Problem Diagnosis

Product isn't about owning the roadmap. It's about owning the clarity. And clarity starts with diagnosis: understanding what's actually wrong before deciding what to do about it.

Most teams don't have one problem. They have several overlapping symptoms that look different depending on who you ask. The PM sees a prioritization problem. Engineering sees a requirements problem. The stakeholder sees a velocity problem. They're all describing the same dysfunction from different angles.

The discipline is in diagnosis, not prescription. Listen for the signals, name the pattern, then match it to an intervention that fits.

Six patterns I see repeatedly

These aren't hypothetical. They come from real engagement feedback, retro patterns, usage analytics, and PM surveys across dozens of teams.

1. "I don't know where to start"

New team members face a wall of content, tools, and processes with no clear entry point.

Signals:

• "I opened the repo and couldn't figure out what to read first"
• Average time-to-first-useful-action is too long
• People defaulting to asking peers instead of using docs

What's actually happening: The system assumes context the new person doesn't have. Onboarding is implicit, not designed. Getting the right people in the room and asking the right questions unblocks this fast, but only if someone takes responsibility for the first-day experience.

Intervention: Build explicit entry points. Role-based sequences. A 4-step wizard (Role, Context, Tools, Comfort). Quick start guides that assume nothing.

2. "Everything looks the same priority"

Content is overwhelming and undifferentiated. No way to distinguish must-read from nice-to-have.

Signals:

• "There's so much here, I don't know what's required vs. optional"
• People reading tool guides before understanding practices
• Content consumed in the wrong order

What's actually happening: The team has information but no hierarchy. Everything is treated as equally important, which means nothing is important. This is a curation problem, not a content problem.

Intervention: Add tier labels (Required / Optional / Experimental). Create curated learning sequences. Organize by job-to-be-done, not by tool.

3. "How do I use AI in my daily work?"

People know AI exists but lack concrete patterns for integrating it into their workflow.

Signals:

• "I use ChatGPT sometimes but not systematically"
• Low adoption of agent workflows despite having tool access
• "I want to use AI but I don't know the right prompts"

What's actually happening: The team has tools but not patterns. They're experimenting randomly instead of following a progression from ad-hoc use to systematic integration. This isn't a training problem. It's a workflow design problem.

Intervention: Codify the progression. Start with "agent as drafter" (generate first drafts, human reviews). Move to "agent as reviewer" (human drafts, agent catches issues). Build shared prompt libraries. Make the invisible patterns visible.

4. "We keep re-discovering what we already decided"

No structure for planning and prioritization. Decisions don't stick.

Signals:

• "We keep re-discovering what we already decided"
• Planning artifacts scattered across tools
• Scope creep without clear backlog governance

What's actually happening: Decisions are made in conversations, not captured in artifacts. The team has meetings but no system. Every planning session starts from scratch because there's no durable record of prior commitments.

Intervention: Establish a two-layer planning system: a strategic plan (goals, constraints, non-goals) and an operational backlog (ordered, described, actionable). Keep both in one place. Reference them in every ceremony.

5. "We don't know if AI is actually helping"

No consistent way to track whether AI tools are improving outcomes.

Signals:

• "We gave everyone Copilot licenses but don't know if it helped"
• No baseline for AI maturity across teams
• "How do I show ROI on AI tooling?"

What's actually happening: The team adopted tools without defining success criteria. They can't measure progress because they never established a baseline. This is an evaluation discipline problem.

Intervention: Run a maturity assessment with behavioral indicators across multiple dimensions. Establish a baseline. Measure again quarterly. Track adoption patterns, not just license counts.

6. "Who reviews what the AI wrote?"

Quality and governance gaps. AI-generated content ships without consistent review.

Signals:

• "Who reviews what the AI wrote before it ships?"
• Inconsistent quality of AI-assisted artifacts
• "We need guardrails, not just tools"

What's actually happening: The team treats AI output as finished product instead of first draft. There's no review loop, no quality standard, and no shared understanding of when AI output is "good enough." This is a process gap, not a technology gap.

Intervention: Build review loops into the workflow. Define quality criteria for AI-assisted work. Create content tiers so people know what level of rigor each artifact requires.

The diagnosis framework

When something feels off, follow this sequence:

1. Collect signals from multiple sources. Don't trust one perspective. Talk to PMs, engineers, designers, and stakeholders. Look at retro patterns, support requests, analytics, and onboarding feedback.

2. Name the pattern. Which of the six patterns (or combination) are you seeing? Most teams have 2-3 active at once.

3. Identify the root, not the symptom. "Velocity is low" is a symptom. "Stories are unclear because the PM and engineer aren't pairing on refinement" is the root.

4. Match to an intervention that fits. Don't prescribe a heavy process for a lightweight problem. A 15-minute ceremony change might solve what a week-long workshop was designed for.

5. Check in two iterations. Did the intervention work? If not, the diagnosis might be wrong. Go back to step 1.

The listening discipline

Building trust with cross-functional teams starts by listening. By narrating your thinking. And by giving people context, not just direction. The best teams aren't led by control. They're aligned by clarity.

When you're diagnosing problems, resist the urge to solve immediately. The most useful thing you can do in the first conversation is help the team see that what felt like six different problems is actually two patterns showing up in different places. That reframing alone often unblocks action.