STEAM? Games: Most Popular Months

December probably thinks this song is about them. They're so vain.

INTRO

When I started in the game industry I heard the apocryphal advice about when to ship a game. It harks back to the days of physical box sales, because holiday sales and marketing visibility: be on store shelves before Christmas.

Then there’s the equally apocryphal quote:

Players won’t remember that we shipped before Christmas. They will remember if we shipped a bad game.

I don’t know if this was actually said or if it just makes a good response to the “ship by Christmas” mantra, but it is surely a truism. We remember when a game is bad.

Because I now has the STEAM? datas, I decided to find out, at least on STEAM?, which month is actually the most popular to ship a game. And has it always been that way?

EXECUTIVE SUMMARY

If you like details and graphs, scroll ahead. Here is the summary for those with less time.

• The most popular month has varied from year to year, but October is the most popular, on average
• Likewise for January, not always in last place, but almost.
• December does NOT fare the way many might expect.
• The favored pattern of months differs between PROFILE LIMITED and UNLIMITED games.
• Differences across months have been diminishing in what I refer to as the “modern era” — since 2019.

DETAILED FINDINGS

Writing a query to group game publications by month is pretty simple. On the other hand, the results are not entirely clear cut. Here’s a graph.

We have the usual problem that, long ago, the rate of STEAM? game publication was glacial compared to now. In 2003 there were three games published, each in a different month. On this graph 1 is indistinguishable from 0. That separation issue is true for many of the early years.

Normalized as Percentages

Plotting data on a logarithmic vertical axis sometimes helps but obscures data in the most recent years for these data.

So let’s look at each month as compared to that year’s total number of published games. Let’s also ignore the years where fewer than 100 games were published, and drop 2024 because we don’t have data for July through December.

That provides a bit more clarity.

Now we’re getting somewhere! Although the years through 2013 are still sort of noisy, the noise is getting smaller. Here are some quick observations:

• October is looking like a leading contender. It’s highest often, but not always.
• January is having a bad time of it, usually coming in last place for the year. But not always, even in recent years like 2019 and 2020.
• 2019 was a year when publication rates were the closest to a uniform distribution, where they’d all be at 8 1/3%
• The annual distribution as a percentage appears to be getting closer to uniform over time.

Smoothing Via Totals

Let’s get some data smoothing by looking at cumulative totals. That is, if we start smoothing out annual variability via integration, does that reveal anything?

So the data are smoother for sure, but we are back to having difficulty seeing the early years and recent years clearly on the same graph.

• We do see that October is clearly not only the winner, but continuing to accumulate a lead over time.
• November is not doing badly but is in a close race with some other months.
• January’s disadvantages are also clear — being when game teams are getting going again after a holiday break, and possibly after pushing hard to publish in Q4. ??

Normalize Totals as Percentage

Let’s normalize the data again. We do that by replotting each month as a percentage of the running total.

Some of the lines are converging so close together that I removed the symbols so we can see better.

• The ascendance of October is clear after the “noisy years”. Once it took the lead it stayed there. But that lead has been dwindling since 2013.
• January is also always the last in line, but it has been improving its share ever since 2014.
• September and November have been battling it out for second place, with December and August charging hard to catch up.
• It’s hard to see August because the lines overlap so much!

Overall 2023 Finish Line

So what does that race look like if the finish line was the end of 2023?

As we expect from the previous graphs, October is the clear winner and January is the clear loser.

November is holding on to second, but September, December, and August are in a pack breathing down November’s neck.

The other months are fairly separated.

STEAM?’s Secret-Sauce Quality Metric

In 2018 the STEAM? service began tagging games as Profile Features Limited. What does Profile Features Limited mean? I’ve not found a STEAM? definition of the term, but the SteamDB site describes it this way:

Steam automatically enables profile customization features as games reach certain player and sales metrics that give confidence that a reasonable number of customers are engaged with the game.

If you see the term Profile Features Limited on a game store page then it didn’t pass those hurdles (yet). We can use the tag as a divider of games into two quality groups as vetted by STEAM?’s secret-sauce metric.

Do game teams whose games pass STEAMs secret-sauce criteria prefer different months? Let’s break out the totals for LIMITED games and UNLIMITED and see what that progression looks like.

The LIMITED tagging began in the middle of 2018, so let’s also take this opportunity to focus on what I’ve been calling “the modern era” — the years since 2019 when publication rates took off.

These totals are just for the years 2019-2023, as those are the years for which we have complete data for all months.

About 75% of all games are tagged as LIMITED, so that subgroup has a large effect on the overall population. That compression makes the UNLIMITED groups monthly preferences look flatter in the graph, so let’s normalize these again as a percentage of the subgroup.

October continues to be the overall favored month and January the overall disfavored month, regardless of subgroup, but other months have changes of position.

Within a subgroup, the UNLIMITED games have different and stronger preferences for certain months as compared to the LIMITED group. Meanwhile it is not surprising that the LIMITED group looks nearly identical to the overall population because it was already 75% of that group.

Teams creating UNLIMITED games more heavily favor the second half of the year but also give more attention to March, April, and May. These months rival December for favor within the UNLIMITED subgroup.

Let’s get a better visual for LIMITED vs UNLIMITED by comparing both subgroups as a percentage deviation from the overall group.

• The LIMITED subgroup has a greater preference for February, May, July and December.
• The UNLIMITED subgroup has a greater preference for March, April, August, September and October. It heavily favors September and heavily disfavors December.
• Both subgroups agree on January, June, and November.

OUTRO

If published games were money then October would have the biggest bank account. It’s a good month to publish because you get exposure for the final three North American holidays of the year. Even Canadian Thanksgiving if you’re early enough!

There may be a trend of slowly decreasing variation between months over time, as a percentage. That said, the absolute differences are still growing because of the ever-increasing publication rates.

When we look at UNLIMITED games alone we see a trend to more heavily favor months leading up to the end of school (March, April, May), at least in North America, as well as the months leading up to the end-of-year holidays.

In any case, this was fun. And maybe more than a little bit inspired by watching the Paris 2024 Olympics Women’s 3000m Steeple and Men’s 1500m finals. ??

Cole Hocker’s race reminds me a great deal of Ellen van Langen’s 800m gold medal win in Barcelona, 1992. (Sorry, that video is terrible — it looks like scanned VHS.)

METHODOLOGY

I’ll start putting this last for those that care, rather than up front, like in an academic publication. ;)

The data I’m using for this investigation is the same data as the previous digs. To be specific, all STEAM? data presented here is for applications of type ‘game’ with a publication date between 2003-01-01 and 2024-06-30. The data were pulled from public STEAM? APIs.

RESOURCES

?? A previous article: STEAM? Data Archaeology

?? SteamDB description of Profile Features Limited tag on STEAM? games

?? A previous article: STEAM? Data - Video Game Ratings and Mature Content

?? A previous article: STEAM? Offerings in 2024 Will Approach 100,000 Games