#1 - July 25, 2024, 6 p.m.
Blizzard Post

Hey everyone! There's been a lot of talk about 6v6 lately, and it's only intensified in recent weeks since we posted our latest Director's Take and made a Twitter post with a promise to talk about it in the future. Well, now is the future, and this is that next piece! So, let's get into it! Wait. Let's not get into it quite yet…

This is a pretty complex and nuanced topic, and it gets to the beating heart of what makes this game tick. Our players have different opinions and preferences about the ideal number of players in an Overwatch match, and so does our team. We’re a microcosm of the broader community and discuss this internally. I’m hoping that this piece will serve as an introduction to our own discussions and a jumping off point to engage with all of you!

I'd also like to set a few expectations and ground rules. First, it's pretty long (creators are going to lose their voices reading this one aloud). If you'd like a tl;dr, the 'What's Next' section should suffice. Secondly, I'm not here to tell you what you ought to like, or even what's empirically better. This is a game, and your preference for fun is relative to your own tastes, ability and experience. Lastly, there are years of decisions that affect such a fundamental piece of Overwatch. Some of you might be intimately aware of these, while others might not. Before diving in, I'd like to set the stage and give some context on some of the larger changes to the game that ultimately culminated in the 5v5 format we have today.

Historical Context

There have been three large format changes to Overwatch, and the first happened shortly after the game launched. This was Hero Limits. Before this there could be any number of a particular hero on one team. You could have six Genji’s, or six Reinhardt’s. If my memory serves me right, the winner of Team 4's first internal tournament was one Winston and five Zenyatta’s. We called it ‘Quinston’. Zenyatta Orbs stacked with no cap to healing at the time. It was glorious... and terrifying!

During early development it was necessary to have duplicate heroes per team. Our first core combat milestone only had four heroes. As the roster grew this was no longer necessary and it also introduced issues. Overwatch heroes have very distinct kits with impactful abilities. Stacking abilities could be fun, but it posed challenges to balancing many of those abilities in game. People reference double barrier, but back then you could have six. Imagine cycling six Immortality Fields! The number of team compositions and strategies available was huge. The variety was great, but we didn't feel like we could ever truly balance the game without severely watering down our heroes, so we made the choice to impose a limit of one of each hero per team. This essentially solved the issue of ability stacking, but it did have a few side effects. It traded some of the creativity in putting together team compositions for a more consistent and balanced experience. It also took away some choice from players. If someone wanted to play Tracer, but you were already Tracer, they would be unable to. A theme of this, and future decisions, was limiting match variety as well as player freedom and creativity in order to create a more balanced, consistent and competitive experience for players. Remember this point, because we’ll touch on it a few more times in this piece. I think the hero limit rule was a great decision, and we would make it again. The other two large format changes have a lot of positive benefits but are not as clear cut.

The second large change to the game's format was the move from ‘Open Queue’ to ‘Role Queue’, which limited team compositions to two heroes from each role. Before Role Queue, there was no enforced 2-2-2 team composition, and players had the freedom to select any role they liked once they loaded into a match. With Open Queue, teams had a lot of freedom of choice about what heroes and team compositions to play, but there were quite a few problems that Role Queue intended to address. Prior to implementing role queue, there was a stark imbalance in the number of people playing each role within each match. Damage was, by far (and still is today), the most popular role and tank was, by far, the least popular. In fact, seeing matches with five or even six Damage players on each team was not uncommon.

This role imbalance frequently led to wildly imbalanced matches between teams. The game played better when heroes from all 3 roles were selected. The disparity caused social issues as players would try to pressure others into playing tank or support roles, even if they didn’t want to. It made balancing heroes more difficult, since we couldn't balance around a certain role ratio for each team. And it even made introducing new heroes more difficult, since any new hero kit had to take into consideration hundreds of different team compositions and abilities that could overlap with it. The move to a Role Queue format was aimed to mitigate these challenges.

Moving to Role Queue solved many of these problems and added some additional benefits to the game as well. With Role Queue in place, matchmaking is based on skill within a given role, not just skill overall. Some players have wildly different skill ratings per role, so the ability to matchmake around set roles provides a big, measurable boost in match quality. Role Queue also solidified the identity of each role, and the fantasy of doing that role’s ‘job’ became clearer.

However, the switch to the Role Queue system did have some downsides, the largest of which was much longer queue times.

Prior to Role Queue, the matchmaker just needed to find 12 players of similar skill levels in order to put matches together. After Role Queue, it needed to find four players of each role to make a match. The total number of players queuing has no effect on this, it’s almost solely governed by the ratio of people queuing for the least popular role. Given the games’ lower relative ratio of players queueing for Tank than other Damage and Support roles, queue times went way up.   We’ll go into details on this later.

The team tried to address this imbalance in various ways. We introduced a priority pass system that gave players who played low demand roles (Tank) a shorter queue when they switched to Damage. This had a positive effect for several weeks, but its influence quickly faded. We also introduced systems to reward players for playing Tank, eventually giving them a loot box and a bunch of XP for queuing for that role. None of these were able to put a dent in the demand for playing DPS. Damage queue times drifted a bit lower over time, but much of this was due to the introduction of Open Queue Competitive, and the permanent loss of Damage players unwilling to wait that long for a match.

There were additional drawbacks with Role Queue. I mentioned earlier that many of our decisions made improvements to the game at the cost of match variety and player freedom. Team compositions were now locked to 2-2-2. This was another step to a more consistent experience within Overwatch, but sometimes consistency can start to feel ‘samey.’ There were some pretty exciting metas in the pre-Role Queue world and swapping to a hero of a different role mid-match was often a much bigger change (and more exciting!) than swapping to one of the same role. Like Hero Limits, Role Queue took away a part of the freedom in selecting heroes that served as the base of Overwatch, but I often wonder if the success of it is as cut and dry as that of Hero Limits. Could there be an in-between version that softens the original problems without completely solving them, and also isn’t quite as restrictive as our current Role Queue system? A bit more on that later.

The third, major format change

These challenges, as well as longer queue times, lead us to the third and most recent major format change to the game— the shift from 6v6 to 5v5. This is almost a blog within a blog, but we’re going to cover three things:

  • Why we switched to 5v5
  • Queue times after the switch to 5v5
  • What the team thinks about 6v6

Here we go.

Why we switched to 5v5

There were many reasons behind the transition to 5v5.

When we launched Overwatch 2, we talked about the greater individual agency that players could have in 5v5. Overwatch is a team sport, and it requires players to rely on each other to be successful. This was even more the case previously and the game could be frustrating if your team was out of sync. Many times, fights in 6v6 couldn’t resolve until ultimate abilities came online. Correct that, until ultimate combos could be executed to break through two tanks worth of mitigation supported by high, uncontested healing output. The world of 6v6 could have really high, highs, but reaaaally low, lows. One of the design goals of 5v5 was to try to raise that floor, even though it was at the expense of some of those high moments. 

Overwatch 2 also has a greater emphasis on FPS gameplay. There’s less damage mitigation and CC, and players get to shoot at enemies more. Crowd Control could take players out of the game for extended periods of time, especially when they were chained back-to-back. A five second Sombra Hack, Mei Freeze on her primary fire, Brig Bash, Ana Sleep, Orisa Halt and Sigma Rock could all happen consecutively to one player! Typically, a Tank who could often survive this onslaught… at least in game, their real-life mental health was another story.  

Damage heroes often complained that their role was the least impactful relative to the strength of super powerful Tank synergies and Support heroes undoing their playmaking. The emphasis on ability cooldown management, ult tracking, shooting barriers instead of players, and not big enough timing windows to just shoot enemies and get eliminations were all factors that contributed to this feeling.

Support heroes also had to focus almost exclusively on healing since there were bigger team health pools and more incoming damage on the field with the additional Tank player. In 5v5 they have more agency to add some damage or duel other heroes.

Tank synergy wasn’t always fun. It slowed the game down, and stabilized it, but sometimes it absolutely ground it to a halt. People point to the double shield meta as an example of this. It wasn’t just about double shields, though. The ability to stack a very defense heavy lineup, especially in game modes that required a team to push through a small choke, caused a lot of matches to feel like there was no progress at all. The game is faster now, and it doesn’t typically stall out. It feels like there is more room to move around a map without the need to earn every inch of it. In 5v5 fights are often less focused on moving around in a death ball formation and there is more room to take angles, flank, and duel other players, which brings in more of that FPS gameplay.

While 6v6 can have high moments, it's also hard to follow. Tracking nine other players is easier than eleven. There are fewer visual effects on screen, especially during big team fights. Overwatch is a pretty demanding game and moving to 5v5 made it a bit more cognitively manageable.

Lastly, one of the biggest benefits of 5v5 was dramatically reduced queue times.  Let’s explore that.

Queue time data and context

The chart below shows Ranked Play queue times per role after the introduction of Role Queue to Overwatch in 2019, compared to queue times per role in Overwatch 2’s 5v5 Role Queue format.  As you can see, the shift to 5v5 Role Queue dramatically reduced queue times for players in all roles, and most notably the Damage role (which is the most popular role).

 RankedQueueTimesChart.png

The biggest factor driving the difference between Overwatch and Overwatch 2 queue times is the ratio of players queueing for each role. We hear players speculate that long queue times are a result of not enough players playing Overwatch – that isn’t the case -- but instead that relative to the other roles, players generally queue tank less, leading to longer wait times for all while we wait for a Tank to free up to find a proper match.

So, in practice, the actual ratio of players queueing for Tank, Damage and Support is much closer to the current 1-2-2 format than the previous 2-2-2 format.  As a result, the 5v5 Role Queue format better mirrors actual player interest in each role, and making the switch resulted in shorter queues.

The sharp drop-off in queue times in October of 2022 is when Overwatch 2, along with 5v5, shipped. The queue times of all heroes dropped and are now much shorter than they were in the 6v6 world. This is almost all because the game required one less Tank in a match. If you look at the current queue times you can see that there is still a lack of Tanks compared to our other roles. Just ‘flipping the switch’ and returning to 6v6 would likely cause a return to previous queue times or worse.

Regardless of the amount of players in-game, queue times are always reflective of how many people are queued per role. In other words, the main driver of increased queue times is the ratio of players queued in each role. Since there are less players queuing as Tanks, queues would likely go back to the way they were before.   

Thoughts on 6v6

The game started its life with six players on each team. I discussed why we moved away from 6v6, but that doesn’t mean it’s without any merit. There were moments and madness that could happen with twelve players in a match that just don't quite coalesce when there are ten. It was almost too much to keep track of. Correction, it was too much to keep track of at times, but that was part of the charm! There are big moments in Overwatch 2, but they don't quite have that same chaotic, over-the-top feeling to them.

Part of that chaos allowed players to feel less pressure in a match. There was an extra player to take up a bit of the slack for someone that was performing poorly. And since that player was a tank, they had quite a bit of impact. There were a lot more hit points and defensive capabilities to burn through, which had a stabilizing effect on gameplay. Teams would fold a bit slower than they do now.

We went over some of the issues of tank synergy earlier but having two of them allowed one to focus on protecting their backline while the other held ground or pushed into the enemy team. The tank role had a lot of depth in understanding how to work with the other tank on the team. This part of the mastery curve took a long time to learn, and for the small group of people devoted to it, that was part of its allure.

We mentioned the difficulty in balancing tanks for 6v6 earlier, but the past year and a half has shown us that there are issues with balancing Tanks for 5v5 as well. For much of OW2 Tanks have been fairly balanced, but when one of them becomes a powerful outlier, like Mauga, that can be felt strongly in the world of one Tank.

What's Next

Ok! That was a lot of setup. I think I've put more effort into this than any essay I wrote in college! So where do we go from here? At the end of the day, we want to make the game that players want to play.

Because of the challenges mentioned previously we are focused on making 5v5 the best experience that it can be. However, because we operate as a service to our players, we always remain open-minded to re-evaluating our decisions based on your actions and feedback, to give you the best game experience we can.  Rather than introduce a disruptive, permanent change to the game, we'd prefer to explore the space further and make changes based on that exploration.

With that said, we're looking at running a series of events to try out different core team composition formats in Overwatch 2. The community has, juuuust once or twice, suggested a test. Why not put various forms of 6v6 in the game in order to gauge the results? We agree, and based on your feedback, we’re exploring how we can test different forms of 6v6 in the game to gauge the results. This is taking some time, however, for reasons that may not be readily apparent.

The first, and honestly largest, reason is the need to do this in a way that allows the game to have optimized technical performance (I’m referring to framerate and memory constraints) with 12 players in a match. There have been quite a few upgrades and additions to Overwatch 2 that have an impact on the performance of the game. These range from new features like allied outlines and enabling healer vision by default, the addition of more technically demanding hero kits, as well as visual upgrades to UI, heroes and maps. In a 6v6 setting, these upgrades can have significant impacts for our players, causing the game to underperform on older systems. Overwatch is a fast-paced game, and maintaining a game that runs smoothly across all our platforms is important for the player’s experience. While a limited time test could arrive sooner, the team is still investigating exactly how long it would take to permanently increase performance across the game. This would be a large effort that would most likely take at least several seasons to accomplish.

Another issue, and one that still gives me anxiety about opening this Pandora’s Box, is what to do with queue times if enough players prefer 6v6 to keep some form of it around. The team was not able to solve this issue previously. While we have some ideas now, there’s no guarantee that they will work. Is there a world where people are willing to live with long queue times to play this format? Maybe, but that’s a pretty risky move to make. We also have tens of millions of new players that have only played 5v5. We want to be openly mindful not to frustrate those who like the game as it is today.

The outcome of these tests would give us a lot of valuable information to consider moving forward. We’d run the test for a few weeks and gauge interest from the community. Then we’d measure the impact it had on the other experiences in Overwatch 2, like 5v5, Arcade, etc. We’d take the lessons from this playtest to see what we can learn about the mode within the current game ecosystem and for the future of Overwatch. We would reflect carefully on the learnings from whatever test we run and explore how to best give players what's being asked for. Whether that’s a world of 5v5, 6v6 or even both, is for future us to figure out.

I’m going to jump on a soap box for a moment. When you look at the changes to Overwatch since its inception, it’s clear that many of those have reduced some of the variety within an Overwatch match. We get feedback from some players that Overwatch can feel “the same” from game to game. While much of this gets attributed to 5v5, we feel that there is more at play here. Besides running experiments with 6v6, we’d like to run some that re-examine the ways we tried solving previous problems, specifically with the goal of bringing some of the freedom back to an Overwatch match without the severity of issues that accompanied it. For instance, we think there could be other ways of putting a team together that aren’t quite as rigid as a set composition, but not as loose as Open Queue. We’ll be running at least one Quick Play: Hacked based on this idea in Season 13. There are some gotchas to this direction, but after reading 400 pages of this blog, I hope you’re all able to see that there are trade-offs to many of the decisions that go into this game (and into any game, really).

To Wrap It Up… For Now

Your feedback has driven both our inspiration and commitment to testing new ideas in-game. We want to be super thoughtful about how we implement new ideas into the live game because there are millions of players who jump into Overwatch each season, and we want anything we put in-game to benefit the community as a whole.

Ultimately, the Overwatch team wants to continue building the game that you want to play season after season.

This blog (or thesis, rather) is just the start of this conversation. This isn’t the last time we’re going to be getting into 5v5 or 6v6, and we can’t wait to share more about our plans in the future.

In closing, we appreciate you so much, and we love seeing your passion and dedication to the game, Team 4, and the Overwatch community.

The team hears you, we see you, and we’re incredibly thankful to be on this ride with all of you. Can’t wait to share more in the upcoming months, and let’s make a great game, together.

 - Aaron Keller