FROM THE BOOK
35 Common PM Mistakes
30 timeless mistakes across strategy, execution, and career. Plus 5 new AI-era mistakes reshaping the role today.
After 30+ years in product roles—from startups to Fortune 500s—I've seen the same mistakes repeated over and over. These aren't rare edge cases. They're the everyday errors that derail careers, frustrate teams, and destroy products.
But the role is changing. AI is reshaping what PMs do and how they fail. Some of the biggest mistakes today didn't exist five years ago.
How to Be a Top Product Manager explores both the timeless mistakes and the new ones, with real stories, frameworks, and practical solutions. Below is a preview of what you'll learn—30 classic mistakes plus 5 AI-era mistakes and how to avoid them.
PM Strategic Mistakes (1-10)
Wrong prioritization, unclear strategy, bad metrics
1. Building Features Nobody Asked For
You assume you know what customers need. You don't talk to them. You build it. They don't use it. Meanwhile, real problems go unfixed.
How to Fix It:
Talk to customers BEFORE writing specs. Find the problem first, then validate the solution.
2. Saying Yes to Everything
Every feature request goes into the backlog. No prioritization. No strategy. Your roadmap is customer-driven chaos, not product vision.
How to Fix It:
Create a prioritization framework. Say NO to 80% of requests. Focus on the 20% that matter most.
3. Ignoring the Business Model
You focus on features, not revenue. You don't understand unit economics. You optimize for vanity metrics instead of real business impact.
How to Fix It:
Understand how your company makes money. Link every feature to business impact: revenue, retention, LTV.
4. Prioritizing by Loudest Voice
The biggest customer screams. The executive demands. They get what they want. Meanwhile, systemic problems fester.
How to Fix It:
Use data and frameworks, not politics. Make decisions defensible, not convenient.
5. Confusing Activity with Impact
You shipped lots of features. High velocity. But revenue is flat. Churn is up. You measured the wrong thing.
How to Fix It:
Define success metrics BEFORE shipping. Link features to outcomes: retention, revenue, NPS—not just "shipped".
6. Shipping Without Measuring
You launched the feature. Now what? You don't know if it's working. You can't optimize. You never learn.
How to Fix It:
Setup analytics BEFORE launch. Know your hypotheses. Measure impact in first 2 weeks. Iterate based on data.
7. Overcomplicating Solutions
Simple problem. Complex spec. Three-month build. Should've been two weeks.
How to Fix It:
Constraint drives creativity. "Build it in 2 weeks" forces clear thinking. Start simple, iterate.
8. Copying Competitors Blindly
Slack has it. We should add it too. You don't know WHY they built it. You don't know if it works for them. You just copy.
How to Fix It:
Understand your customers. Build for them. Competitors are playing their game, not yours.
9. Assuming You Know the Customer
You've been in the industry 10 years. You think you know. You don't. Markets change. Customers evolve. Assumptions age.
How to Fix It:
Talk to customers every month. Run research sprints. Validate assumptions constantly.
10. Optimizing for Vanity Metrics
Daily active users are up! But LTV is down. Retention is falling. You optimized for the wrong metric.
How to Fix It:
Define north star metric aligned with business. Everything else supports that one metric.
PM Execution Mistakes (11-20)
Specs, communication, stakeholder management
11. Writing Specs That Confuse Everyone
50-page PRD. Nobody reads it. Everyone interprets it differently. Launch chaos.
How to Fix It:
1-pager max. Talk to the team. Document decisions, not details. Spec is alive, not carved in stone.
12. Skipping User Research
You're too busy to talk to users. "We know what they need." Probably wrong.
How to Fix It:
Block 4 hours/week for customer conversations. Talk to 5 users before committing to big features.
13. Over-Promising Timelines
Sales says "4 weeks." Engineering says "12 weeks." You commit to "6 weeks" to keep everyone happy. Now everyone's angry.
How to Fix It:
Estimate with engineering. Commit to what's realistic. Buffer for unknowns. Under-promise, over-deliver.
14. Micromanaging Engineers
You question every decision. "Why are you using that library?" You erode trust. Talented engineers leave.
How to Fix It:
Define the problem and desired outcome. Let engineers own the solution. Trust their expertise.
15. Ignoring Technical Debt
"Ship features now, fix later." Later never comes. Codebase rots. Everything gets slower.
How to Fix It:
Reserve 20% sprint capacity for tech debt. It IS a feature. Ignore it at your peril.
16. Poor Stakeholder Communication
Radio silence for 3 months. Then launch day. Everyone's surprised. Nobody's prepared.
How to Fix It:
Weekly updates. Share early. Get feedback often. No surprises on launch day.
17. Feature Factory Mentality
Shipping 15 features per quarter. None of them connected. No strategy. Just output.
How to Fix It:
Quarterly themes. Everything connects to strategy. Depth over breadth. Fewer, bigger wins.
18. Analysis Paralysis
More data. More research. More meetings. Never launching. Competitors ship while you analyze.
How to Fix It:
Ship the MVP. Learn in market. Iteration beats planning. Better to be 30% right quickly than 100% right late.
19. Accepting Scope Creep
"Just add this one thing." Then another. Then 5 more. MVP becomes massive. Launch delayed 6 months.
How to Fix It:
Define scope and lock it. "That's v2." Say NO. Protect the MVP from bloat.
20. Launch and Abandon
You shipped. You moved to the next project. Nobody's supporting the launch. Adoption dies.
How to Fix It:
Plan for post-launch. Marketing, support, monitoring. First month determines success, not launch day.
PM Career Mistakes (21-30)
Breaking in, growth, long-term strategy
21. Waiting for Permission to Lead
You think you're not "PM enough" yet. You wait for a title. Meanwhile, PMs who shipped own the conversation.
How to Fix It:
Own something now. Work on a feature. Own the outcome. Title follows actions, not vice versa.
22. Staying in Your Comfort Zone
Year 3 PM doing the same job as year 1. No growth. No challenge. Slowly becoming irrelevant.
How to Fix It:
Every year, take on something hard. New product line. New market. New skill. Stay sharp.
23. Not Learning Technical Basics
You don't code. You don't understand architecture. You can't have real conversations with engineers. They don't respect you.
How to Fix It:
Learn the basics. SQL, APIs, databases. You don't need to code, but you need fluency.
24. Burning Bridges with Engineering
You pushed them too hard. You ignored their concerns. They're looking for jobs. Your best engineer's on their way out.
How to Fix It:
Respect engineers. Listen to their concerns. Build trust. You need them more than they need you.
25. Ignoring Company Politics
You focus on product. You ignore politics. Meanwhile, your VP's favorite PM gets all the resources.
How to Fix It:
Understand the org. Build relationships. Don't play politics, but don't be naive about it either.
26. Focusing Only on Features
You shipped 50 features. Your retention is terrible. You never built habits. You never owned the full journey.
How to Fix It:
Own engagement, retention, LTV. Features are means to outcomes, not the goal.
27. Not Building Your Network
You need a job. Nobody knows you. You reach out cold to old contacts. Nobody replies.
How to Fix It:
Build network constantly. Coffee chats. Conferences. Online. Year 1, not when you're desperate.
28. Chasing Title Over Growth
You got promoted to Senior PM. Same skills. Same problems. Title doesn't make you better.
How to Fix It:
Chase growth, not title. New challenge beats new title. Titles follow growth naturally.
29. Avoiding Hard Conversations
Engineer's underperforming. Feature's not working. You avoid it. Problem gets worse.
How to Fix It:
Have the hard conversation early. Data-driven. Respectful. Fix problems before they compound.
30. Neglecting Data Skills
You can't write SQL. You can't analyze data yourself. You wait for analysts. Slow feedback loop.
How to Fix It:
Learn SQL and analytics. You don't need to be expert, but you need speed and self-sufficiency.
PM AI-Era Mistakes (31-35)
New mistakes in an AI-transformed product world
31. Using AI for Decisions Without Human Judgment
You asked Claude "what feature should we build?" It gave you an answer. You shipped it. Users hated it. You abdicated your job to a chatbot.
How to Fix It:
Use AI as a tool for thinking, not a substitute for judgment. Talk to customers. Trust your instincts. AI is a copilot, not the pilot.
32. Ignoring AI Hallucinations and Limitations
AI wrote your spec. It sounds confident. You didn't fact-check it. It contains fabricated features, wrong API details, impossible timelines.
How to Fix It:
Always verify AI output. Fact-check specs. Validate API endpoints. Understand where AI confidently lies. Never trust without verification.
33. Not Understanding AI Bias in Your Product
Your AI-powered recommendations work great for US users. For users from other countries, accuracy drops 40%. You shipped discrimination without knowing it.
How to Fix It:
Test AI features across demographics. Measure fairness, not just accuracy. Understand training data biases. Own the ethical implications.
34. Over-Relying on Prompt Engineering Instead of Real Understanding
You spent 2 weeks perfecting prompts. Now you have a janky spec. You think AI solved your problem. It just delayed discovering you don't actually know what you want.
How to Fix It:
Use AI to accelerate your thinking, not replace it. Start with customer understanding. Use AI for drafting, iteration, brainstorming—not discovery.
35. Not Testing AI Features Thoroughly Before Launch
AI features are different. They fail in weird ways. Silent failures. Hallucinating responses. Off-by-one errors in recommendations. You tested happy path. Shipped anyway. Disaster.
How to Fix It:
Test edge cases obsessively. Stress test AI outputs. Run canaries with real users first. Monitor hallucinations post-launch. Set guardrails.
Why AI-Era Mistakes Matter
The first 30 mistakes are timeless. They'll still be killing products in 2035. But mistakes 31-35 are unique to this moment. AI is democratizing product development, but it's also introducing new failure modes.
The PMs who thrive won't be those who use AI best. They'll be those who use AI wisely—amplifying human judgment, not replacing it. The biggest mistake is thinking AI solved the hard part. It didn't. Understanding customers, making trade-offs, and shipping with conviction—that's still the job.