WoW Factor: Updating data in the wake of Battle for Azeroth’s release announcement


Hooray, we have a release date for World of Warcraft: Battle for Azeroth instead of just a release window! And contrary to what many skeptics (myself included) expected to get, it is actually quite a bit faster than other releases. But as you all have no doubt noticed by now, my love of math means that I’m hardly sore about this. It just means that there’s another data point to consider when we look to the future.

So let’s talk about this new piece of information while using the same information from the column in which I made a reasonable estimate, based on this new information. Again, I think it’s important to note how much faster this expansion is actually releasing compared to prior expansions; it’s significant, even if it means that the people predicting things like June were being wildly wrong about “optimistic” predictions. (After all, pessimistic predictions were equally wrong, just in the other direction; my own estimates were off by 2-3 months.)

Announcement to release

The Burning Crusade: Announced 10/28/05, Released 1/16/07 (13 months)
Wrath of the Lich King: Announced 8/3/07, Released 11/13/08 (14 months)
Cataclysm: Announced 8/21/09, Released 12/7/10 (13 months)
Mists of Pandaria: Announced 10/21/11, Released 9/25/12 (11 months)
Warlords of Draenor: Announced 11/8/13, Released 11/13/14 (12 months)
Legion: Announced 8/6/15, Released 8/30/16 (12 months)
Battle for Azeroth: Announced 11/3/17, Released 8/14/18 (10 months)

Whoa! Right away, that’s a surprising number if you look at everything taken together. Battle for Azeroth is officially the absolute fastest turnaround time out of every expansion in the game’s history to date, edging Mists of Pandaria out by a month (another expansion that had no small amount of people being concerned that it was a quick, throwaway update to the game). This doesn’t change the averages in any significant fashion, but it does mean that a 10-month turnaround is now plausible enough that “11-13 months” is a reasonable expectation for future releases, rather than an outlier.

We’ll likely never know exactly why the turnaround is different, although I can certainly make guesses. We also can’t know if this is going to be a release like Cataclysm, which was clearly pushed out well before it was ready; of course, Cataclysm was coming off of a huge content gap that did nothing good for its reception, while Legion doesn’t have players anxiously awaiting its ending with the same degree of fervor. Regardless, it’s a surprising deviation from what had been fairly uniform dates up until now.

Take aim.

Expansion lifespan

Vanilla: Started 11/7/04, Ended 1/15/07 (26 months)
The Burning Crusade: Started 1/16/07, Ended 11/12/08 (22 months)
Wrath of the Lich King: Started 11/13/08, Ended 12/6/10 (25 months)
Cataclysm: Started 12/7/10, Ended 9/24/12 (21 months)
Mists of Pandaria: Started 9/25/12, Ended 11/12/14 (25 months)
Warlords of Draenor: Started 11/15/14, Ended 8/29/16 (21 months)
Legion: Started 8/30/16, Ended 8/13/18 (24 months)

You wouldn’t be wrong to argue that Legion has a lifespan closer to 23-and-a-half months, but that’s a point of fine argument that just makes life way more annoying than it needs to be. And the funny thing is that this is exactly what the averages originally predicted, a timescale between 23 and 24 months for a lifespan. I had, at the time, focused on the gap between “long” and “short” expansion life cycles, but once you ignore that the averages line up perfectly.

Thus, this information actually looks more accurate for future prediction purposes. It’s obviously not foolproof, but its mean and median both wind up at the exact same points, so I think this one also deserves far more consideration than it tends to get. Most predictions tend to focus on content gaps or the like; lifespans and announcement times seem to produce more accurate information.

Do we have a word for when you didn't like data before and like it even less now?

Beta timescale

Here we have to be a little bit more speculative, because the beta has not yet actually started; we’re still in the alpha. I’m going to go ahead and predict that we’re getting beta either this month or in early May based on the “soon” promise, but we will still have a little wiggle room.

Vanilla: Beta March 2004, Launched November 2004 (8 months)
The Burning Crusade: Beta October 2006, Launched January 2007 (3 months)
Wrath of the Lich King: Beta July 2008, Launched November 2008 (4 months)
Cataclysm: Beta June 2010, Launched December 2010 (6 months)
Mists of Pandaria: Beta March 2012, Launched September 2012 (6 months)
Warlords of Draenor: Beta June 2014, Launched November 2014 (5 months)
Legion: Beta May 2016, Launched August 2016 (4 months)
Battle for Azeroth: Beta April/May 2018, Launched August 2018 (5-4 months)

Hey, look at that, both versions line up perfectly with the averages and the overall length of beta test periods. So it’s a bit more speculative than the metrics elsewhere in the column, but gosh, it works just fine. It also means that the game will have a roughly analogous beta timetable to Legion, which seemed to work out just fine.

We had more data there, though, gauging the time between announcement and beta as a useful measure of when beta was likely to start…

The Burning Crusade: Announced October 2005, Beta October 2006 (12 months)
Wrath of the Lich King: Announced August 2007, Beta July 2008 (11 months)
Cataclysm: Announced August 2009, Beta June 2010 (10 months)
Mists of Pandaria: Announced October 2011, Beta March 2012 (6 months)
Warlords of Draenor: Announced November 2013, Beta June 2014 (8 months)
Legion: Announced August 2015, Beta May 2016 (10 months)
Battle for Azeroth: Announced November 2017, Beta April/May 2018 (6-7 months)

Still all over the map there. You can suss out a bit of a pattern if you really try, but it doesn’t seem to be terribly predicative. (That’s certainly possible, for the record; it might just be difficult as hell to figure out when beta testing is going to start working from the announcement.) Announcements to betas aren’t even moving in any consistent pattern.

Me auctions sold!

Last patch/last raid to expansion release

Ah, the classic metric that everyone uses!

The Burning Crusade: Last patch 8/22/06, Released 1/16/07 (5 months)
Wrath of the Lich King: Last patch 3/25/08, Released 11/13/08 (8 months)
Cataclysm: Last patch 12/8/09, Released 12/7/10 (12 months)
Mists of Pandaria: Last patch 11/29/11, Released 9/25/12 (10 months)
Warlords of Draenor: Last patch 9/10/13, Released 11/13/14 (14 months)
Legion: Last patch 6/23/15, Released 8/30/16 (14 months)
Battle for Azeroth: Last patch 8/29/17, Released 8/14/18 (12 months)

Is this all over the map still? Yep. However, we do have slightly more uniform data with the newest release date. A year from the last patch isn’t set in stone, but it seems to be closer to the way to bet than anything else. And what about the last raid?

The Burning Crusade: Last raid 6/20/06, Released 1/16/07 (7 months)
Wrath of the Lich King: Last raid 3/25/08, Released 11/13/08 (8 months)
Cataclysm: Last raid 12/8/09, Released 12/7/10 (12 months)
Mists of Pandaria: Last raid 11/29/11, Released 9/25/12 (10 months)
Warlords of Draenor: Last raid 9/10/13, Released 11/13/14 (14 months)
Legion: Last raid 6/23/15, Released 8/30/16 (14 months)
Battle for Azeroth: Last raid 11/28/17, Released 8/14/18 (9 months)

Well, that rather skews some information, doesn’t it? But I think it speaks to a rotation of philosophy for development that’s closer to what informed the team back in the earlier days, offering the last raid of the expansion as a part of lead-up content rather than a preamble to same. Instead of having nothing to do for a long chunk of time, players had something to do pretty consistently through the expansion lull, even if it didn’t work out to another numbered patch.

I can only hope that you find this aspect of analysis and prediction as fascinating as I do. If not, well, stay tuned next week for a much less data-driven column. Until then, you can tell me how much you hate it down in the comments below or via mail to I think it’s neat.

War never changes, but World of Warcraft does, with a decade of history and a huge footprint in the MMORPG industry. Join Eliot Lefebvre each week for a new installment of WoW Factor as he examines the enormous MMO, how it interacts with the larger world of online gaming, and what’s new in the worlds of Azeroth and Draenor.
newest oldest most liked
Subscribe to:
Randy Savage


Sally Bowls

Yes, I do find this fascinating and appreciate the effort. Well done and thanks.

Without disagreeing with the above, I see some minor additional issues.

1) Calendars – not just months -matter. So, IMO, dates will probably be slid a month either way to avoid times like Christmas, New Years, Confederate Memorial Day, Blizzcon, et al.

2) I could be wrong but I think the realism on annual releases helped. A number of games try for an annual release. And there would be advantages to smaller, annual releases. And Blizz devs said they were trying for that. Recently, they said they realized they were never going to achieve that. Recognizing and planning for that, along what is certainly the world’s largest dev budget, seemed to have allowed them to improve. E.g., I think WOD would have been much better received if Legion were ready sooner and the WoD content was spread out over 9 fewer months.

3) Filler matters and is not reflected in dates. E.g., 11.5 months from last raid to launch where there is nothing new would feel worse to me than 12 months but with things like Allied races. Bliz is going away from the Sunwell/Ruby Sanctum filler raids, which is probably wise. But a question is how much do you delay the next release (TANSTAAFL) in order to provide more intermezzos?

Jaymes Buckman

How do you equate Sunwell and Sanctum?

Sally Bowls

I am not sure I did so accurately, but I seem to recall the rumor that SWP was added because TBC was taking longer than anticipated. And most considered RS as filler rather than strategic content.

Jaymes Buckman

I think that that rumour started because Illidan felt thematically like the end boss to some, but the Sunwell was far too well developed to be made in haste. That whole patch was massive. No argument on the Sanctum.

Patreon Donor
Kickstarter Donor
Loyal Patron

One odd point is that the expansions that introduced new classes (Wrath, MoP, Legion) consistently had longer lifespans and were preceded by shorter-lived expansions than the ones that did not.

The apparent alternation between “long” and “short” expansions could be an artifact caused by a lot of semi-random factors and changes made by the team over time in how they plan and work on expansions through a decade and a half.

However, it’s seems there is at least some information there that we can use to speculate on what the development lead times say about the relative “cost” of a new class vs. other things. It does seem odd that expansions including an addition that we assume is costly to make vs. more routine content consistently seem to take less time to build out than others that do not include that feature.

Kickstarter Donor

Have to pick on Eliot for some unknown reason. :P

Loyal Patron
Patreon Donor

I still have a lot to do so I was hoping Eliot’s original date was right.

Dug From The Earth

This almost… ALMOST seems like a really lengthy, wall of text way to say “Hah, I told you so”


…or least an overidentification of such. o.O