The team that shipped and the team that didn’t
When Maya Smith arrived at Brindlepath, the company had two product teams of about the same size and seniority. One shipped steadily, recovered quickly from mistakes, and had a reputation with customers for listening. The other shipped in bursts, argued constantly, and had lost two engineers in six months.
Maya’s first instinct was to look at skills. The struggling team had good people: a strong product manager, an experienced designer, talented engineers. What it lacked was harder to see. In meetings, the engineers rarely questioned the product manager’s requirements, then quietly built something different. The designer learned about features after they had been specified. Nobody admitted a mistake until it reached a customer. Decisions were relitigated in the hallway after they had been made in the room.
The healthy team was not more talented. It was safer, clearer and closer. Its product manager, designer and engineering lead met with customers together, argued openly about what to build, and wrote down what they decided. When something broke, the first question was “what happened?” not “who did this?”
Maya spent more of her first year on the struggling team’s dynamics than on any feature. It was the best investment she made.
The roles on a product team
A modern product team is small, cross-functional and long-lived. It owns a meaningful part of the product, a problem area or a customer journey, and has most of the skills it needs to discover, build and run solutions without waiting on other teams.
The core is the product trio: a product manager, a product designer and an engineering lead.1 Together they own discovery. They talk to customers together, decide together what to test, and share responsibility for the outcome. Around the trio sit the rest of the engineers and, depending on the product, specialists in data, quality, content, security or clinical practice.
Each role has a distinct center of gravity:
The product manager is accountable for value and viability: that the team works on the right problems, that solutions create value for customers and the business, and that the work fits the strategy. The product manager is not the boss of the team, and does not own design or engineering decisions.
The product designer is accountable for usability and the experience: that customers can understand and use what the team builds, and that it feels coherent with the rest of the product. Great designers are involved from the first customer conversation, not handed requirements to decorate.
The engineering lead is accountable for feasibility and quality: that solutions can be built, run, secured and maintained, and that the team’s technical choices serve the long term. The best engineering leads shape what gets built by proposing simpler or more powerful approaches the product manager had not imagined.
Specialists bring depth the trio lacks. In healthcare, a clinical advisor, often a nurse or physician, can tell you in five minutes what would take five weeks of research to learn. A compliance partner involved early turns regulation from a late-stage veto into a design input.
The lines between roles are not walls. The best teams overlap generously: engineers who talk to customers, designers who understand the data model, product managers who can read a pull request. What matters is that accountability is clear, even when contributions are shared.
What makes a team effective
Between 2012 and 2015, Google studied over 180 of its own teams to find out why some were more effective than others. The researchers expected to find that the best teams had the best individuals, the right mix of skills or personalities, or the most experienced leaders. Instead, they found that how a team worked together mattered far more than who was on it.2 They identified five dynamics:
- Psychological safety: team members feel safe to take risks and be vulnerable in front of each other.
- Dependability: team members get things done on time and to a high standard.
- Structure and clarity: team members have clear roles, plans and goals.
- Meaning: the work is personally important to team members.
- Impact: team members believe their work matters.
Psychological safety stood out as by far the most important, underpinning the other four. The idea comes from Amy Edmondson’s research on hospital teams, which found that the best-performing units reported more errors, not fewer, because people felt safe to speak up about them.3 For a product team, psychological safety shows up as engineers who challenge a requirement before building it, designers who say “I don’t think this works,” and product managers who admit their hypothesis was wrong.
How a product leader builds safety
Psychological safety is not niceness. It is the confidence that you will not be punished or humiliated for speaking up with ideas, questions, concerns or mistakes. A product leader builds it through small, repeated behaviors:
- Go first on mistakes. Talk openly about your own wrong calls and what you learned. The team takes its cue from you.
- Ask, then listen. “What am I missing?” and “What would make this fail?” are the most valuable questions in product management.
- Separate the decision from the person. Critique ideas, not people. When something goes wrong, ask what happened in the system, not who is to blame.
- Thank dissent. When someone disagrees in a meeting, thank them, even when you still disagree.
- Close the loop. When someone raises a concern, tell them what you did about it, even if the answer is “nothing, and here’s why.”
Brindlepath’s struggling team improved within two quarters once its trio began meeting customers together and running blameless reviews after incidents. The engineers who had quietly built something different started saying so in the room.
Structure and clarity
Safety without clarity produces pleasant teams that do not ship. The other dynamics, especially structure and clarity, need deliberate work too.
Give each team a mission, not a feature list. “Make sure every high-risk discharged patient has a confirmed owner within 24 hours” gives a team room to think. “Build the ownership queue” does not.
Organize teams around customer problems or journeys, not technology layers. A team that owns “the navigator’s morning” can move faster than three teams that own the front end, the back end and the database, and must coordinate every change.
Make decision rights explicit. Who decides what to build, how it looks, how it is built, and when it ships? Write it down. A simple model is that the product manager decides what and why, design decides how it works for people, engineering decides how it is built, and the trio decides when it is good enough together.
Keep teams stable. Teams that stay together long enough to build trust and deep domain knowledge outperform teams assembled for each project. Move work to teams, not people to work.
Hiring and growing product managers
Hiring is the most important decision a product leader makes. A great product manager can make a mediocre team good; a poor one can make a great team miserable.
When I hire product managers, I look for five things above experience or industry background:
- Curiosity about customers. Do they ask about the people who use the product before they ask about the product?
- Clear thinking and writing. Can they explain a complex trade-off in plain language, on paper?
- Judgment under uncertainty. Given incomplete information, can they make a reasonable call and say what would change their mind?
- Influence without authority. Have they moved people who did not report to them?
- Ownership. Do they talk about outcomes they were accountable for, including the ones that failed?
The best interview exercises mirror the real work: a short written analysis of a product problem, a role-play of a difficult stakeholder conversation, a review of a real customer interview. Brain teasers and generic “design a product for X” questions tell you little.
Growing the team
Product management has few formal training paths, so most growth happens on the job. A product leader’s job includes building a development system:
- Clear expectations by level, so people know what “good” looks like as an associate, a product manager, a senior product manager and a leader.
- Regular coaching, focused on real decisions: reviewing a strategy document, debriefing a customer interview, preparing for a difficult conversation.
- Stretch opportunities with support: owning a new area, leading a launch, presenting to the board.
- A learning habit: shared reading, internal talks, and time for people to learn the domain deeply. In healthcare, that might mean shadowing clinicians every quarter.
Remote and hybrid teams
Most product teams now work partly or fully remotely. Google’s study found that whether teammates sat together had little to do with effectiveness. What matters is that the dynamics are deliberately supported.
Remote teams need more written communication, more explicit decisions, and more intentional connection. Write decisions down where everyone can find them. Use meetings for discussion and decisions, not status updates. Overlap working hours enough for real-time collaboration. Make customer conversations visible to everyone by recording them, with permission, and sharing highlights. And invest in occasional in-person time, especially for planning and for new team members.
Brindlepath in practice: turning a team around
The struggling team from the opening of Chapter 8 did not change because of a reorganization or a new hire. It changed through a handful of practices, introduced over two quarters.
First, the trio started meeting customers together. Every week, the product manager, the designer and the engineering lead joined at least one customer conversation, and spent fifteen minutes afterward agreeing what they had heard. Within a month, the engineers stopped quietly building something different from the requirements, because they had heard the same customers say the same things.
Second, the team adopted blameless reviews. After a production incident in which a configuration change hid patients from the queue for a morning, the review asked what in the system had allowed it, not who had made the change. The answer, a missing check in the release process, was fixed in a week. The engineer who had made the change led the review, which did more for psychological safety than any speech could have.
Third, the team wrote down its decision rights. A one-page agreement set out who decided what, and how disagreements would be escalated. Hallway relitigation dropped sharply once there was a clear place to raise a disagreement.
Fourth, Maya changed how she ran reviews. She stopped presenting conclusions and started asking questions: what are we missing, what would make this fail, what did we get wrong last time? She went first in admitting her own mistakes.
Six months later, the team’s anonymous survey scores on the five dynamics had risen on every one, most on psychological safety. Its delivery measures had improved too. The two engineers who had been considering leaving stayed.
Common mistakes
Hiring for résumé, not for the work. Impressive company names say little about whether someone can write clearly, make decisions under uncertainty and influence people they do not manage. Test the real work.
Handing requirements over the wall. A product manager who writes specifications and hands them to design and engineering loses the ideas those disciplines would have contributed, and their commitment.
Mistaking harmony for safety. Teams that never disagree are rarely safe; they are usually silent. Psychological safety shows up as open disagreement about ideas.
Reorganizing too often. Every reorganization resets trust and domain knowledge. Change team structures deliberately and rarely.
Bringing specialists in at the end. Clinical, security and compliance experts consulted only before launch can only say no. Consulted at discovery, they shape a better product.
Leaving growth to chance. Product managers who never get feedback or coaching plateau early. Development needs a system, not good intentions.
Try this
At your next team retrospective, ask everyone to answer five questions anonymously on a scale of one to five, one for each dynamic in Figure 8.2: Do I feel safe taking risks on this team? Can I count on my teammates? Are our goals and roles clear? Does this work matter to me? Do I believe it makes a difference? Share the averages with the team, without names, and discuss the lowest-scoring dynamic. Agree on one change to try for a month, then ask the same five questions again.
How AI changes this
AI joins the team as a tool, and sometimes as a teammate. Product teams increasingly work alongside AI assistants and agents that draft, analyze, test and monitor. The roles in Figure 8.1 do not disappear, but each role’s work changes. Engineers review more code than they write. Designers explore more options faster. Product managers spend less time on synthesis and status, and more on judgment and alignment.
Smaller teams can do more. A trio with strong AI tooling can prototype, test and ship at a pace that once required a larger team. That is an opportunity to keep teams small and focused, rather than a reason to cut them to the bone. The constraint that remains is human: the number of customer relationships, decisions and reviews a team can handle well.
New skills matter. Every role now benefits from knowing how to instruct, evaluate and correct AI tools, and how to tell good output from plausible output. Make this an explicit part of your development system, not something people pick up by accident.
Safety matters even more. When AI writes a first draft, people need to feel safe saying “this is wrong,” including when the draft came from a senior colleague’s prompt. And they need to feel safe admitting when they used AI and it failed. Teams that hide their AI use learn nothing from it.
Agents need owners. If your team deploys AI agents that act in production, someone must own each one: its instructions, its permissions, its quality and its failures. Chapter 17 covers how to lead teams that include agents.