We wanted to address some concerns we've heard from a few developers about store traffic to their games in October. Back then, after some developers let us know about the issue, we did some investigation, found a bug in one widget, and pushed out a fix a few days later (along with making some tweaks to the algorithm for that widget).
Here's the history of what happened:
On October 5th, we deployed some code to Steam that changed the defaults for our search queries to factor in sales and wishlist activity more than before. This change was meant to fix an issue where customers would search for a game, but that game would be much lower in the search results than other less relevant games.
This change had the unintended side-effect of de-boosting tags in the "More Like This" section on a game's store page. What that meant was that instead of seeing products similar to the game, you saw games that shared just one tag in common, but were effectively the most popular games on Steam. This isn't how we wanted that feature to work.
On October 9th, to fix the issue, we changed the algorithm for the "More Like This" section to increase the boost on tag similarity. However, we still kept a part of the change that factored in sales and wishlist popularity. On October 30th, we increased the weighting of tag-similarity again based on the results we were seeing.
To compound matters, from October 10-17, we were also running an experiment in the same "More Like This" section to test out a new algorithm which we hoped would be more effective in showing customers games that we think they would be interested in. This experiment ended up showing fewer products to a subset of customers, which led to a decrease in overall impressions in that area.
As a result, overall impressions and views for the "More Like This" section did show a substantial decrease for about two weeks after the initial bug. But traffic recovered around October 19th, and has stabilized since.
In general, we're always trying to show games to customers that we think they will enjoy, no matter who made them, what the budget was, when they came out, etc. We're constantly exploring ideas and trying new things to try to figure out the best ways to do this. The fact is, traffic is going to shift whenever we do this work; it may go up or it may go down for any individual product. However, not all impressions and views are equal – in the end, what matters is that we show customers games that they find interesting.
