How the Game Industry Turned Accessibility Into Profit
Exploring the production mindset shift, which changed niche options into vital marketing tools
While the game industry certainly had its fair share of deserved controversies in recent years, there is at least one area where it has significantly improved, and that is accessibility.
Game companies frequently communicate the specific options their game has to offer to those who require them. Finding extensive subtitle parameters or colour-blind filters has almost become a norm.
Some companies invest a lot of time and effort to push the boundaries of accessibility.
Naughty Dog implemented hundreds of accessibility options in The Last of Us Part II. Microsoft engineers designed an adaptative gamepad to meet the needs of gamers with limited mobility. Ubisoft produced audio-described trailers for vision-impaired people.
Everybody contributes to raising the bar, to establish new standards, and it’s fantastic to witness the video games opening the broadest possible audience. No one should be left behind.
As a player, I’m astonished by these changes, seeing the results of these efforts in action. As a game developer, I’m intrigued.
Why would companies invest this much for what appears to be a minority of their player base?
I’ve never worked directly on such features, but from experience, I can tell they can add design, technological and production challenges. Nothing impossible, of course, and a well-planned feature won’t be a big challenge, but ultimately some of them can be expensive.
Profits drive companies, so in cynical terms, why do they choose to invest in an area with a lower yield? The answer is in part ideological, but there has to be more.
The equation is tricky; you can’t solve it with the basic decision-model everyone uses. In this piece, I’m going to explore the limits of the RoI-driven prioritisation technique and then detail the Tetraclass model, a more advanced approach. It will help us figure out how publishers can recoup the high costs of accessibility and other less-intuitive investments.
The limits of R.O.I. (return-on-investment)
Product managers typically rely on a prioritisation technique based on the return-on-investment. It’s a quick way to evaluate any feature should you be able to assess two factors:
- Its impact on the overall experience (the more, the better)
- Its cost, which is usually equivalent to the implementation time
When you divide (or subtract) the former by the latter, you get an easy-to-compare score for each feature and can quickly sort them in a list.
At the top of the list will be the highest R.O.I. features, those which add significant value for minimal efforts. They are probably the ones you want to work on first. Conversely, at the bottom, you have the time-wasters, elements whose costs exceed their impact: you’ll implement them later or not at all.
Like all models, it’s both useful and misleading.
It’s useful because it’s relatively quick and efficient at putting vastly different items on equal footings to help make decisions. It’s misleading because you also lose most of the subtleties in the process. Relying only on the R.O.I. to figure out the production strategy doesn’t help to address the more complex and nuanced topics, like accessibility.
Limit #1: Unclear definition
The first issue with the R.O.I. approach lies in the vague definition of “impact”. Evaluating the fun-potential and frequency of a gameplay mechanic or item can be reasonably straightforward, should the team agree on the target audience needs.
How do you score the impact of a save system, though? Of taking time to debug correctly? Of colour-blind filters? Some features have potential, not to increase the value if you add them, but to lessen the experience of players if you don’t. It makes them incredibly hard to score.
The premise of the return-on-investment model is to enable comparison, and it already shows its limits when we try to balance positives against negatives. You’re back at relying on intuition to arbitrate such cases.
Limit #2: Dealing with edge-cases with averages
The second flaw of an R.O.I. approach is how it deals with split audiences. For instance, a colour-blind filter is a typical example of a feature with significant effect for colour-blind people and none on the others.
The impact of polarising features (such as a PVP mode, in-game cinematics or most accessibility options) is hard to evaluate because each person places a different importance on them.
You can do averages to circumvent the problem, but this is another cheap maths technique which eliminates all the useful nuance. On average, a human has less than two legs and more than one. You don’t design pants with averages.
Limit #3: Hard to leverage to your needs
The third limit I see in this prioritisation technique is what you can do with it. Applying a model paints a clearer picture of a situation, but it should also help take concrete actions to change this situation.
For that matter, the R.O.I. approach isn’t too helpful as it encourages primarily to reduce costs to quickly shift the rankings (because “impact” is a relatively obscure notion, cf limit #1). Whether you break down a feature into smaller chunk to evaluate individual R.O.I. or directly cut within to try to save costs, you will most likely decrease the end quality as well.
It can be a healthy way to optimise production and to achieve more in the allowed time, but, I’ve never personally encountered an occasion where this approach improved the overall quality of experience. It’s sometimes better to leave something for later if you can’t go big.
Evaluating R.O.I. is an efficient method, although limited. But if that were the only frame of thinking people used, the game industry would never work on accessibility features.
What would happen if you could more precisely evaluate the impact on satisfaction for each feature, for each audience? What if instead of reducing costs, you’d try to make a feature even bigger by increasing investments?
A few years ago, I came across a relatively unknown model which compliment perfectly the R.O.I. and addresses its flaws. By using it, I was able to approach the equation differently and see new solutions.
Introducing: The Tetraclass model
The Tetraclass model (also named Llosa matrix after her original designer Sylvie Llosa) helps us visualise the impact of a feature on the consumer’s satisfaction of the product.
Here is what it looks like:
The matrix has two axes:
- The potential of a feature to add satisfaction
- The potential of a feature to add dissatisfaction
Therefore, the matrix has four categories:
- Key (top-right): Features with the potential to impact the satisfaction in either direction depending on their quality.
- Plus (top-left): Features which will only increase the satisfaction when done well (no penalties if missing or poorly done).
- Basic (bottom-right): Features which only create dissatisfaction when done poorly (no bonus when done well).
- Secondary (bottom-left): Criteria with very few impact on either satisfaction or dissatisfaction.
To use this model, we examine each feature one by one and evaluate where they stand on these two axes. Let’s see concrete examples to see how it works.
The coffee example
I’m ordering a coffee, how do different criteria impact my satisfaction?
- [Key] Taste. When the coffee is good, I’m happy. If it’s not, I’m displeased. This criteria alone can impact my appreciation significantly, in either direction.
- [Plus] Chocolate treat. When there is one I like, I’m delighted by the little attention. When there is none, I don’t even notice.
- [Basic] Dirty cup. It’s the bare minimum, I don’t feel positive emotions for a clean cup, but if it’s dirty, it adds a lot of dissatisfaction.
- [Secondary] Cup colour. I don’t care about it; the colour is what it is. There is no impact potential, positive or negative, on my experience.
A video game example
When I purchase a game, what can impact my general appreciation?
- [Key] Length. Most players have specific expectations: too long, too short, fair enough; length can both impact positive and negative satisfaction.
- [Plus] An additional game mode. There is potential to bring extra variety and surprises, but by definition, this isn’t the primary focus; nobody would notice its absence.
- [Basic] Bugs. I expect my experience to be smooth; bugs directly hurt my enjoyment of the game and create dissatisfaction.
- [Secondary] Colour-blind filters. For 94.5% of players, they’re neither something to be excited about nor something which hurt their appreciation of the experience. They’ll just ignore them.
In both examples, you might disagree with my categorisation: like the R.O.I. model, there is an element of subjectivity (which can spark more in-depth discussions within the team on player preferences).
It takes a bit of time to map all the critical features of the experience on a tetraclass matrix. Still, once done, it gives us a new lens to understand our target’s expectations, personal preferences and how to approach development.
Strength #1: Better production strategies
While you probably won’t make groundbreaking discoveries by examining the Key and Secondary categories, but the new dimension helps to figure out better priorities with the Plus and Basics, since these two have a very different impact on satisfaction.
One of the pitfalls you can try to avoid is to under-invest on Basics while focusing too much on the shiny Pluses. Early in the development cycle, Key features should be the central focus, as they have the potential to make or break the game. The second next important category is the Basics.
Even later in the production, you should (re-)secure the Basics first (accessibility, comfort, debug, etc…) before dedicating your resources to the Pluses. The human mind tends to focus a lot more on negative than on positive: eliminating dissatisfaction sources is more important in the long term.
Strength #2: Leveraging spatial visualisation
The second significant strength of the Llosa approach lies in its second dimension: on a graph like this, we’re now able to represent features with different shapes. In the example below, I went for two circles colours linked by a thin line to illustrate how different audiences perceive the same element.
This allows us to preserve a lot more the nuance of polarised features, such as the middle one in the graph: it’s in the Key category for some, in Secondary for the others. If we were to average out these specificities, we’d probably not tackle this element early enough and generate frustration for parts of the audience.
Accessibility features typically fall in this category: using a Tetraclass matrix helps to visualise their impact on a diverse audience and plan for them.
Strength #3: Moving features around
The third massive advantage of the Tetraclass model is how we can use it to our advantage. Gaining a new perspective on your product is excellent. Leveraging a model to redirect your efforts on what you believe should be the priority is even better.
While the R.O.I. encourages to break down features into smaller chunks, Llosa Matrix helps to bundle features. For instance, instead of judging each accessibility setting individually, we consider them as a whole and adjust their position on the graph.
Each option alone doesn’t bring much for its cost, but once all put together, publishers have enough to communicate on it. Suddenly, what was a “don’t care” for a large chunk of the audience turns into a “plus”. Your satisfaction in the product also depends on your moral values, not just the features you’re using as a player.
Marketing campaign on the accessibility features raise the general awareness of all gamers; they don’t just convince the targeted players.
In this piece, I detailed how using a different prioritisation technique than the common one can shed new light on a complex equation and even inspire solutions. While the R.O.I. method is undoubtedly quick, this efficiency comes at the cost of losing a lot of nuances.
The accessibility issue got solved with a change of mindset on priorities, and everyone won. Publishers make money out of it; new people have access to video games; the dev teams are proud of their work.
Each game, each product is different: figuring out the ‘why’ is always better than nailing down the ‘what’. I’m personally convinced, there are other underdeveloped aspects of gaming left right now we could examine without return-on-investment in mind.