50/50 Track Record: Why Big Companies Fail at Innovation
HipChat: 3-year obsession → sold to Slack
Stride: Rewrite disaster → killed
StatusPage: Dead on arrival → acquired instead
Jira Product Discovery: 4-year bet → 8,000 customers in year one
"The company has a tendency to over-invest. You need to create scarcity. Let's not drag the rest of the company into it."
The Core Problem
Why Success Metrics Kill New Products
$100M
success bar at Atlassian
300K
existing customers to sell to
6 mo
patience window
The metric trap: New products judged by mature product KPIs (monthly active users, growth rate)
The reality: Early stage products should have low MAU—they're not ready yet
The consequence: Team gets shut down before product finds fit
The solution: Separate success metrics for Wonder/Explore/Make/Impact stages
The asymmetryA startup that fails learns from that. A big company that fails looks incompetent. So big companies often stop things too early.
Point A Framework
How to Protect New Bets Inside Large Orgs
Wonder: Prove there's a real market problem. Validate that Atlassian should care. Must answer "why now?"
Explore: Test solutions with customers—not building, just validating with prototypes and mockups
Make: Build the MVP. Team expands. First production code. Real metrics begin.
Impact: Scale and commercialize. Move out of incubator. Becomes a real business unit.
JPD timeline
Year 1-2: Wonder & Explore (solo, then 3 people). No code written.
Year 3-4: Make phase (alpha/beta, internal dogfooding, 8,000 customers at launch).
One year after launch: Already one of fastest-growing products in Atlassian history.
The Point A advantage
Psychological safety (can borrow people, no firing). Access to 20+ internal teams. Customer research pool. Analyst relationships. 3-6 month reassessment windows, not quarterly.
The Scarcity Move
Frame Failure to Win Autonomy
Tell leadership: "70% chance this doesn't exist in 6 months"
Stay small (7 people, not 50) to maintain speed
Refuse process help: "We're a bet. Let's not drag the rest of the company in"
Build hacky prototypes: "Not going to scale, not respecting design guidelines—but testing fast"
No planning sessions for resources; iterate and test with customers next week
Why this worksWhen everyone knows "this might fail," help comes with fewer conditions. Teams avoid gold-plating. You get runway to test truth, not process.
Contrarian
What Big Company Innovation Gets Wrong
✗Rewrite the product to competeINSTEAD →✓ Never rewrite. HipChat/Stride failed because rewrite took 3 years; Slack won in the meantime.
✗Bundle new products with existing onesINSTEAD →✓ Let new products stand alone. Bundling (like MS Teams + Office) kills competition; you can't force adoption this way.
✗Measure early bets by mature KPIsINSTEAD →✓ Ask different questions at each stage: Wonder/Explore are about learning, not growth.
✗Give new bets all the company resourcesINSTEAD →✓ Create scarcity. Small teams move faster. Big teams slow down. Earn help by proving concept first.