It takes a long time to make a game. It takes a long time to gather enough wishlists to get visibility at the launch of your game. However, many developers fear that if they talk about their game, and create a Steam page too early, their wishlists will “get old” or the players will “get tired.” Because of this fear, many developers chose to develop in private without publishing their Steam page until closer to release with fewer wishlists.
Is this fear justified?
In today’s blog I am diving deep into a case study for the 2021 game Song of Iron, a beautiful action game that underperformed a bit despite collecting almost 90,000 wishlists pre launch. The game’s solo developer, Joe Winter, wondered whether the softer sales were due to his game’s wishlists “getting old.”
Let’s investigate.
The Story of Joe Winter and the Song of Iron
Joe Winter is a AAA-grade animator who worked on Halo Infinite and Halo 5. You can see his amazing reel here:
Like so many indies before him, he quit his job to work full time. His game Song of Iron is a side-scrolling 2D action adventure game built in Unreal. The game is beaaauuuuuutiful. And not surprisingly this is one of those games I describe as “Seeing is believing” where a single screenshot, gif, or trailer can make the game go viral and rack up tens-of-thousands of wishlists.
In fact as soon as he launched his Steam coming-soon-page, he created a Reddit post with the classic “I am working on this game” but paired it with an amazing trailer.
The post earned 135,000 upvotes and netted the game more than 13,000 wishlists within 7 days.
This is what I mean by seeing is believing.
From there, Joe ran a textbook perfect marketing campaign to capitalize on a beautiful game that does well visually. He relied on a bunch of Reddit posts, trailers, and big gaming events like Xbox showcase and The Game Awards.
Joe’s marketing strategy was inspired by David Wehle’s GDC talk (see Joes commentary on that here) which emphasized putting a Steam page up early so you can capture wishlists if you go viral while sharing details of your game’s development.
Here is a high level list of Song of Iron’s biggest wishlist gains:
| Date | Action | Result |
| February 29, 2020 | First day of Steam page | 52 wishlists day 1 |
| March 3, 2020 | Announcement Reddit post | 13,702 wishlists |
| July 23, 2020 | XBOX Summer Showcase | 577 wishlists |
| August 24, 2020 | 1 year anniversary Reddit post | 8,500 wishlists |
| Dec 12, 2020 | The Game Awards | 2,945 wishlists |
| March 26, 2021 | ID@Xbox games showcase (also Trailer #2 announce) | 1302 wishlists |
| June 20, 2021 | Steam Next Fest | 18,341 wishlists |
| July 16, 2021 | Release date announce trailer | 3,249 wishlists |
| August 31, 2021 | Launch Day | 3,048 wishlists in 1 day |
Total wishlists at launch: 89,082
Total deletes: 7,928
Net total wishlists on launch day: 81,154
Delete ratio: 8.9%
What is “old wishlists theory”
The “old wishlists” theory says that if someone wishlists your game when you first announce it, by the time you finally release, they are more likely to be “tired” of waiting and your constant marketing efforts and not buy it. Those early wishlisters will have a lower conversion rate.
Here is a THEORETICAL graph I drew up in MSPaint about how “old wishlist” theory would look if you were to graph conversion rates over time after a theoretical game launch. Each red line represents a group of people based on when they first heard about the game and wishlisted it.
Theoretically, the bars on the far left of the graph represent the early wishlister cohort and their conversion rate would be low (sub 10%). Zooming out, the conversion graph would look like a wedge that goes up and to the right where the older the wishlist, the lower the conversion rate. The graph looks like the rocks on the edge of the ocean, eroded down by the relentless barrage of marketing that you inflicted up on the population. The far right of the graph would have the “freshest” population of people who just heard about the game, so they convert at a much higher rate.
So if you believe in “old wishlist” theory, you should hold back and not talk about your game because if you start promoting your game too early you will actually be worse off by release.
Low wishlist theory is often evoked when a game with thousands and thousands of wishlists underperforms. People say “Ah, too much marketing too early, those people didn’t show up.”
So if the “old wishlist” theory is correct, then all the people who wishlisted Song of Iron from his viral Reddit post in March 2020 should have a low conversion rate.
The reality
The cool thing is that Valve does provide data to prove whether old wishlist theory is true or not.
Valve tracks when people wishlist a game and whether they bought the game. It is called wishlist cohorts. If your game has been released, from Steamworks you can download a CSV that records (on a monthly basis) how many people wishlisted then purchased the game.
If you mashup the wishlist cohort data with the wishlist data, then use a couple pivot tables, you can generate a graph that shows the conversion rate for each month of your marketing activity.
Here is the cohort conversion data for Song of Iron:
Let me explain it:
Valve only provides cohort data at a month-long resolution. So, looking at the graph above, you can see 14,801 people wishlisted the game from March 1st, 2020 through March 31st, 2020 (represented by the red line and the right axis). Joe Winter pulled this data for me in early January 2025 which means to date, 6.83% (the blue bars and the left axis) of those March 2020 wishlisters have purchased the game.
Now that you understand how the chart works, you will notice that the blue bar bounces around somewhat randomly. Some months are a high of 10%, some months are a low 5%. But there is no, clean clear wedge up and to the right. It doesn’t look anything like my theoretical “old wishlist” graph where the oldest wishlists would convert the worst, and the new wishlists would convert better.
In fact, one of the best converting months (June 2020) was 1 year and 3 months before launch.
Other interesting facts we can glean from this graph
- FRIENDS AND FAMILY: February 2020 has the highest conversion rate (15.38%) because he launched the page on the last day of February (side note, it was leap day) and gained only 54 wishlists that day. Those wishlisters are probably Joe’s friends and family and anyone directly following him already. So they are going to convert VERY well.
- STEAM NEXT FEST: June, 2021 was Song of Iron’s participation in Steam Next fest. It is the lowest converting period despite it being just 3 months from launch (they are among the freshest wishlists!)
- STILL DO NEXT FEST: Just because Next fest converted at the lowest rate (5.51%) that doesn’t mean Next Fest isn’t worth it. That is still 991 sales. We don’t know WHEN those people converted; some could have done it during a 20% sale or during the 70% off sale that occurred in 2024. But if the average sale price was $16.00 that is still (ballpark) $15,856. That still means Next Fest is worth it.
- POPULAR UPCOMING IS FINE. The last bar on the far right would represent people who discovered the game in the popular upcoming widget. Some theorize that people who see your game in the popular upcoming widget on the front page of Steam convert at a lower rate. That doesn’t seem true. The average conversion rate is 8.02% and that month converted at a much higher 9.13% rate.
The trend month over month
If I remove February 2020 (the month that only had 54 wishlists) then apply a trend line to the graph, it is as flat as can be.
If “old wishlist” theory is correct, that trendline would be quite low on the left and high at the right.
The average conversion rate is 8.02% and the standard deviation (how much the number vary from the average month) is only 2%.
There is no evidence that wishlists “get old.”
The 6-month sampling
I have heard a few publishers say something along the lines of “We only like to announce games 6 months before release date because we find “wishlists get old.”
Let’s examine this behavior.
If I crop the Song of Iron conversion data set just so the last couple months are shown then, YES, it looks like wishlists get old. But that is just a sampling error.
I have looked at a bunch of other games and the last couple months are usually higher converting than the baseline. That is because in the final months, developers are running their strongest marketing push to content creators, maybe running some ads, getting their trailer featured, and their game is showing up in Popular Upcoming. All of those activities have very high conversion rates (content creators are the best)
But just because the last couple months convert really well, does not mean the wishlists that convert A LITTLE less well are bad. 8% conversion is still better than the 0% conversion if your page wasn’t available.
Here is the thing, we indie developers are so small scale, we cannot get enough visibility in a 6 month period. Song of Iron ran a nearly perfect marketing campaign full of trailers, viral Reddit posts, and headline shows like XBOX Summer Showcase. There is no way Joe Winter could have squeezed all those activities into the last 6 months of his game’s marketing. You don’t get to choose whether you would like to collect 86,000 wishlists in a year or squeeze them into the final 6 months. As an indie you do not get to pick when you get your wishlists; you have to get what we get whenever they occur. If you delay your announcement, you miss out on the wishlists forever.
So why did Song of Iron underperform?
I asked Joe what he thought
“I definitely did a better job marketing the game than I did building it I think. I think I set pretty high expectations for players and I fell a bit short. Nothing crazy, you probably noticed the controls are not the best. It can really feel like a one man job at times and I think that came through. Difficulty was a bit all over the place also with players. Some struggling to ever get past the first boss, while others just breezed through. People who finished the game seemed to have a much better response overall than those that didn’t. Which makes sense why finish if you don’t like something. But it also tells me that I buried the lead a bit too much, didn’t do enough to keep people enticed. and since it’s such a short game, that lead to early quitting and a lot of returns”
Here are my thoughts on why the game underperformed. Joe is an amazing artist and animator. He has AAA experience working at the highest level. However, he is new to coding, design, and sound. His game took off because when he showed trailers or screenshots his best skill set was front-and-center. Everyone looking at the game assumed that the entirety of the game was at that same AAA level of visual quality. However as Joe said “controls are not the best… and difficulty is all over the place.” People couldn’t detect that from the trailers. Therefore there was a cognitive dissonance between the absolute top tier visuals and the less than stellar “game feel” and shorter play time.
Unfortunately the problem with AAA-looking real-time action games where you play as a dude with a sword is that you are going to be compared against the #1 dude with a sword game like Dark Souls. When players see Dark Souls quality graphics, they expect Dark Souls quality controls and combat depth. If you can’t deliver that, they will not buy. It’s not that Song of Iron’s controls are that bad, it’s just that they are not as good as Dark Souls.
Song of Iron has a “Mostly Positive” rating. This means people are saying “eh it isn’t bad, but it isn’t Dark Souls.” They aren’t going to leave a bad review, but they aren’t going to tell their friends that they need to go right away and buy it.
I wrote about this in my blog post “Why don’t some games sell well.”
Side note: If you are making a souls-like, you are underestimating how good your controls have to be. Souls-likes sell well on Steam but ONLY if their controls are perfect. It is a very very difficult genre to get right and I see so many great looking, interesting Souls-likes fail because the fans have some tiny complaint about the most minute detail regarding their combat. Get used to the complaint of “jank.” Some genres you can get away with jank. Souls-likes are not one of them.
Shoppers do not consider the age of the game when buying
I have looked at the cohort conversion history for many games and written about many others: Specifically Cosmoteer had a 4-year marketing period and sold wonderfully. I also looked at the cohort conversion chart for Zero Sievert.
Here is the Cosmoteer cohort chart:
As you can see Cosmoteers average cohort conversion rate is 19.31% (more than double Song of Iron’s average conversion rate of 8.02%.)
When a game fits a shopper’s expectations, like Cosmoteer, every single cohort converts well regardless of age. From the shopper who saw it the first month the steam page was announced all the way to the last one, converts at the same higher rate. When a game is missing something, like Song of Iron, every cohort converts at the same lower rate. The game’s hype slices across the entire timeline of when people wishlisted it. Time is not a factor.
It kind of makes sense. If the entire internet is buzzing about a game and you are deciding whether to buy it, you don’t say “Well, this game sounds neat, and everyone is talking about how good it is, and every streamer is playing it, BUUUUUT the developer did take too long to make it so I am just going to sit this one out.” That is ridiculous. People are either excited to play a game or not. It is that simple.
Do demo’s hurt sales?
The Song of Iron demo was live ONLY during Steam Next Fest in June 2021. During that period the game had its lowest conversion rate of 5.51%.
Don’t jump to the conclusion of “Demo’s Hurt Sales! So I shouldn’t post a demo.”
Demos reveal the truth. If a game is fun to play, a demo proves it. If a game has some issues, a demo reveals it. Basically a demo reveals the future.
If you look at the Song of Iron forums for discussion about the demo you can already see the complaints people had for the release:
I think the reason Steam Next Fest conversion rates were so low comparatively is that a lot of people tried the demo, were concerned with what they played, then said “if this is fixed I will buy it.” Then when the game launched and earlier reviews confirmed what they saw with the demo, they stayed away.
Although hard to stomach, this is why I think demos are such a great marketing tool: they are the proverbial canary in the coal mine. They tell you truly how your game will perform upon release. Whatever result comes from releasing your demo, don’t interpret the demo as the cause, it is the symptom. Don’t interpret demos as “hurting sales.”
Summary
I really appreciate Joe opening himself and his game up to examination like this. It is hard to take a deep look at a game that didn’t perform as well as was expected.
When a game underperforms a lot of incorrect conclusions can be drawn and I think it is important to dig deep to find out whether they are true or not.
Lately I have see a lot of people say “Wishlists don’t matter anymore” or they don’t “work” any more. The truth is more subtle between works / doesn’t work. Wishlists are potential. In an earlier post I used the analogy of the bowling ball vs the feather. The number of wishlists you have just indicate how high you are standing before you launch your game. If the game has everything perfectly aligned to what the market wants it will soar and float like a feather. But if there is something that isn’t quite right with the game, it is like a bowling ball that falls straight down (or somewhere in between).
This doesn’t mean that wishlists aren’t important anymore. You just need to go in with your expectations realistic. It’s totally possible for 100,000 wishlists to convert at 5% (5000 copies) or you could convert at 30% (30,000 copies).
You don’t know until you launch.
Valve doesn’t know either. In a blog post for Tape To Tape (a game with 53,000 wishlists at launch) I documented how Valve didn’t give them a popup deal at launch until they could verify that their wishlist conversion rate was high.
But I still see games perform well with huge wishlist totals. Tiny Glade did well with 1,000,000 wishlists gathered. Games like Sulfur had a fantastic launch after gathering 160,000 wishlists. Basically, wishlists are an indication there is the possibility things could go well, but are not a guarantee.
Nobody knows how you will convert until you launch your game. Wishlists are not a promise.
BONUS: When should I launch my Steam page?
I have answered this question on every single Q&A and I should do a deeper blog post on this but here is my quick rule of thumb:
- You have decided the specific genre and sub-genre of the game (e.g. you know for sure you won’t be changing from a metroid-vania to a souls-vania).
- Although art is NOT final, you have the art direction decided (e.g. you know you will be cel shaded, and not gritty grim dark.)
- You have in-engine screenshots of at least 3 distinct environments to show you are a deep game and not an asset flip. They don’t have to be fully functional environments, but they shouldn’t be concept-art fake-ups (you cannot fool Steam shoppers).
- You have a 30-second trailer that shows real gameplay. You can do some smoke and mirrors here if the AI isn’t fully finished, or the sound isn’t fully configured, but it should show how the game will actually function.
Extra consideration: If you think your game has something special to it, you should time your “announcement” to coincide with a BIG marketing beat like a festival like The Game Awards, or ID @ Xbox, or another 3rd party festival that matches your game like Wholesome Direct.