Okay, interesting answers. I didn't consider SEO conventions, but I understand your compromises. Here are a few other thoughts though...

I was more thinking about the lack of (even slightly increased) space between the filter category titles and the sub-filters themselves than between those - who seem quite fine as they are now.
But that's based on the experience of someone who works more often with mockups following very closely editorial instructions tied to printing restrictions, plus aesthetics and considerations related to litterature. You'd almost never see information presented that closely to any title, especially in Europe, even when the difference only amounts to an outline followed by the same space value (between a title and subfilters or sub-filters themselves). Not that it's profoundly dramatic of course...
Take the non-capitalized way you types your tags and labels ("first time on sale...", "price improvement", "showing 1-50", etc.). On internet, that's quite common nowadays, but even Gog - in a very european way - followed the common capitalized convention (for the first letter of every tag), since the opposite would be unthinkable in any professional written production. Even the oddities of "RoboCop" or "GameSieve" are usually explained by technical regulations : an organization's given the right to dispose of its name in any way it sees fit and this extends to commercial titles, IPs, etc. - hence the difference between names types on covers and those reported after that in magazines. French language being very subtle, less pragmatic and more conservative on that matter compared to english : Gog would be correct, since it contains pronounceable (vowel) syllabes ; GOG wouldn't for example, since this convention is usually saved for unpronounceable acronyms (like HTML).

No, I did understand their use, but I admit I was (unconsciously) considering your search engine's operation differently at the beginning - as the result of a request rather than a catalog from which elements are substracted - and am not sure I understood precisely your explanations ("their behaviour... exclude the filter (where the regular link is require), and when filters overlap (making AND-selection possible), that's something which is simply impossible with the OR-selection of multi-select checkboxed filters."). With only checkboxes (using "and" associations), no crosses or exclusions, one would simply select which parts of the catalog he wishes to display, reducing the initial view to a simple list of all products based on default choices that one can further refine with his researches (using words or +/- operators). Your deductive method's fairly elegant - and, yes, it allows for a more precise control -, but I still find it weird to deal with both those checkboxes and check marks on top of the search engine's complications (without using filters). In the end, the result doesn't follow a common visual convention and I'd expect a wide gap in user practices due to people understanding only parts of Gamesieve's numerous aspects.

Of course, you'll probably have detailled statistics of use in a few months, but, as far as I'm concerned, I wouldn't need most (if not all) of the sub-filters of the "Discount quality" category - especially the distinction between annual relative values and absolute ones for example - and, hypothetically, I could even do more or less the same without the "Current discount" one : as long as my current selection appears within a certain price range - and your "Current price" choices are already quite precise -, the result, especially if I've entered any keywords, feel precise enough. In fact, even though several of those sub-filters are interesting for complex statistical analysis, they probably won't be of much use - if not even confusing - for an user simply browsing for a genre, a serie or a single game.
Here's the catch : if one knows what he's looking for, a few basic options are enough ; if one doesn't (and wants to explore a genre or search a bit for inspiration), your other criterias have him covered (within the releases of a certain period, OS, price range, etc.). Then comes more peculiar needs related to specific pieces of information - and most of current wishes expressed through this thread seem to go in this direction (with release dates on Gog or details about engines used for example) -, but here again, those don't really require the fine sieving possibilities offered by such checkcrosses or the two first (discount) categories. They're hypothetically interesting for sure, but their actual effect doesn't convince me that much.
So, simply as food for thoughts of course, even though the current system makes sense, I could ditch or bump at a lower place the one, if not even the two first filter categories, wouldn't notice the conversion of the checkcrosses in checkboxes, and would worry more about the lower categories' visibility - which are probably be more useful for daily visitors, apart from people seeking precise marketing statistics about discounts. This, moreover, fuels the discussion about the interest of historical prices and discounts, in addition to the interest of relative value - within a given period (current year, etc.) - compared to, somehow, absolute benefits - since a game's release.

Edit : I struggled to post that reply because Gog's forums seems to forbid the use of the words "in-s-ti-tu-tion(al)". Fun fact !

A general update:

I've been fleshing out my ideas for importing your library+wishlist and enabling permanently hiding games. I'm confident now that I'll get to where I want to be with all of it, but even stripping away all fancy features and focusing on the core minimum, it remains a hefty amount of work.

I've been doing little but working on that (and playing games to recover), and doing that at a pace which isn't sustainable, so am now needing to walk away from it for a week or two, to get back to it with renewed energy, and hopefully some better pacing.

I expect to be asking for a first few volunteers for a data import by the start of September, though probably without any interface yet to show you the result - I primarily feel I need to get a better idea of how much trouble outdated products will give me.

---

I feel I also owe you all a further reply about title search, for which I need to sit down and write out the logical steps I've gone through way back when, to see if there's any point at which I can break that chain and usefully offer what is "expected", or if it all holds up and I can clearly explain why offering that really would be worse than what I have now. I'm not going to actually get to this until after import e.a. are finished, but this here is stating that I have not forgotten about that topic, and promising that I'll get back to it. Eventually. (Which I hope will be earlier than "soon".) ;)

I'm in for "data import by the start of September, though probably without any interface yet to show you the result".

Hi there.

First of all, cool project. I can only give you my first impressions so far, but if I'll keep using your site, I'll try to keep track of any grievances and/or good impressions. Then again, let's be real most users decide whether they'll use a site or not within the first one to five seconds, so hopefully even this much will be of some use to you. I'd like to warn you, it will sound negative, but I'm actually quite impressed so far. It's just that negatives pop up very quickly and might need to be addressed, while praising you would feel insincere so early and wouldn't have much constructive value.

Now then.

The very first thing that struck me is the layout, of course. Honestly, I don't like it too much. The visual separation is so slight it looks like a wall of text to me, which felt overwhelming right off the bat. And since it's a single long column, I'm forced to scroll scroll scroll my boat, even though my 4K screen could merrily fit multiple such columns. Some things seem misaligned or aligned in a bit of a strange way (more on this below).

I'm not convinced I need to see "XX% better than previous year's low" when I already have "XX% better than previous all-time low". It could be an interesting option or something hidden in some deeper info, but the way it is now, it's kinda confusing. Remember, we need to go brrrr over a bazillion games and swipe faster than the most arrogant Tinder user. Each and every piece of clutter makes this harder. Simplicity, visual cues and proper alignment make it better.

The rating section is aligned to the right side, and that's about 70 cm of physical distance in my case. Not everyone uses a TV for monitor of course, but I'd argue it's still too far anyways. Having to constantly jump left and right with my eyes is both tiring and distracting, since the lack of visual separation makes it easy to slip and jump across rows.

If the ratings stay where they are, maybe consider making them uniform. Same width for each and every rating line, always keeping Verified and Total on two separate lines, right-aligning the container but justifying the content. Personally I'd also re-order them like this:

Verified : 5* (20)
Total : 5* (40)

The search are seems completely misaligned (aligned to nothing), and it probably doesn't need the label. Consider the usual Type to search written inside the input box which disappears upon typing. Saves spaces and eye movement, since you jump straight to the input box, instead of jumping to the "Search games" label and only then the input box.

The left-aligned information in each game row could also use some better harmony. Consider replacing centered text with justified one (I realize this can be very taxing to finetune when trying to optimize for many resolutions), and maybe avoid having too many font sizes and styles (there seems to be 7 to 9 different text styles within a single game row). The Includes and Expansions sections are leaking into the area that otherwise looks like a dedicated column for tags.

The lack of visual separation and visual cues/anchors for the eye to snap to, combined with multiple different font sizes, styles and alignments, and sometimes an overwhelming amount of information make the central (and most important) column somewhat unpleasant for me to process. Nothing I could quickly scan through. The game thumbnail is almost too small to be identifiable, and plenty of games are easier to identify by thumbnail rather than by name (at least for me). Oh and by the way, if I zoom in, I become completely lost as to what belongs to what, that is unreadable for me.

Finally for this section, scrolling the noodle is quite choppy for some reason. Even Facebook is smoother for me, and their code is horrible! So it's probably not my browser. Based on your project (which proves your capability, persistence and attention to detail, that is obvious already to me), my guess would be that your code isn't even more horrible than Facebook, but rather that there's some particular mismanaged thing.

I've ran a quick profiler run, just reloading the page and immediately scrolling up and down. I tried to be as fair as possible and did the same thing with Facebook's default feed. I'm not profficient enough with Profiler to analyze the results without spending hours of my time, but even just looking at the total times spent by different tasks seem to point to some specific problem with your site:


GameSieve

59 ms Loading
178 ms Scripting
778 ms Rendering
2879 ms Painting
1308 ms System
1702 ms Idle
6903 ms Total

Facebook feed

313 ms Loading
6305 ms Scripting
964 ms Rendering
134 ms Painting
2239 ms System
261 ms Idle
10216 ms Total

Notice that your site spent almost 3 seconds Painting out of 7 seconds recording, and it was clear from the timeline that scrolling around triggers Painting over and over and over like crazy. Facebook spent only 134 ms on Painting out of 10 seconds of recording. 778 ms spent on Rendering also don't seem right. Facebook with all of its images, videos, multilayered containers and all shouldn't have spent less of the total time Rendering than GameSieve which is mostly just plaintext and a few small images. So unless I happen to be the only one who noticed very laggy/choppy scrolling on your site (not sure how would Facebook be snappy in that case but whatever), you should look for the source of all those Painting tasks. Run the Profiler yourself and you should find some clue. Might be wise to fix this sooner rather than later.

Some final thoughts...

I really like the whole idea of a more robust rating of the actual value of a discount, rather than just showing the current discount which can be completely misleading. For one game a 40% discount might be a rare chance to grab that game at the best price since it usually never goes lower than 30%. For another game a 95% discount might actually be a daily occurrence, no big deal.

I am not so sure how happy GOG might be about this, since in some sense it is disrupting their business model. It might be a good idea to check with them, to avoid a situation where they don't do anything now because your site has a niche user base but only after months of hard work when your userbase gets so big it starts to eat at GOG's profits. You know, GOG doesn't show the best historical price for a reason... dunno, I'm no expert on either marketing or law. It's just that I actually wanted to make a similar project, but decided not to exactly because I was afraid of this, and I kinda felt bad. Not that I wanna make you feel bad... uhh, sorry, it's just something I'd recommend thoroughly thinking through, if you haven't done so yet.

I am not such a fan of the default sorting. Ideally, I'd want to be able to set the primary, secondary and tertiary sorting myself, instead of a single preset (you said the default was highest discount then compared to historically best discount then release date, right?) or individual classical sorts. Sorting by date is something I'd probably never use, GOG's own Highest Discount then Alphabet sorting is pretty close to ideal for me. The biggest downside of that are games that never do discounts below, say, 60%, because they'll be in the middle of pagination, and I'll have no real way of knowing whether a game does or doesn't go lower in the first place. If I'll end up using your site regularly, access to historical prices will definitely be the number one reason. That's the single most valuable thing your site offers.

You could of course make it even more tempting by adding user voting on tags, including adding new tags. That could make your database actually superior over GOG's own, because I don't always agree with their tag assignment, and I miss several tags there. I might end up using your site more than GOG in theory.

Oooh, before I forget. Since you make money by affiliate links, maybe consider adding a very visible note that coupon extensions like PayPal Honey are scam and steal your affiliate income.

( Exposing the Honey Influencer Scam by MegaLag: https://www.youtube.com/watch?v=vc4yL3YTwWk )

P.S.: All of the above is my own opinion, my biases and preferences. I don't have a PhD in UX, typesetting, or anything really. I did work as a webdesigner for a while though, and I do try to think about layout and design on a deeper level even now, so I'd like to think I'm not giving you completely trash advice that will ruin your site.

I'll have to do a Columbo here because I forgot to mention one important thing: the left panel could use some re-wording.

For example, the very first section, Discount Quality, is rather confusing to me. Both the "discount quality" term itself (I don't think I've ever heard those two words together), as well as the "improved" terminology. Not sure what would be better, but if I imagine hearing someone say:

"Oh did you know, Doom is now sold at a discount of improved quality."

That doesn't sound like anything anyone would say. Sorry for not being able to offer anything constructive here.

The other thing that bothers me a little is the "or more" and "within 10% of the best", etc. Those are just too wordy, why not use simply "90%+" or "> 90%", and "Top 10%"? That's what I would expect and immediately understand, without doing a double take.

Same for all similar parts where you use words instead of symbols: "at most", "up to", "to", etc. Not sure how many people would agree with me, but seeing it worded out like this every single time makes my eyes kinda... stare into the distance blankly. There's something about the repetition which makes it look like a scenery or something, it's hard to describe. But common math symbols don't do that, plus they only take a single character.

I would 100% highly prefer symbols over words, especially when repeated.

up to 2
from 3 to 4
from 5 to 6
from 7 to 8
above 9

is much more fatiguing for me than

≤ 2
3–4
5–6
7–8
≥ 9

( Alternatively 3..4 and 9+ )

You might also consider making certain sections collapsed by default, so that the user immediately sees all the filters, not just all options for first two to four filters.

Hi! Thanks for all the extremely detailed and useful feedback! I'm not going to take action on any of it in the next couple of weeks (I hope), as I need to take a break for personal sanity, but I will come back to a lot of this.

Thanks for the screenshot. That's rather eye-opening. Is this really your default browsing experience? I was operating under the assumption that anyone with a screen like that would be running at some high-DPI mode, or at the very least have set a much larger default font-size, which'd make my site render like your zoomed in view.
How do other sites deal with this? Any examples where you feel they get it right? (I can make my site's font-size increase to 200% based on your screen width, but form controls and images wouldn't scale along, so that wouldn't really lead to a decent experience...)

I'm showing it because it's my default and primary sorting criterion: there are so many games which hit their all-time low in a one-off event 3-4 years ago, and I don't want to constantly rank them based on doing worse than that, when realistically they'll never hit that same price again.
All the same, I am vaguely planning on offloading some detailed price information to a by-request secondary section, so will take a second look at which number deserves to be primary at that point.

The left of the input field is aligned to the left of the box covers. But I've heard this feedback so often by now that I'm probably going to admit that this is not good enough, and I'll just left-align the entire thing, or left-align the label with the covers (though then on small screens I need to jump it back to what it is now). (Of course, the last time I heard this feedback, my field turned out to not actually be aligned due to different font metrics, where now it is. Maybe I should instead look into lining out the info box to only the middle column?)

This I won't do, since that makes for bad accessibility. The label (text before) and placeholder (text within) have very distinct meaning and reasons for existing, and I'm not going to join the vast number of sites which get it wrong.

Your zoomed in screenshot looks like 300% zoom compared to the 4k shot (so 1280px equivalent), while 200% zoom (1920px equivalent) would be closer to the expected default experience which I'm designing for.

Which browser is this with? I haven't heard any prior feedback at all about this, and personally am experiencing buttery smooth performance, which I'm really pleased with. Looking at a profiler myself, I do see a lot of painting, but nothing excessive. Could be an effect of information density times 4k resolution? Though if I try to emulate a 4k screen by setting zoom to 50%, I still don't see it. Anyway, I'll dig into this more - thanks for the signal!

Yes, exactly! That's the core concept I've been missing everywhere else, and so decided to implement myself.

I've thought about this whole area a lot before I decided to apply for affiliate status - asking myself critically if my site wouldn't be a parasitic presence. (Since I really appreciate GOG, that's not something I want to be.) My conclusion is that price trackers have existed for years, and they're all GOG affiliates, so simply making information about historical lows available is not something GOG has a problem with. I also think I'm a net positive to GOG, exactly because I highlight deals which otherwise wouldn't be (easily) recognized, and thus drive a (tiny) stream of sales which otherwise wouldn't have happened. And if people use my site rather than one of the multi-store price trackers, they won't get tempted by e.g. a better deal at Steam.

GOG manually reviews affiliate requests, so someone at GOG is aware of my existence and approved it - beyond that I've tried to get in touch with them with questions about API usage and to see about contributing back data fixes, but that's met a deafening wall of silence.

Currently it's price improvement over the last year, followed by price improvement over all-time low, followed by how long since the all-time low happened, followed by how many times the game has been on sale, followed by rating, followed by discount, followed by release date.
The basic idea on that entire sorting is that I want to surface "unusual(ly good) deals", and each step in how that sorting evolved is basically a way to push down tired old deals which we see all the time.

Most visitors would never come up with a sort like that, probably opting for just sorting by discount and getting the 95% whale rock games at the top. I feel strongly that my way is better. :)

I see that one getting a surprising amount of usage - I suspect (but deliberately don't track) that a lot of it is by repeat visitors who just want to see what's new.

I'm very glad I found gamesieve, I check it often and use it as my main way of keeping track of games and prices. Just wanted you to know someone else is happy with all your hard work!

Sale stats for games the way I count them:
4094/5649 (72%) games on sale. 2792 (68%) of those sales at the best price ever or better, 144 improved.
7752/11095 (70%) products on sale. 5441 (70%) of those sales at the best price ever or better, 282 improved.

Updated after a belated ending of a previous sale, and near-simultaneous addition of some new games:
3950/5652 (70%) games on sale. 2676 (68%) of those sales at the best price ever or better, 132 improved.
7514/11100 (68%) products on sale. 5231 (70%) of those sales at the best price ever or better, 259 improved.


This is the first big sale since I doubled the currencies I'm handling (by adding EUR and USD for all countries with local currencies), and together with the API being pretty slow to respond, processing time of all prices has gotten seriously out of hand. I don't know yet how I'm going to improve this, but it's clear I must spend some time on that.

In the last two days, GOG has removed the Vlad Circus Bundle (containing the two Vlad Circus games) and the Bad Cheese demo. Does anyone know of an existing thread where it's worth mentioning such removals, assuming there's interest?

Since no actual games were removed, I kinda feel that Games removed from the GOG catalogue, by year isn't the right place.

(Heh, and I see my last two posts there were about a demo and a bundle. Clearly I'm not very consistent with my feelings. Maybe I should just ask there...)

Completely invisible on the front-end, and I won't know if it actually makes a difference until the next time GOG has significant performance problems, but this should be somewhat mitigated now, specifically preventing a bug on my side which could only occur when the running time of a harvest exceeded certain expectations, which then caused it all to take double the amount of time.

Well, thinking that the faster updates, or reduced delays, will be visible :)

Heh, true! :)

(Did you actually notice any delays when the sale started? Or is this more theoretical?)

I just looked once after the sale started to go through the list of improved discounts for price below $5, since I couldn't be more specific and go right up to the $5.44 that the PSC code I got bought.
So theoretical, but just saying that people will notice any discrepancies between what's listed and what's on the store, so while it probably couldn't really be said that they'll see the lack of them, it will nevertheless be a visible improvement in the sense of reducing reasons for complaint :)

I've seen the following pattern too much: A (presumably) new visitor searches for "cyberpunk" (e.a.), doesn't know nor notice the collapsed grouped products, searches for "cyberpunk ultimate", (presumably) still doesn't consciously see that section, and leaves.

Breaking expectations is hard, even if the result is ultimately more useful.

To hopefully mitigate this pattern, I'm now auto-expanding the grouped products for the first search result if it's scoring significantly higher than the next result (so that's only when searching and sorting by relevance). And I'm doing this with a visible transition. (Ick!?)

Give it a couple of weeks to get used to. If any of you are still bothered by it after that time, let me know about it here and I'll figure out an opt-out or a way to limit the behaviour even further.

---

In happier news, I'm finally starting to make visible progress on the wishlist / owned products import. Still a looooong way from done, but a lot of necessary infrastructure has finally taken shape in such a way that as of today I'm seeing actual functionality doing something when interacting with GOG. @mrkgnao: I'd currently expect something like 2-3 weeks before it'd be useful (for me, not for you) to have you as a test subject for an initial import. Will PM you when that time actually comes.