How Much is a Steam Game Wishlist Worth?

The answer: it’s complicated, but… $3?

This guest post is by systemchalk, data analyst extraordinaire and guest blogger here. He is happiest when wrestling with numbers. He writes here, streams here, and tweets here. We’ve edited his wording a bit, but his analysis is presented here with his approval.

Imagine a game (released or upcoming) and guess how much an individual wishlist for that game is worth. This question is surprisingly difficult to answer. There is considerable discussion about wishlists as an indicator for future sales and ways of getting more of them, but there has been less attention paid to the people who have wishlisted and when they buy the game and at what price (with notable exceptions coming from Jake Birkett and Simon Carless). This blog will help fill this gap by estimating the value of wishlists for three releases from Kitfox Games.

The take-aways:

• a wishlist is worth most when gained before launch
• after accounting for price differences, wishlist value across cohorts are similar between games
• a wishlist’s value lowers over time, as wishlisters intend to buy the game at a lower and lower price
• conversion rates are surprisingly stable in the long run

And before we dive into the specifics, keep in mind a few BIG limitations:

• We only really have 2 games to compare across multiple years of launch: Moon Hunters (2016) and The Shrouded Isle (2017). Perhaps in a few years we’ll have more data from more games.
• Wishlists weren’t tracked until February 13, 2017, which removes almost a year of data from the Moon Hunters set (the ‘correction’ mentioned below)
• Steam is changing all the time. So, it’s possible in 2018 or even in 2020, some core factor changed that made this data relatively inapplicable.

However, we used relatively similar discount strategies & timings across both titles, so if we assume that we can compare wishlist patterns across titles, these should be good candidates for doing so.

Why estimate wishlist value?

Estimating a value for a given wishlist is inherently interesting because it provides a benchmark for quality and can be compared against the costs and alternatives for acquiring these wishlists. In addition, estimating a value for a wishlist requires us to look at the individual behaviour of wishlisters and so allows us to answer more specific questions about this sought after group.

Here are the estimated values for a wishlist for the games Moon Hunters, The Shrouded Isle and Fit For a King, broken down by how many months from release the game was wishlisted (negative numbers are pre-release).

Pre-release wishlists have the highest estimated value and the value falls over time as wishlisters take advantage of deeper discounts. There are differences between the three games in that they were all released at different times, and Moon Hunters has a higher base price ($14.99, available wherever fine games are sold!) than The Shrouded Isle and Fit For a King ($9.99). These differences will be addressed in the breakdown below.

The estimate comes from calculating the average price at which a wishlister converted into a sale and is then discounted by the propensity for a wishlist in that cohort to convert into a sale. The rest of the blog will break down these components and what they say about wishlisting behaviour.

At what price do wishlists convert?

Steam does not provide information to connect an individual wishlist to an individual sale, but it does provide information about days in which each wishlist cohort (by month) converted into a sale as well as daily sale and wishlist information. We calculate an average price for each day that accounts for foreign exchange, discounts, and taxes (but not Steam fees), weighted by the number of sales. These daily averages are then weighted by the number of conversions for a monthly wishlist cohort and turned into an average for the whole cohort. This means that the average price for each cohort reflects the price at which the wishlist was converted, regardless of how far in the future the conversion happens, and better accounts for wishlisters responding to sales than taking unweighted averages.

Here are the average conversion prices for the three games:

Moon Hunters’ higher average price and higher estimated value of a wishlist reflects its higher price than the other two games. The sudden jump for The Shrouded Isle and Fit For a King are due to the fact that there were wishlists that converted the same month before a sale and so it is almost certain the average price for this cohort will go down over time as the games go on sale.

This highlights the important fact that these graphs are not static. Unless every single wishlist in a cohort converts, there is always the potential for a future conversion at a price that is different from the current average. Older cohorts are more likely to be stable, though there are many instances of wishlists waiting for years before converting. It also means that while each series has been shifted to report cohorts in relation to the release date, a game like The Shrouded Isle’s average prices reflect sales activity over a longer period of time than Fit For a King.

If the focus is a direct comparison between two games, it is better to decide on a cutoff for sales activity (for example, one year after release), especially when there are differences in discounting. We have not restricted the sales information here since our focus is on estimating the ‘true’ value of the wishlists (which would ideally include all sales that will ever happen to the game), but it recommends caution when making comparisons between the different games in this blog.

Finally, there is a significant data issue that needs to be addressed. Moon Hunters did not have wishlist conversion information until February 2017, a year after release. When comparing cohorts that are present for both Moon Hunters and The Shrouded Isle, the conversions happened at a similar fraction of the base price, with Moon Hunters consistently falling slightly below. In order to recover the missing cohorts we used the fraction of the base price for The Shrouded Isle (with a slight downward adjustment) and calculated a price for Moon Hunters. The similarities between the two series up to the first year are due to this correction, not due to any observed behaviour. While this may not be a universally accepted solution, it should be noted that the assumption is weaker than it first appears, since it is merely assuming that nothing happened in the first year to break an observed relationship between the two games.

How many wishlists convert?

The price at which wishlists convert is one factor in the value of a wishlist. Simply multiplying the wishlists by the average value at conversion will overestimate the revenue generated by wishlist conversions because not everyone ultimately buys the game. Since we cannot know which individual wishlist will convert at what price, the average price is multiplied by the probability that a given wishlist converted. In order to calculate the propensity to convert we took the number of conversions for each cohort and divided it by the number of wishlist additions for the given month (no correction was made for deleted wishlists. These are treated as a failure to convert). Here are the propensities to convert for the three games:

As before, the decision to use all available years for each series makes comparisons between games difficult. Comparing the propensity to convert to those reported for Ancient Enemy on Jake Birkett’s Twitter, for instance, it would seem that Moon Hunters and The Shrouded Isle did quite a bit better at converting. However, the series for Moon Hunters and The Shrouded Isle reflect years of opportunities for these wishlists to convert and at lower prices, while the Ancient Enemy series reflects behaviour a couple months after release, and so we should expect to see the conversion rate to go up over time.

One interesting result of using all the available data is to see that in the long run the propensity to convert is more stable than what might be expected, with Moon Hunters and The Shrouded Isle averaging out at about 0.3 (though recently this has increased for Moon Hunters). In fact, the relation was close enough that when correcting the missing data for Moon Hunters, no adjustment was made to the propensity to convert (again, the assumption being that the first year of Moon Hunters behaved the same as the remaining years in relation to The Shrouded Isle). It would suggest that for these two games that about a third of wishlisters bought the game, though some waited longer to get a lower price. The downturns at the end of each series likely do not necessarily reflect a change in behaviour for the later cohorts but rather the fact that not enough time has passed to accurately reflect the long run behaviour. The drops for the last cohort for The Shrouded Isle and Fit For a King in particular should not be surprising since our explanation for the high average price was the result of a few people who couldn’t wait to get their hands on the game after wishlisting.

This partially explains why the propensity to convert for Fit For a King is noticeably lower. The median time to conversion for The Shrouded Isle is 7 months, and so it is likely that half of the cohorts reflect the conversion behaviour of relatively early conversions. However, the short time period does not explain the entire difference. When compared to The Shrouded Isle’s sales activity in its first year, Fit For a King is about 0.1 lower across all cohorts. Based on what we see in the other games, Fit For a King may see an increase of 0.05 to 0.1 over time, but it remains to be seen if this game has a lower propensity to convert, or if its wishlists are simply waiting for a deeper discount.

Conclusions

When comparing the three games, the average price of conversion as a fraction of the base price was remarkably similar. Since the propensity to convert for Moon Hunters and The Shrouded Isle was similar, this means that Moon Hunters’ higher wishlist value is due to its higher price, while Fit For a King’s lower wishlist value is due to its lower propensity to convert.

The propensity to convert is more stable in the long run than what might be commonly expected. This suggests that wishlist behaviour after release is largely a negotiation over price. Pre-release cohorts (the ones most likely to be associated with fans of the game or the developer) are the most valuable, even though they have the same option to wait that everyone else does.

The value of a given wishlist by cohort can also depend heavily on the developer outreach in that period (events, etc), especially pre-release.

Finally, this long-term analysis method highlights that at least some amount of individual variability in behaviour can be accounted for by wishlist cohorts.

Post-Script

I’d like to take a moment to thank Kitfox Games for letting me look at their data and being so encouraging with the results of the analysis. Sales data is often closely guarded, but sharing this kind of information allows us all to become better informed, even if sometimes we’re only verifying if something is or isn’t an outlier.

In this spirit, if you have some sales data you’d like to see receive a similar analysis (either privately or for your own blog), I’d be delighted to look at it. You may reach me at systemchalk@gmail.com.