Benchmarks

This post is part of my series on marketing benchmarks that will help you understand how your game is doing and what is normal. To learn more, check out the full benchmark series by clicking the button

A good demo is a critical component of your game marketing strategy. If you have a great demo, streamers will want to play it (sometimes repeatedly over months), the festivals you enter will earn you more wishlists, and fans will fall in love with your game and be ready to buy it day one. 

So how do you know if you have a good demo?

The most important metric nobody monitors: Median play time. 

I love looking at median playtime because it is actually measuring the time players are spending with your game. There is no hiding between splashy trailers, funny tweets, or faked screenshots. You can’t really “game” the system when you are looking at median playtime. 

Benchmark

Based on my research here is the median of the median playtime of demos for each tier of income:

🥉 7 minutes
🥈 18 minutes
🥇 38 minutes
💎 65 minutes

Data Source

The benchmark data you see above comes from 3 separate surveys I ran. I asked developers to share the median playtime number that is found in the Steamworks backend. The surveys were from Steam Next Fest February 2022, Steam Next Fest October 2022, and one general survey for anyone who at some point had a demo.

I didn’t record playtime data for the June 2022 Next Fest because Valve added a gamified reward for players who tried demos. Some crafty Steam users created bots to play every single game in the fest and it completely tanked each game’s median play time.

In the end I collected 130 separate games and their median play time. Here is the full graph with the combined data for the games in the surveys. Each bar represents an individual game

High level results

Across all demos here are the numbers: 

Average playtime: 19.75 minutes
Bottom 30th percentile = 10 minutes
Median playtime: 14 minutes
Top 70th percentile = 22 minutes

Games that have released 

But does a high play time mean that the game will do well? 

62 of the games in my survey have been released. I estimated their revenue based on the number of reviews and then grouped each game into the 4 revenue tier buckets. 

Here is the graph of all the released games, their earning tier, and what their median play time is. Each bar is a separate game.

Legend:

Magenta = Diamond Tier
Gold = Gold  Tier
Gray = Silver Tier
Brown = Bronze Tier

Just looking at the graph you can see the gold and diamond tiered games clustered at the top of the graph. Then the bronze tiered games clustered at the end of the graph.

If I look at each bucket and find the average and median it breaks down like this:

TierAverageMedian
🥉Bronze Tier8 minutes7 minutes
🥈 Silver Tier19 minutes18 minutes
🥇 Gold Tier:43 minutes38 minutes
💎Diamond Tier56 minutes65 minutes

What about that Diamond Tier game on the right side of the graph with such a low median play time? That is a weird one. The game is a co-op party style game. My hunch is that for demos a lot of people just poke their nose into the game to see if it will be fun, realize it is a party game and then leave expecting to play it later with their friends. Maybe party games don’t adhere to the typical rules?

But what about linear games?

One of the reasons some games have super high median playtimes is because they are “sandbox” or rogue-like, or procedurally generated games where you can replay them multiple times and get new results. So they are going to have inflated numbers. But isn’t this data biased against linear story games?

Well, in general, Steam players like LONG games with lots of replayability and randomly generated fun. Short, linear games are a harder sell.

In one of my surveys, I asked developers if their game was “Linear / Story Based” or “Replayable”

Here is a graph of all “Linear” games from my survey

Here is a graph of all “Replayable” games 

As you can see the top earning replayable games had a almost 20 minutes longer median playtime than the linear one.

Here is what the averages for each looked like

30th PercentileMedian70th Percentile
Linear games9 minutes14 minutes20 minutes
Replayable games10 minutes18.5 minutes28 minutes

So yes, a replayable game will have a longer median play time than a linear one but not by much. When I look at these aggregated numbers, no game had a median play time that was longer than 30 minutes. 

I also asked developers how long they designed their demos to be and most said 30 minutes. So linear or not, the average player will not play your game for the entire time you designed it for. 

That is the biggest issue, not whether it is linear or not but whether people want to play through it to the end.

How to find your own median play time?

If you have a demo or game on Steam you can check right now what your median playtime is.

  1. Go to Steamworks Sales & Activations Reports dashboard  
  2. Click Products and pick your Demo from the list
  3. Look at the table and find the value next to “Median time played”
  4. You can also click the median playtime to get some extra numbers.

Side note.

That detailed chart called “Users bucketed by time played” might seem interesting. You might look at that bump at the far right and think it means that someone LOVES you. But in my survey almost every game I surveyed had 0% there. I think the occasional time I saw that at 1% it was probably because some player somewhere left their computer on overnight with the demo running in the background.

If you want to get super specific, you can also embed Google Analytics tracking or Unity Tracking into your game. That will allow you to filter median play time for specific date ranges because the Steam provided numbers are a lifetime aggregate. If you make major gameplay improvements to your demo, you might not see those improvements in the median data.

How to fix a low median play time?

Well… fix your game.

First, figure out why people think about your game. 

Send a survey out to your mailing list and Discord server and ask open ended questions about what they liked or thought about your game.

Run a live playtest with a naïve user. Maybe they got confused or stuck on a hard boss or confusing puzzle.  

Embed analytics into each phase of your demo to figure out when people dropped out of your demo. Maybe your tutorial is too long. Maybe your tutorial is too short. 

The developer behind Do Not Buy This Game added an event to every phase of his demo and he could see where people dropped off. In fact he noticed a disproportional drop in the second event in his demo and was able to trace it down to a bug.

Also, maybe your game is just too hard. The developer behind Tower Escape told me his demo started at a 15 minute median play time. So he made some balance changes and made the default difficulty setting to “easy.” His median play time is now sitting at 42 minutes.

So monitor, improve, and monitor again. Keep an eye on your median play time!

Additional Information

If  you want to know how to playtest and get feedback from players, check out these two blog posts

Watch my talk all about demos:

Read these helpful blog posts

Check out my class

Check out my comprehensive course on selling a game on Steam: