Since I think a major issue with GOG's catalog is titles just getting "lost"...

Does anyone know how GOG populates "you may like these products?"

I can't figure it out. For instance, I went to the page for Order of Battle (turn-based strategy war game) and one of the listed "you may like these products" games was Deponia, a point-and-click fantasy adventure game. And this is just the latest example of many odd "recommendations" from the "you may like..." lists.

I imagine it's either something random or there's a lot of fandom crossover between those two titles. For whatever reason.

They tend to favor the most popular ones / bestsellers (or so it seems to me).

from my exerience - totaly random

I would say they list fairly related titles, of the same genre or whatever, but they also include some random game with no apparent connection to the others, which I assumed was to invite the user to explore new territory: you like strategy games, so here's more of that, but hey, check this out, you might like it too!

They employ a sophisticated AI system - a monkey pulling game IDs out of a hat! Agreed. The tagging system is a good idea, only that the number of tags they use in the backend far exceeds what is available in the game catalog search section... the fact that they are missing a way to easily search for DOSBox or ScummVM-powered games is kind of ridiculous.

So, what you are saying is: It's totally random, then...

That would be the case if none of the games were related, but that's not what I see, most of them are, it's just that they add, intentionally or not, some odd title to the bunch.

Question: do you own any other games by Daedalic Entertainment (creators of the "Deponia" series)?

Maybe even "Black Guards" or "Black Guards 2" (which would be both: made by Daedalic AND turn-based (RPG) strategy games)?

Then the "you may like these" recommendations would be based upon Dev/pub, and/or genre.

Even your wishlisted titles may play a role in this.

To give you an example:

If I go to the "Black Guards" page...what I get shown under "you may like these..." are a DLC for "Black Guards", and the "special edition" to "BG", a strategy game set in space, and a manager game.

Now, the DLC and the "special edition" are easy explained - I own "Black Guards".

The "strategy in space" game is also easily explained - I own several strategy games (but not this one (or any of the series) in particular).

And the manager game is also easily explained - as I already other manager games (even more so explainable, since some of these manager games, that I own, revolve around the same theme).

And in regard to the strategy game that gets recommended to me: if I go to that game's page, the "you may like these..." recommends me four DLCs to that very game.
Not too hard to understand, why it does that, hm?

I do own Black Guards and Black Guards II... and if those were recommended, it would make some sense (they were not recommended). But Deponia is a fantasy point-and-click adventure game. There seems no relationship in either genre or gameplay -- that I can see -- to Order of Battle except that both are indeed video games and both are on GOG. ;)

Sorry, I think I didn't explain myself very well in my comment.

What I wanted to say was: the recommendations are based on games/genres that you already own, respectively have shown interest for in the past (wishlisted), and/or are developed/published by the same devs/pubs as games you already own/have wishlisted.

So: if you get recommended "Deponia", I would be willing to wager, that you already own other point 'n' clicks or that you already own other games by Daedalic (or have them in your wishlist).

Does that make more sense?

Edit:
Kinda proves my point. :)

There are two The Dark Eye Point&Click adventure games (Chains of Satinav, Memoria). They were also developed by Daedalic. The Dark Eye Fans probably have bought all four games (Chains of Satinav, Memoria, Black Guard 1 and 2). As a result, you get also recommendations for other Deadalic adventure games.

Internet stores are using subgroup/knowledge discovery algorithms to detect patterns within their user database. Best example is Amazon with things like "Customers who searched for "<insert product name>" ultimately bought". These algorithms are basically doing statistical analysis of the data.

The algorithms of GOG detected the subgroup of The Dark Eye fans and they have an intersection with the group of adventure fans because of the two The Dark Eye adventures. If most users who bought Black Guard 1 and 2 start buying HuniePop you'll see it also recommended to you.