and not after one, or two? Well, this anime season (Fall 2015) I re-coded the project entirely into a so-called MVC model. A model that strictly separates logic and content of a website. At least sort of. With that the Piwik analytics were changed within that project and started anew.

A Little Background

Previously, I pretty much created my own JavaScript Framework for this project, whereas that wasn't MVC. The JavaScript is responsible for the logic like requesting data from the server and it was inserting it into the website. In other words, it was meddling around in the content of the website and depending on context, deleted, changed DOM elements or created them.
That's not bad as is, websites worked like that for ages. It worked. Pretty well, if I may say so.

However, there came some issues with it, after a lot of improvements. For example changing the way DOM elements were inserted was incredibly tedious and not feasible after a while. This makes adding certain new features nearly impossible, same with removing (not visible) bugs.

As I said above, I thought nothing of this project in the beginning. If you are into programming, you probably know that tiny idea you get, you play around with it and discard it afterwards, because it doesn't work as you wanted or it worked, but it is not big enough to stand on its own feet / feasible to let it running. In hindsight, this is one of the rare cases where a project proves itself anyway.

Given these circumstances I never really advertised this project and I am not really fond of doing so even now for other reasons, see below. I used a link to my website in a more or less disguised signature and mentioned it a few times where people were (often) wrongly discussing the progression an anime goes through while airing by recalling false numbers and basing their arguments on that and no-one could prove them otherwise - until now.

This implies of course that I only have a very small userbase visiting this project in the first place but nontheless surprisingly a lot, showing me, with a bit more "advertising" I could reach even more. The more the website improved, the more visitors it got on average.

What I Learned

Quite a few things, but first things first. In the First Post In A New Environment post, I mentioned how I think no-one will ever read these blog posts but that I enjoy writing about my projects and will do it because of that. This is one conclusion and an important one I draw from this project:

People Are Lazy

There is an About page on the A/R Stats project, answering some important questions, like when the ratings are collected from MAL, the source behind this project. More often than not, even people actively discussing ratings in-depth, are completely oblivious as to why the ratings are collected as they are and make wrong assumptions about it.

When I started the About page I thought, this is a crucial piece of information and if people want to dig deeper, know about how things work, they'll find that page and read it. Earlier it was even prominently positioned next to the other menu items.

Wrong. People won't. People are lazy. Not all, but most of them and they rather make assumptions about what they see/already know than read up on it.
Not that I am blaming anyone here, it's simply a neutral observation that comes as a surprise to me.

Website Usage

I use Piwik analytics on this project or rather on this domain in general. I prefer it over Google Analytics since I can install Piwik on here as well and control the data independently. With that I have a pretty good overview on how the project is used. Piwik respects Do-Not-Track settings of your browser as well as doesn't save the entire IP of my visitors.
The project started in Fall 2012, at the end of November I started to use Piwik in it, the last major overhaul was released on October 1st this year (2015), so that is pretty much three years of data.


  Timespan Visitors Re-visiting Page Views Ø Time On Page
Year 1 Nov, 27th 2012 - Oct, 1st 2013 432 149 1.645 2m 36s
Year 2 Oct, 2nd 2013 - Oct, 1st 2014 3.076 1.119 10.095 2m 3s
Year 3 Oct, 2nd 2014 - Oct, 1st 2015 10.026 3.824 35.256 2m 15s

Pretty steep increase in visitors. Again, I don't really promote this project. It started out with 1 visitor per day on average to like 27 now and there's no reason to doubt it will stop here. That's in comparison with "professional" sites nothing, yes, but given the circumstances that I didn't think much of this project at first, it's somewhat great. Try to imagine all the re-visiting users of this last year, all 3.824 in one place.
Visitors alone is each hit of a visitor who is not recognized as coming back. So the number of re-visiting is telling you a bit more here.

The re-visitors increased from 34-36-38% while overall visitors increased by 712%-325%. This is noteworthy as well for the sole reason that no matter your improvements and in the last year there were several major ones, the revisiting number only changes slowly in comparison to the overall amount of visitors but that number halved in percentage.
This hints at some possibilities:

  • More people delete their cookies on a regular basis and are seen as a new visitor next time they visit
  • Perhaps slowly hitting an invisible wall by not advertising "enough"
  • Maybe the majority of people visits only once and forgets about the site, I read a few times already "what was that one site that...", "I found a site some time ago but can't find it again"
  • People don't find what they want and leave immediately / don't find anything of interest
  • Perhaps even all of the above in any combination

The time someone stays on the site stays relatively consistent. Occasionally the peaks are at 3-4mins, observable on a weekly basis, while lows are at 1 1/2m. Overall that is perhaps not much, but given my own surf behaviour, if I want to look for some stuff, I am probably not much longer on sites either (except perhaps when shopping or reading lots of text).

If I learned anything from being a moderator in a heavily chat-based browsergame, the time perception is way off for most people and I bet you won't realize how short you are on websites either. Often times when someone is violating some time-based rules, I get the answer "..but I played [insert some crappy MOBA title] for xy minutes", or "I was afk for at least xy minutes", while in truth it was like 5 minutes, no kidding!

The Average User

Technically, depending on your point of view, there is no average user. But let's put that aside and take a look at how people actually access A/R Stats. This time not divided by year since the changes are comparatively small.

Chrome Mobile6871
Mobile Safari3444
Chrome has a massive advantage here. When the project started it was almost 50/50 for Firefox and Chrome with a few percentages more for Chrome. In the last two years on the other side Chrome always was like 10-15% more than Firefox on a weekly basis. I am not really suprised, given the bloated nature of Firefox these days.

That is perhaps kind of ironic if you consider A/R Stats is programmed with Firefox in mind. I prefer Firefox due to the addons and there are a few that Chrome doesn't have, although the number of addons for Chrome increased rapidly, so did my usage of Chrome as well.

But to see that Firefox has like a 1 in 4 share is surprising. Gotta focus more on bug testing in Chrome in the future. 9% overall use A/R Stats on mobile. You know, I am very reluctant developing with mobile in mind. That's because of the very nature of this project. Graphs on mobile screens are meh, the majority uses desktop anyway and performance is another huge topic.
Developing for PCs is pretty similar to developing for mobile but at the same time quite different. This leads to several concerns to keep in mind and address. Something I should keep an eye on. The re-code with Bootstrap is pretty much an acknowledgment of that.

Speaking of mobile. How about screen resolution? The table shows the Top 5 and the next higher resolution after 1080p. This table is among the most interesting to look at. Over all three years I have roughly 1320 entries of *different* resolutions. 1320! That's insane. However, even though mobile is on the rise, the desktop typical resolution of 1920x1080p is the winner here. Behind that the laptop typical resolution. Laptop could be considered mobile but a lot of people don't have a "full-blown" PC anymore. I've included the 2880x1800 resolution just to show how small the number of above 1080p currently is, 2880x1800 is the next higher resolution I could find. Currently, though, 1080p is reigning by a few points before laptops.

Getting Constructive Feedback

Like pretty much in any project, feedback is important. Feedback helps improve things that one thought would be great in theory but fail in practice. Feedback helps getting rid of bugs that your own testing didn't find.

It's like in exams, at the end you go over what you've written to fix grammar errors and such and when you get your results back, you discover you still overlooked several errors while you exactly remember going over that specific sentence. A particular usability expert says, as a creator you know too much about what you create to find all the issues. You may find some but not all. Something I experienced several times as well.

Being a creator of something puts you in a different position here. You know, there are sometimes things you've got to read up first for days, maybe do research for weeks, work on for several days, make some compromises. There were features I worked countless hours on but weren't well-received. There were other features no-one really notices that I put weeks of research into and never will get any kind of feedback on. Time that will never be appreciated. To some extend that is understandable though, most of the people just don't know.

Given the above something I discovered is, that many people don't pay attention to what's not working and don't care for the why. They want it working and no excuses, some may be forgiving, many aren't.
If something doesn't work they start to rant about it off-site, in the context they discovered it, sometimes in very destructive ways. One of the early versions was criticised as "what's this daanst excel shit".
I'd be lying, if I say that didn't sting a lot and that comment is some years old itself but still in my head (as we all know negativity stays longer in your head). Sure, being a developer and not a designer I was aware that the design is/was lackluster but the function was and is still there. I also put quite some time into it and am open to suggestions. Confusingly, there never was anything like excel going on in the first place..
But on the other hand, the feedback I get on-site is much more friendly and constructive. In numbers I got only a few mails but each and every mail is invaluable for several reasons:

  • someone took the time to give feedback on something they care about, even if little
  • they give you insight in how they use the site or want to use it
  • they may give you pointers on things that don't work as intended, even if not explicitly mentioned
  • giving new ideas
  • most obvious: report bugs / issues / problems

That may have to do with the way I implemented the mail address. On click it opens your mail program with a pre-inserted sentence in the body: "Please be constructive and thanks for your time! :)"
In usability it's often times small things like that making a great change. I am sure it's the same with digital communication as well.

In the current newer version there is a form to send me a message and that one is limited to 500 characters to prevent abuse. In there, I don't have anything like that but already two messages through it. Seems like "people are lazy" kicks in again, eh? *winkwink*
But in all seriousness, these are constructive, too. But somewhat colder in general anyway. I am very interested to see how this will turn out in the future.

My assumption here is, given the on-site nature of the contact and the signs of a human being behind this project, appreciating that you take your time with feedback makes people realize exactly that and makes them write nicer. Of course you've got to keep in mind that many people who make destructive comments are too scared to do it to your "face".
But overall the direct feedback was overwhelmingly positive. Perhaps when people think they are ignored they get upset and their tone changes out of frustration. Something that can happen when the userbase/feedback greatly exceeds the capabilities of the receiving end.

Observable for example in the MAL forums, no-one has a clue who the new developers are, how to get in contact with them. The on-site administration seems to ignore most of the user complains. Similar for the stuff on getsatisfaction while they shutdown the on-site forum, most likely to prevent a shitstorm they knew would come. Or in the comments of Youtubers, when people leaving insults and get a response they often times apologize right away, maybe including an attempt at justification. This is probably an entirely different issue, a certain form of assumed entitlement of getting a response that ties in with the above-mentioned appreciation that someone took the time.

However, that's one reason I prefer not to advertise as much. It's off-site where people can be destructive as much as they want and I can't control it in the sense that I rely on others (moderators) and it's much more indirect. On here, it's my turf, my rules so to speak. You send me a hateful mail, I block you. Simple.
Having experienced this side of the medal I learned to appreciate the effort others put in their project much more, though.

The Future of A/R Stats

I am intrigued how this project will evolve further. How the "userbase" will de-/increase (hopefully the latter) and how the feedback will be. There are still some great ideas in the "Maybe-List" that I have to look at and maybe some great ideas come from users. Let's see how long this project will be able to run.
Maybe another three years for A/R Stats?