What Is a Feature Factory?
John Cutler coined the term "feature factory" to describe product teams trapped in a cycle:
- Stakeholders request features
- Teams build features
- Features ship
- Nobody measures if they worked
- More features get requested
The feature factory feels productive. You're shipping constantly. Roadmaps fill with checkmarks. But none of it connects to outcomes that matter.
Signs You're in a Feature Factory
1. Success is measured by output, not outcome
- "We shipped 12 features this quarter" (output)
- vs. "We improved activation by 15%" (outcome)
2. Roadmaps are lists, not strategies
- Features grouped by timeline, not by goal
- No clear thesis connecting items
3. Research happens after decisions
- Features get committed before validation
- Research is for "confirming" not discovering
4. Everyone is a PM
- Sales drives features for specific deals
- Executives have "ideas" that become requirements
- The highest-paid person's opinion wins
5. Nothing gets killed
- Bad features stay because removing them is harder than keeping them
- The product bloats over time
- Complexity grows faster than value
Why Feature Factories Form
Feature factories aren't caused by bad people. They're caused by:
Misaligned incentives
- PMs evaluated on shipping, not impact
- Engineers measured by velocity, not value
- Sales compensated on deals, not retention
Short-term pressure
- Quarterly targets demand visible progress
- Board meetings need "new" announcements
- Competitive threats feel urgent
Research avoidance
- Research takes time; building is faster
- Negative findings threaten existing plans
- "We already know what customers want"
Measurement failure
- No baseline to compare against
- No tracking of feature usage post-launch
- Success defined by launch, not adoption
Escaping the Feature Factory
The exit path isn't about doing research (though that helps). It's about changing what counts as success.
Shift 1: From features to outcomes
Replace: "Ship invoicing feature" With: "Increase enterprise retention by reducing billing friction"
The feature becomes a hypothesis, not a requirement. If it doesn't achieve the outcome, you try something else.
Shift 2: From roadmaps to bets
A roadmap implies certainty. A bet acknowledges uncertainty.
Each item should state:
- What we're building
- Why we believe it will work
- How we'll know if it worked
- When we'll evaluate
Framing as bets invites learning instead of defending decisions.
Shift 3: From stakeholder requests to validated problems
Stakeholder requests are solutions (often bad ones). Your job is to find the underlying problem.
When someone says "build X":
- What problem are they trying to solve?
- Who else has this problem?
- What evidence supports this priority?
- Are there other solutions?
Shift 4: From launch-and-forget to learn-and-iterate
Every feature should have:
- Success metrics defined before launch
- Tracking implemented before launch
- Review scheduled after launch
- Decision criteria for iteration or removal
If you never evaluate, you never learn.
Practical Tactics
Tactic 1: Outcome-based planning
Start planning with outcomes:
- "What business results do we need this quarter?"
- "What customer behaviors would drive those results?"
- "What changes to our product might enable those behaviors?"
Features emerge from outcomes, not the other way around.
Tactic 2: Feature audits
Quarterly, review:
- Which features shipped 6+ months ago?
- What % of users actively use them?
- Which should be improved, maintained, or removed?
Tactic 3: Bet reviews
Monthly, review active bets:
- Are metrics moving as expected?
- What have we learned?
- Should we double down, pivot, or stop?
Tactic 4: Saying no with data
When stakeholders request features, respond with:
- "Here's what our data shows about this problem."
- "Here's how we'd measure success."
- "Here's what we'd need to deprioritize."
Data shifts conversation from opinion to evidence.
The Cultural Shift
Escaping the feature factory requires changing what gets celebrated:
Old culture:
- "We shipped on time!" (even if nobody uses it)
- "The customer asked for this!" (one customer)
- "We need to stay competitive!" (copying others)
New culture:
- "We moved the metric!" (outcome achieved)
- "Research showed this pattern." (evidence-based)
- "We killed a feature that wasn't working." (learning)
The feature factory dies when shipping is no longer the goal. Impact is.