Focus
Which languages should I localize my game into?
The birth of ROI-based localization
Yuri Petyushin
Yuri Petyushin
Yuri Petyushin is the chief process officer for Allcorrect Group. He likes building things — systems, teams, businesses and houses.
T

here are a lot of articles out there on the subject of localization ROI, but I’m sad to report that most of them are general overviews written by localization companies themselves that provide very unsophisticated guidelines such as “localize into popular European languages first, then into Asian languages” or vague assurances that boil down to “invest in localization, and we promise it’ll pay off. Just trust us — we sell localization services every day.”

There are a few exceptions to this sad state of affairs.

CSA Research has conducted two studies entitled “Can’t Read Won’t Buy” (2006) and “Return on Localization Research” (2010). These studies show the effect of localized content on buying behavior in a number of different industries. Both works can be purchased on CSA’s website.

According to a report from Distimo covering 200 apps and 12 countries, localization can increase revenue by 26% in a week. This one data point is tirelessly and shamelessly quoted by many translation vendors, often without a link to the original report.

A 2011 study by the European Commission revealed that 42% of Europeans never purchase products and services in other languages.

OneSky has created a decent app localization guidebook that provides general guidelines on how to increase your app’s ROI.

Additionally, our own company conducted a survey of 10,000 Chinese and Russian video game players (2014-2015) that shows how localization quality influences buying behavior.

The problem is that none of these studies answer the simple, practical question of “How much money will my game, given its specific genre, make, and is it worth it to localize it?”

This is why in 2017, we set out to create an ROI-based localization framework that would be based on hard data (as opposed to being survey-based) and geared specifically to video games.

In order for the answer provided by this framework to be specific and relevant rather than just general, we chose the mobile games market as a pilot research project. In order to answer the main question of “What will my localization ROI be?” we have broken it down into several questions that represent various stages of our research.

  • 1. Which mobile markets are the largest?
  • 2. Which markets are the most profitable for a given genre?
  • 3. Does localization increase revenue?
Figure1: Top ten mobile game markets.
Figure 1: Top ten mobile game markets.
Figure 2: Which markets bring the most revenue.
Figure 2: Which markets bring the most revenue. China is not represented on this graph because reliable data for Android could not be obtained. We assume, however, that China would be plotted somewhere next to Japan.
  • 4. Will localization pay off, and what will a given project’s localization ROI be?

Question #1: Which mobile markets are the largest?

This stage of our research began in 2016-2017 with a simple analysis of mobile game markets by size (Figure 1). Since profitability is more important than revenue, we had to compile data from Statista, Geenapp, Chartboost and StatCounter in order to come up with a relative indicator that we call “marginal profit.” It represents the difference between the total market value and the cost of a mobile game install by all users. Here’s the formula: marginal profit = market revenue – number of users × consumer price index (CPI). Marginal profit is thus an indicator of a market’s desirability.

Question #2: Which markets are most profitable for a given genre?

Averages are fun. Everyone in the world has plenty to eat. On average. However, in reality, some parts of the world are experiencing famine while others are experiencing obesity epidemics. This brings us from the general “mobile games markets” (average) to the role-playing game (RPG) genre (specific), which served as our pilot research project. The top mobile RPG markets by revenue are shown in Figure 2.

In terms of profitability, Figure 2 shows how average revenue per download relates to revenue and downloads. As you can see, countries tend to cluster in the bottom left corner, although there are clear outliers — these are the juiciest markets. Average revenue per download is represented by the size of the circles. The bigger the circles, the better!

Question #3: Does localization increase revenue?

Now that you’re about to dominate the entire world with your awesome new RPG, the logical question to ask is “Do I need to localize my game, or can I just publish it?” People in Japan and Korea speak English, so they’ll buy it anyway, right?

If you’ve read CSA Research on this topic, you know that the answer is “No.” However, we need hard proof backed up by data, so here are some simple steps that our research team took to figure out whether or not localization really increases revenue:

  • 1. Using Priori Data, we pulled data from 55 countries and two platforms for a total of 200,000 games.
  • 2. We went to the title-specific pages of those games and fetched data on languages and countries where the games were published (our proprietary robot Nastya did a lot of clicking and ended up having to take a break to recover from carpal tunnel syndrome).
  • 3. We merged the two data sets. We compared, cleaned and verified the data.
  • 4. We found missing data points, swore, found workarounds, made calculations and assumptions, pulled our hair out, lost sleep and eventually came up with a decent database.
  • 5. We analyzed the data, looked for patterns and tested our hypothesis.

Our first conclusion was that about half (47%) of all the games in our database are only published in one country (see the pie chart on the left in Figure 3). These games comprise a very similar part of the revenue chart (41%) for all of the games reviewed. The games published in a single country thus only make just slightly less than their “fair share” based on their quantity. However, the bulk of this revenue is generated by only a handful of games. These turned out to be huge Chinese titles that rake in so much cash in their local market that they don’t even bother to expand. In fact, just three Chinese games represent 28% of the entire iOS market.

Unfortunately for most companies other than Tencent, most companies are not Tencent. The pattern above is therefore an exception to the rule and does not apply to most other games.

Figure 3: Deep dive into well-published games.
Figure 3: Deep dive into well-published games.
Looking at these pie charts, you can easily spot another anomaly: games published in 53-54 countries earn far more than their “fair share” of the pie — 30% of all revenue — despite the fact that they can’t even be seen on the “quantity” chart.

Zooming in on the segment of well-published games in Figure 4, you’ll see that the size of the circles (revenue) of games published in 53-54 countries is substantially larger than those published in 48-52 (at least twice the size). The question is “Are their circles larger because they’re published in many countries, or are they published in many countries precisely because they have big circles?”

You can see that one significant difference between these two groups (circled in blue) is the number of languages into which they are localized (the Y axis). This could be interpreted to mean that adding languages (localization) significantly increases revenue. But does it really?

Figure 4: Revenue by number of languages and number of countries. Bubble size = revenue.
Figure 4: Revenue by number of languages and number of countries. Bubble size = revenue.
To answer this very critical question, we took a closer look at well-published games and analyzed which markets they actually make money in. If you look at Figure 4, the localized games bravely explore the full extent of both the X and Y axes of our chart, which represent the number of downloads and revenue. In this case revenue is represented as the percentage of all revenue generated by the top 250 games for this specific country.

There is definitely a strong correlation between revenue, downloads and whether your game is localized or not. Publishing a game without localizing it therefore provides a much lower ROI.

Question #4: Will localization pay off, and what will a given project’s localization ROI be?

Now that we’ve established that localization is, in general, a very good thing, the next logical question is which languages you should start with. You could just consider the most common languages that the well-published games on the chart below are localized into, but if you’d like to understand whether or not your localization will pay off and estimate your ROI, keep reading.

To calculate ROI, we came up with this simple formula:

Revenue for top-n rating
Expected ROI depends on where you land in the top 250 list for any specific country, so we broke it down into categories (K1 = top 1-15 in the country rating, K2 = top 16-30, and so on).

The cost of marketing is estimated as CPI multiplied by the average number of downloads for that category (K). More on that later.

Localization costs are estimated based on average localization cost per word for any specific language multiplied by the average number of words in an RPG (the number we used is 70,000 words). We also spread out localization costs into Spanish between Spain and other Spanish-speaking countries (based on the size of the Spanish-speaking population in the target markets). We did the same for other languages and countries, with each country having up to three languages in our calculations. (So the localization costs associated with breaking into the Swiss market are comprised of proportionate values for German, French and Italian).

If this all seems way too complicated, here’s a simplified version of the formula. You basically just look at your revenue and subtract your marketing and localization expenses. Divide that by your localization costs, and you get a percentage representing your localization ROI.

ROI
Then we applied this formula to our data and this is what we got (Figure 5).

Needless to say, there are multiple ways in which we can and will improve this research:

  • The costs of breaking into new markets don’t include engineering and design expenses associated with changing the interfaces and such.
  • We can’t really assume that all installations are paid.
  • The categories (K) are based on revenue rather than downloads, which would be more appropriate (this will be fixed in later versions).
  • Our study is focused solely on mobile RPGs.
  • The study does not include platform fees (this will be fixed in later versions).
Figure 5: The ROI of East Asian markets.
Figure 5: The ROI of East Asian markets.
However, the significance of this study is that it’s the first and only research that provides answers to practical financial questions asked by game developers that are looking for ways to expand and multiply their revenue. It provides a common framework for comparing different markets and measuring them with a single yardstick that is meaningful and easy to understand. The lively hour-and-a-half-long debate that the research generated during LocWorld Silicon Valley, where it was first presented in 2017, marked the birth of ROI-based localization and confirmed its relevance to everyone who participated in the round table, including Tencent, Blizzard and other game developers.