When continuous deployment backfires // Intentional churn & re-engagement // In which I finally talk about videogames // Part 1 of many

February 2, 2022 · writing

One of the most startling videos about business that I’ve ever seen was actually about a video game.

The game is nine years old, the video is three years old, and I still think about both of them constantly.

The video is long (and good, if you’re into that sort of thing), but I’ll summarize below:

https://www.youtube.com/watch?v=tmuy9fyNUjY

Hm, okay, there’s a lot in there.

Here’s a counterintuitive appetizer:

As good engineers, they (obviously) began their operations with continuous deployment (i.e., small, incremental improvements released every day).

Can you guess what happened? Massive churn:

Can you guess what solved it?

Intentionally delaying new improvements into 13-week (3-month) batches:

Whaaaaat!? * If you’ve spent any time in startups, this is a real mind-bender. Not just a little “hmm” bender, but a profoundly existential, rethink-everything-I’ve-ever-learned sort of revelatory apocalypse. This chart violates everything we’ve ever been told about *everything throughout the Lean decades.

Cool, so what’s happening here?

First, PR. The press won’t write about incremental everyday improvements. Everyday progress is a non-story. But they will write about big exciting batches of improvements. (That’s a values tension: iteration speed vs. PR uplift, and almost makes you think that startup advice shouldn’t be given quite so generally.)

Second, empathy and understanding for the customer journey. For many businesses, daily engagement is a vanity metric. (For an easy example, think about your laywer, physiotherapist, or plumber — you pay them not to see them — and then extrapolate to tech.) Just because you want daily engagement (as a business owner) doesn’t mean that your customers want it.

Rather than fighting this fact, GGG (the game developer in question) has embraced it. Here’s one of the most remarkable statements I’ve ever seen from a business (full quote below):

Here’s the money-quote (20:10):

The philosophy here is **players will stop playing your game.

That’s unavoidable.** Something’s gonna happen that makes them stop playing. They get a new job or whatever, they get burned out, a new game comes out.

The key is to give them a lot of opportunity to re-engage, and that means they need to really clearly know when they’re next coming back.

It is inexcusable for a player to be able to leave your game without knowing the date that they are returning to it.

And so that’s why we want to make sure that users quit with a plan to come back.

Before we clearly signaled when the next release was, players would get horribly burned out and then they would just never play again. Whereas now, they can quit before they get burned out, and that’s amazing for us, because then users are fresh [when they come back]. 

Have you ever seen a healthier long-term business mindset? A more humble and empathetic view toward the customers’ realities and journeys? A stronger willingness to stand against short-term vanity metrics? 

To summarize the above:

1 // For their particular business, continuous deployment was a company-killing error. Not because continuous deployment is fundamentally flawed, but because all “best practice” advice must be put into context.

2 // By batching their feature releases onto a consistent and predictable schedule (every 13 weeks, in their case), they were able to unlock both the support of PR and the planned return of previous players.

Or in other words: by violating widely accepted technical best-practices and ignoring weekly/monthly retention (i.e., the golden standard of KPIs), they built a massive growth engine.

Makes you think, huh? 

Anyway, I’ll need to do a couple sequels to this post (at least 2 more), since I didn’t even get to the point I originally wanted to make about this video. 

We’ll get there. In the meantime, play more videogames.

🎮📺

← all writing