75 Lessons Learned from Live Game Development

I conceived of and shipped live services games to 20 million players; here's what I learned.

May 25, 2023

In the course of making games for over 20 million players—based on popular IP like Game of Thrones and Star Trek—I learned a lot of lessons. This is just a few of them, although this includes some of the most painfully learned lessons.

BIG PRINCIPLES

Every "rule" in gamedev has a great exception
You probably won't be the exception
Hard: Shipping; Harder: Shipping fun game; Hardest: shipping fun that scales.
What few are willing to hear: making a great game is unreasonably hard, and therefore makes unreasonable demands of the team
Shots on goal (trying ideas, shipping, iterating) matters most

PHASES OF PRODUCT

You can only build 1 game vision; converge team & purge fiefdoms
0 to 1 is about shared vision
Data is everything in the Live Ops stage; velocity of insight -> product improvement is predictor of success

PEOPLE

Be wary of hiring people who harbor deep reservations about making money
Be very afraid if the team building the game isn’t playing it
Toxic / abusive team members suck
Cynical team members suck
Veteran devs are AWESOME, but beware of jaded veterans with a sense of entitlement.

HIRING

Many people will say nearly anything to land a job at a game studio
Hire optimistic people because game development is incredibly hard
Work with people who want to make a great game right now more than anything else in their life
Focus on “tell-me-about-the-time-you-did-X” questions (behavioral interviewing) and don’t waste time on hypotheticals

TEAM MIX

Great teams with experienced craftspeople are foundational
Adding fresh blood who need to learn/prove (e.g., interns) has a high ROI and better connects you to current culture
Live Ops teams who run great games are usually not right to build a NEW game

FEATURES

Smaller is almost always better than larger
But “large enough” is even more important
Replayability is gold
It is better to cut mediocre features

MARKETING

For mobile games: what Kotaku says doesn’t matter
Meta critic score is one of the worst-designed metrics that exist
App store featuring will not save you
The sins of design will not be countered by marketing spend

TOOLING & TECH

Tools need to be native to the user (Unity for programmers, spreadsheets for designers, website forms for PMs, etc.)

Manual processes suck
Don’t build backend yourself
Don’t build a 3d engine
Technology prowess doesn’t win

PRODUCTION

Games look like a mess until at least 90% done
People think they are close to 90% when it's actually more like 50%
It will take longer than you think

ITEMIZATION & ECONOMY

Good shop design is critical—needs as much attention as any other part of the game
Getting a rare item should be clear and awesome for the player
Don't cap the size of your economy / monetization
Pay attention to what people buy

UX

Show your game to people IN PERSON, and take lots of notes on what happens when they play
Pay attention to how performance scales as the size of data (inventories, etc.) grows

INNOVATION

Innovation is good for capturing new markets, attention to craft craft is good for capturing established ones
UX is often the difference between good and great
You can’t really “lean startup” a game because too much of it is based on a complete experience

NARRATIVE

Storytelling on mobile must connect with the audience within seconds
Narrative != Dialog
Narrative is represented through game systems
Humans find meaning through story

DATA

Instrument everything, you’ll probably want the data at some point
Make it easy to access data
Dashboards that don’t display actionable intelligence might be cool but aren’t useful

PLAYERS

Love your players. (Contempt for players is doom)
Players will speed run content faster than you expect
Your most valuable cohort will probably be the people who show up in the beginning.
Optimize single player experience before multi

DON'T PANIC

Don’t panic if things look like a mess for the first 80% of the development timeline
Panic if things look like a mess when you’re about to ship :)

CURIOSITY

Team members should be really curious about what successful games like yours have done before
Pay close attention If people on the team are experts on your target audience, and they don’t like the game
Pay close attention if more junior members of the team aren’t learning from the more senior members; all of the reasons this might be happening are bad
Be wary of know-it-alls.. unless they've shipped top-revenue games🙂

VELOCITY

Any feature that depends on more than 1 programmer will be exponentially harder and riskier to implement
The best features can usually be prototyped in days if not hours
Code in the fewest languages possible, ideally just one

LICENSES

Aligning incentives is critical
Licenses can break through noise when nobody else does it; when everyone does it, worse
Avoid over-licensed IP
The IP biz dev person’s incentive is usually short-term (annual bonus) your incentives are long term (decade+)
Walk away if licensor demands % of gross rather than a % of net (of UA)
Walk away if licensor demands unreasonably strong control over when/if you ship

MONEY

Debt for UA is like taking treasure from a pit of vipers — more ways to die than win
Biz models that work in largest games may not work to scale your much-smaller game
Be wary of backers long on money and short on gamedev experience
Backers who don’t understand games will lose hope at your darkest hour
Every game development process has a “darkest hour”
Your forecast will be wrong

SHIPPING

It is better to delay than ship before it is ready
But it is better to ship before you run out of money than not ship at all
If you can’t ship before running out of money, the culprit is probably that your scope was too great—not that the capital was too small

FINAL THOUGHTS

If some stuff on this is contradictory, that’s OK. Games contain multitudes.
If you haven’t failed in game dev you haven’t been at it long enough
When someone declares the future of games, it usually isn’t
Fear is the Mind Killer

Metavert Meditations

Discussion about this post