I Analysed 62,000 Indie Developers. Here's What Actually Predicts Success

Main ideas / lessons

• Steam indie success is extremely unequal (power-law / “Extremistan”): a tiny fraction of games capture most sales.
• Survival in the market matters more than early expectations: many developers quit early; those who continue accumulate advantages.
• Power-law mechanisms imply a “stay in the game” strategy: compounding advantage, digital scale, and rare breakthrough events all reward persistence.
• Use structured risk allocation (“barbell strategy”) to avoid the “messy middle” where risk is high enough to hurt but reward is capped.
• Common dev practices (scope, niche validation, algorithmic visibility, etc.) align with power-law mechanisms—even if they’re not framed that way.
• Avoid metric-only thinking (McNamara/quantitative fallacy): numbers like sales matter, but you also need measures of player joy/community and intrinsic motivation.
• Ultimate driver: meaningful stamina—developers who love what they make are more likely to keep going long enough to reach upside.

Key concepts and supporting reasoning

• Pareto distribution / power-law pattern

  • Sales concentration follows a “power law” shape: a few games soak up most sales; everyone else gets by.
  • Analogy: Vilfredo Pareto found that roughly 20% of people own 80% of land.
  • Framing: Nassim Taleb calls such environments “Extremistan”—a small number of huge outliers dominate outcomes.

• Large-scale dataset

  • Analysis covers 62,466 developers who shipped at least one Steam game.
  • Focus is not only on sales, but also on retention/survival: whether developers make additional games.

Methodology / “instructions” presented

A) Long-term survival approach (“stay in the game”)

• Goal: increase the odds of being among the “survivors” who reach later releases and therefore a higher chance of hits.
• Mechanism-based logic
  • Compounding advantage
    • Ship a game → gain reusable code, mailing list, community, and knowledge.
    • Each additional game starts from a higher baseline.
    • Staying longer lets advantage compound.
  • Scale (digital goods)
    • Physical products have upside capped by time.
    • Digital copies cost ~nothing to deliver, so potential scales far more.
    • Winners can grow exponentially.
  • Rare events
    • Career-defining breakthroughs are unpredictable but real (e.g., a viral streamer, supportive community, hit game).
    • You can only capture them if you’re still producing when they happen.
• Practical implication
  • Don’t aim merely to “survive early”; aim to keep creating so you’re present for rare upside events.

B) Barbell strategy for resource allocation

• Core structure: like a dumbbell
  • Side 1: safe bets (keep the lights on)
    • Projects with low risk of financial ruin (near-zero).
    • Low risk of losing time/resources to something that prevents pursuing your dream.
    • Examples mentioned: contract work, and other lower-risk efforts (context-specific).
  • Middle: messy middle (avoid)
    • Projects with medium risk but mediocre reward.
    • Risk: they consume effort without meaningful chance of upside.
  • Side 2: moon shots (uncapped potential)
    • High-variance bets where only one success is enough.
    • Allocate ~10–30% of resources to these (rule of thumb described).
    • Worth attempting because upside is not capped—as long as safe bets cover failures.
• Time horizon strategy
  • Use safe income to fund extended development cycles:
    • The speaker describes funding 6–12 month stints working on a moonshot, then returning to contracts if money runs out.
• Explicit caution
  • This is the speaker’s interpretation; others may structure it differently (e.g., more safe work + evenings on risky bets).

C) Game-building strategies connected to power-law outcomes

Framed as ways to align with compounding/scalability/network-effect mechanisms:

• Manage scope / smaller projects
  • Smaller games reduce financial risk and let skill compound faster.
• Validate niches / specialize
  • Specialization builds cumulative advantage with a targeted audience.
• Maximize algorithms
  • Strong market research and positioning can drive visibility, potentially leading to network effects/virality.

Data-driven claims and “what it means”

• Sales concentration

  • ~5% of games account for ~90% of sales (as stated by the speaker).

• Developer continuation rates

  • Only one in five developers (≈20%) went on to make another game.
  • Survival curve logic:
    • For developers making a second game: roughly 60% chance they won’t make another (i.e., quit before the third).
    • For later stages (third → fourth): quitting drops to around 20–30% chance of not making another.
  • Interpretation: hazard of quitting is highest early; continuing longer reduces quitting likelihood.

• Linking survival to “hits”

  • Define a “hit” as ~25,000 copies (speaker approximation of ~quarter-million revenue for a mid-price game).
  • Estimated odds:
    • First game: about 1 in 10 to hit that goal.
    • By the fifth game: closer to a coin toss.
  • Survival bias acknowledgement:
    • The low overall probability is partly because most people never reach game 5.
    • The speaker argues survivorship bias is not a flaw—it’s the point: success accrues to those who stay.

Examples / sources of long-game developers mentioned

• Sokpop (Dutch “video game boy band” concept)

  • Ships small games under a shared brand + Patreon; spreads risk across many releases.
  • Eventually had a major hit: Stacklands.

• Gugonix

  • Built to roam as a hobby alongside a day job.
  • After modest success: lived frugally, reinvested carefully → became a sustainable full-time career → Shell Diver.

• Bite Me Games and Andy from Arenas

  • Ran content/community for years before breakthrough.

• Clockwork Games

  • Learned from two quiet releases, pivoted to co-op, validated community → success with In Sync.

Closing lesson (philosophical / motivational)

• McNamara fallacy / quantitative fallacy

  • Named after Robert McNamara; the military emphasized enemy casualties as a main metric.
  • Numbers looked favorable, but they didn’t reflect what truly mattered (morale, local support, lived reality).

• Application to indie dev

  • Sales and other hard metrics are useful because they’re easy to measure.
  • But metrics can’t fully capture:
    • player joy,
    • community value,
    • emotional impact (e.g., a review saying the game cheered someone up).

• Final claim

  • Developers who go the distance share a key trait: genuine love/meaning in what they’re making.
  • This meaning creates stamina that data alone doesn’t capture.
  • Therefore: use data strategically, but don’t let it become the only driver.

Speakers / sources featured

• Ross (speaker): data scientist; analyzed the Steam developer dataset; main presenter
• Vilfredo Pareto: Pareto distribution (the 20%/80% observation)
• Nassim Taleb: framing of “Extremistan” and risk under uncertainty; also referenced the barbell strategy idea
• Robert McNamara: namesake of the McNamara fallacy; metric obsession example (Vietnam War era)
• Game Oracle: referenced platform/tool for market research/analysis
• Mentioned developer/publisher names
  • Sokpop, Gugonix, Bite Me Games, Andy from Arenas, Clockwork Games
• Game titles referenced
  • Stacklands, Shell Diver, In Sync