Shirley Zhang, Chenxi (Rita) Xiang, Alexis Tuchinda
Team Norms
We all agree to:
- Show up to planned meetings on time or say in advance when we cannot make it
- Finish our parts by the internal deadline we agreed on
- Ask when we feel like we need help
- Raise concerns or disagreements early and directly with the team, not behind anyone’s back
- Keep project discussion in our shared group chat so no one misses decisions
- Reply to team messages within 24 hours
- Divide work fairly and rebalance when someone’s week gets busy
- Make sure everyone has a chance to speak in meetings before we make a decision
- Check in on how the team is working every couple of weeks
New questions:
Rita:
Rita CS177 Week2 Interview Questions & Screener
- Looking across all your experiences, what is the most important lesson you have learned as a PM?
- Has your understanding of what makes a great PM changed over time?
- Looking back, what do you wish you had learned or prepared before your first PM internship?
- For students who want to enter product management, what skills or mindsets would you recommend they start developing now?
- Which parts of the product are you directly responsible for? And which cross-functional teams—such as engineering, design, data, user research, or marketing—do you collaborate with most closely?
Shirley:
Shirley Screener and Interview Questions
- When you have many competing requests, meetings, or opinions, how do you decide what to prioritize?
- Can you tell me about a recent situation where you had to balance competing priorities or push back on an initiative?
- How do you create clarity for the people working on a project, especially when the requirements or constraints are still unclear?
- How do you balance defining the problem or requirements with actually moving into implementation?
- How do you help technically oriented teammates consider multiple possible solutions instead of immediately jumping to implementation?
- What information do you look at when trying to understand why a particular product decision or initiative is important?
Alexis:
- What are the most difficult, frustrating, or inefficient parts of your workflow right now?
- Walk me through the last time this happened.
- How often does this happen?
- Walk me through the last time you used AI for work.
- Overall, where has AI made your workflow easier, and where has it created additional work or difficulty?
- Is there a problem you encounter frequently but have essentially accepted as part of the job?
- Is there a problem that happens less often but is particularly costly when it does?
Alexis Screener & Interview Questions
Affinity Mapping (Human)
- Background of the PM affects how they approach/ what limitations they have/ cross-functional
- PM roles are increasingly unclear because roles have to combine engineering, design, and business responsibilities
- Different tracks, including platform-centered roles, B2B/B2C roles, feature-centered roles, and strategy-centered roles.
- Non-technical PMs don’t have enough background to properly audit what AI tools generate… Therefore, there is a limit on how much interviewees are willing to let AI do.
- There is some expectation that product managers become a little more technical to double-check AI work
- Technical fluency changes the quality of PM decision-making. Both interviewees showed that understanding APIs, frontend/backend constraints, model behavior, data pipelines, and system architecture helps PMs diagnose problems faster and communicate more precisely with engineering.
- Need to make decisions with limited information
- “Make effective decisions with incomplete information”: develop a broad understanding of how different parts of the product and organization fit together
- Cross-functional communication becomes harder as companies grow; information gets lost across organizational layers.
- How to reduce unnecessary complexity: Tara repeatedly emphasized starting with fewer features and learning from real usage, while Tianrun described simplifying complex AI workflows into clear stages. A strong PM is not necessarily the person who proposes the most features, but the one who knows what can be removed.
- Scale affects how much impact/ who they need to be in charge of
- Too many meetings can become overwhelming and leave little time for focused thinking or execution.
- New employees may struggle to understand why certain initiatives are prioritized over others.
- Main challenge of getting people to do things – “herding cats”
- Incentive challenges – tradeoff between the effort of doing the work versus the capacity to get promoted/paid
- Giving teams a blank canvas can make decision-making hard, and it is much better to give them a few options to choose from instead
Affinity Map (AI second pass)
- Getting things done without authority (“herding cats”): 5 of 8: Terry, Tamara, Tara, Tianrun, Madhuri
- PMs can’t direct engineers. They persuade, negotiate the lowest-cost option, and follow up personally because assigned work silently drops.
- Unclear ownership tends to surface near deadlines.
- Incentives are misaligned: people pick work that helps them get promoted, not what’s best for the product.
- Workarounds: explain the reasoning behind the work, offer engineers 2–3 options instead of a blank canvas, and check every piece of the project.
- Meetings and ops crowd out focused, future-looking work: 3 of 8: Madhuri, Terry, Tamara
- “Meeting hell” grows with company size, and PMs end up spending most of their time on context gathering and follow-ups.
- Workarounds: “conceptual Thursday,” 5-minute and 30-minute task blocks, and declining meetings where they won’t contribute.
- Unclear problem definition and fragmented context: 4 of 8: Madhuri, Young, Dave, Terry
- Requirements are often unclear, and teams start building before they agree on the “why.”
- Lack of definition causes friction between product and engineering.
- Workaround: map the business, design and technical context, plus the people involved, before deciding.
- Deciding under uncertainty: prioritizing and knowing what not to build: 6 of 8: Terry, Tara, Tianrun, Tamara, Dave, Madhuri
- PMs have to balance feasibility, customer value, and whether anyone will pay.
- Scope grows without adding customer value, and leadership’s priorities can be hard to read.
- Workarounds: ship the smallest valuable version, focus on the top 20% of work, and cut features.
- AI output outpaces verification: 5 of 8: Kara, Tamara, Young, Tianrun, Terry
- Nontechnical PMs can’t audit AI-generated work, so there’s a limit to how much they’ll let AI do.
- “AI slop” and vibe-coded tech debt pile up.
- AI agents behave differently from run to run, so evaluating them becomes the PM’s job.
- Workarounds: build structured evaluations and become more technical.
- Build speed outruns customer feedback and validation: 4 of 8: Young, Dave, Tamara, Terry
- AI makes prototyping cheap, but customer access, feedback, and organizational processes stay slow.
- Getting customers is especially hard in healthcare.
- Turning feedback into requirements takes real effort.
- Workarounds: rely on personal networks (LinkedIn plus a CRM) and use AI prototypes to reject weak ideas early.
- The PM role is ambiguous and increasingly technical: 6 of 8: Madhuri, Tara, Tianrun, Kara, Terry, Tamara
- The role varies by company, size, and track (platform, B2B/B2C, feature, strategy).
- Startup PMs cover UI, sales and coding when no one else can.
- Technical fluency improves how well PMs diagnose problems.
- Communication and ownership break down at scale: 3 of 8: Madhuri, Terry, Tamara
- Information gets lost across organizational layers.
- Large-company processes and risk aversion reduce PMs’ ownership; Tamara called it “soul-crushing.”
AI second-pass comparison
What did it catch that we missed?
The AI consistently paired each problem with the person’s workaround, such as “conceptual Thursday,” the three-option framing, declining meetings, and structured evaluations. Those workarounds are what the assignment asked for, but also change some of the framing of the issues to be solution-related. It also surfaced themes our human maps skipped: AI pricing and token costs (Young, Kara, Terry), AI adoption moving faster in product and engineering than in the rest of the organization, and the overall tension that AI speeds up building but not validation.
What did it flatten into something generic?
“Cross-functional collaboration” lost the specifics. Tara described negotiating with frontend and backend for the lowest-cost solution, and Tianrun described a power balance that shifts depending on whether the PM sits closer to the model layer or the application layer. Terry’s frustration as an early-career PM with no authority was split into several neutral-sounding items. The AI over-split instead of synthesizing. Alexis’ version lists 26 problems, many overlapping (AI slop vs. tech debt vs. prototyping noise), with no clear distinction of which ones hurt most.
Did it assert anything that isn’t in the transcripts?
Mostly no, but some claims go beyond the evidence. “Execution toward judgment” and “learning velocity” are interpretations; Tianrun didn’t say them. Some items in Alexis’ extraction were inferred from neutral or positive comments. For example, “customer feedback is hard to turn into requirements” came from Kara praising AI summaries, and network tooling became its own product problem from a single LinkedIn mention. “Role requirement rather than a complaint” is the AI’s own judgment.
Long List
- P1–P2: Shirley’s interviews (Madhuri, Terry)
- P3–P4: Rita’s interviews (Tara, Tianrun)
- P5–P8: Alexis’ interviews (Kara, Dave, Tamara, Young)
| 1 | Meetings and coordination leave no time for focused work. Every team (customers, engineering, marketing) pulls the PM into meetings. | P1, P2, P7 | 3/8 | “Sometimes there’s like so many meetings throughout the day that there’s no time to do actual work.” (P2) |
| 2 | PMs spend most of their time rebuilding context before they can decide anything. Business, design, technical and stakeholder context is scattered, and requirements are often unclear. | P1, P2 | 2/8 | “Sometimes a metric changes not because of anything you did, but because another frontend or backend team launched a new feature that you didn’t know about.” (P3) |
| 3 | Decisions have to be made with incomplete information, balancing customer needs, feasibility and business priorities. | P1, P2, P3, P4 | 4/8 | “Assumptions that seem reasonable based on observation do not always translate into an effective product strategy.” (P4) |
| 4 | Getting people to act without having authority over them (“herding cats”). PMs have to persuade and negotiate, and teams’ incentives don’t always line up. | P2, P3, P4, P7 | 4/8 | “Most of PM is logistics and organization – ‘herding cats’ – getting humans to do things.” (P7, notes) |
| 5 | Assigned work silently drops. PMs are accountable for results, so they have to personally follow up, and unclear ownership often surfaces near the deadline. | P1, P2 | 2/8 | “Just because like someone is working on it doesn’t mean they’re going to like deliver on it without you intervening.” (P2) |
| 6 | AI output is hard to verify, especially for nontechnical PMs. This covers AI-generated PRDs, prototypes and code (with tech debt), and evaluating agents whose behavior varies from run to run. | P2, P4, P5, P7 | 4/8 | “Not technical enough to audit” AI-generated work. (P5, notes) |
| 7 | Hard to prioritize and cut scope. Long to-do lists hide the important work, it’s hard to know what to remove, and it’s hard to see why leadership prioritizes what it does. | P1, P2, P3, P4 | 4/8 | “You don’t need to add features just to make a product look ‘complete.'” (P3) |
| 8 | Operational work crowds out future and strategic thinking. Day-to-day firefighting takes over, so nobody plans several sprints ahead. | P1, P2 | 2/8 | “If somebody is not actively working on the future, nobody can actually think.” (P1) |
| 9 | Communication breaks down as companies grow. Information gets lost across org layers, and large-company processes slow shipping. | P1, P2 | 2/8 | “PMs all meetings pretty much … a little soul-crushing in bigger companies.” (P7, notes) |
| 10 | Engineers jump straight to technical solutions, and a blank canvas stalls decisions, so PMs have to frame a few options. | P1, P3 | 2/8 | “I’ve dealt enough engineers who are only solution oriented.” (P1) |
TAM
There are approximately 850,000 active product managers globally as reflected on LinkedIn, including roughly 200,000 in the U.S. Of that population, 85% of product managers are earlier-career PMs, APMs, Mid-PMs, and SPMs, making them the majority of the addressable population.
On the spending side, the broader project management software market is estimated at 8.7B – 11.27B globally in 2026, based on estimates from Grand View Research, The Business Research Company, and Mordor Intelligence. Those firms also forecast substantial continued growth, with estimates reaching $19.8B by 2033, $16.87B by 2030, and $23.09B by 2031, respectively.
The market demand is not simply for more project-management software. Organizations increasingly need “project scheduling, workflow automation, resource planning, cross-team reporting, and governance as their portfolios become more complex.” At the same time, tooling is becoming increasingly crowded: a global monday.com survey commissioned through Qualtrics found that 82% of respondents use work or project management software for organizational efficiency, while 57% reported an increase in the number of workplace tools they use. That creates three distinct sources of market demand in 2026: first-time adoption, replacement of inadequate tools, and consolidation of fragmented tool stacks.
Sources: CorpWaters, “How Many Product Managers Are There?”, APMIC, “2026 Project Management Software Market Report”
Add two or three sentences comparing your cold and round 2 interviews. What changed? Include one line: one idea from this week’s reading we used, and where.
In our cold interviews, we mostly recorded how PMs worked day-to-day: their schedules, tools, and general responsibilities. In round 2, our screener and shared question set helped us dig into the problem side of the role instead. We asked more systematically how the PM role changes with company and team size, how PMs work with engineering, design, sales, and leadership, and who actually has the final say on decisions, especially for AI products, where control over the pipeline can sit with either the PM or the engineering team. We also replaced many closed-ended questions with requests for specific examples, and asked follow-ups on each part of the workflow interviewees described. This led to concrete stories instead of general descriptions of the job: one PM walked us through a per-token pricing tradeoff where leadership and engineering disagreed, and another explained how promotion incentives at large companies discourage taking risks on good products. Finally, asking what skills PMs look for when hiring interns showed us how they themselves define a good PM, which pointed to the skills they find hardest to develop. We used Torres’ story-based interviewing in our round 2 questions, asking PMs to “tell me about the last time” they pushed back on a plan or juggled conflicting priorities, rather than asking what they usually do. Collecting specific past stories this way surfaced real tradeoffs and workarounds that general questions had missed in our cold interviews.
