The Guerrilla PM Playbook

The short version

Most PMs work without a research team, dedicated data analyst, or full-time design support. The Guerrilla PM playbook runs world-class product management in 2 hours a week, four 30-minute blocks: Monday discovery call, Tuesday competitive scan (AI agent does 80%), Wednesday prototype session in Claude Code, Thursday data review (AI agent runs queries). The system compounds. By month three, you're operating at a level that PMs waiting for support functions never reach. Tools cost under $100/month. Scrappiness forces clarity. When you can't spend two weeks on customer research, you ask better questions. Block one hour next Monday and call a customer.

Most PMs work without a research team. Most work without a dedicated data analyst. Most don't have a design team that can work on their projects full-time. And yet, the expectation is that you'll do everything: discover what customers need, validate ideas, design solutions, analyze results, manage the roadmap, gather competitive intel, run product strategy.

The old playbook says you can't do world-class product work without world-class support. That's not true anymore. It's never been more false.

I run this playbook with minimal structure: 30 minutes of customer discovery, 30 minutes of competitive research, 30 minutes of prototyping, 30 minutes of metrics review. Two hours a week. That's it. No research team, no data analyst, no design support. Just 2 hours and the right tools.

And outcomes-wise? My solo-PM teams outperform teams with full research, design, and analytics support. Not because they're better at their jobs. Because scrappiness forces clarity. When you can't spend two weeks on customer research, you ask better questions. When you have to prototype quickly, you learn faster. When you have to analyze your own data, you actually pay attention to what it says.

The Reality: Most PMs Are Solo

Let's start with the uncomfortable truth. According to State of Product Management reports, fewer than 20% of PMs have access to a dedicated researcher. Fewer than 30% have a data analyst. Fewer than 50% have design support. And in startups, those numbers are closer to 5%, 10%, and 20%.

You're not the exception. You're the norm. You're a solo PM without the resources you think you need. And you've probably been waiting to have those resources to do your job "properly."

Here's the reframe: you don't need those resources. You need a system. And the system is simpler than you think.

The "2-Hour Weekly Investment" That Changes Everything

The system is this: every week, you invest 2 hours in four activities. Each activity takes 30 minutes. Each activity compounds over time.

Monday: 30 min Discovery Call

You talk to one customer (or a prospect, or a churned customer, or a support escalation). Unstructured. You ask them about their world, their problems, what they're trying to do. You take notes. You listen for patterns. That's it.

One call. 30 minutes. You don't need a research agency to tell you what customers care about. You need to actually talk to them.

Tuesday: 30 min Competitive Scan

An AI agent does 80% of the work here. You give it a simple prompt: "What did our top 5 competitors ship this week? What are they saying about their latest features?"

The agent pulls together the updates, launches, product moves, positioning changes. You skim it. Takes 30 minutes. You now know what competitors are doing. You don't spend three hours manually checking competitor websites.

Wednesday: 30 min Prototype Session

You don't have a designer. So you prototype yourself. Not in Figma. In Claude Code. In code. You build rough, interactive prototypes that let you test ideas quickly.

An idea comes in on Monday. Wednesday, you've built a prototype. Thursday, you test it with customers. Friday, you know if it works.

No designer needed. Faster than waiting for someone to wireframe something.

Thursday: 30 min Data Review

An AI agent runs the queries. You ask: "What happened to activation this week? Did our new onboarding change anything? How are the five segments performing?"

The agent builds a dashboard, pulls the metrics, writes a summary. You read it. You spot patterns. You make decisions based on data, not vibes.

30 minutes a week on metrics. Over a year, that's 26 hours of analytics work. Enough to catch real shifts and inform strategy.

How Each of These Compounds Over Time

The genius of this system isn't any single activity. It's that they compound.

By week four, you've done four customer calls. You're noticing patterns in what customers say. A theme emerges: customers are struggling with X. That becomes your focus for the month.

By week eight, you've run eight competitive scans. You know what your competitors are doing, what worked for them, what didn't. You're not surprised by anything anymore. You're ahead of moves.

By week twelve, you've built a dozen prototypes. You're fast at it now. Ideas that would take a designer two weeks to mock up, you prototype in an afternoon. You learn 10x faster.

By week twenty-six, you've analyzed metrics fifty times. You're reading them like a language. You spot anomalies. You understand what drives your metrics. You're not guessing anymore.

The compounding effect is real. By month three, you're operating at a level that most PMs never reach because they're waiting for support functions to exist.

The Tools That Make Solo PM Viable

This playbook only works if you have the right tools. Here's what you need:

For Discovery:

• Calendar tool (you already have this)
• Recording tool (Loom, Gong, or just native video)
• Note-taking tool that syncs with your docs

That's it. You don't need a research platform. You don't need Respondent or Userbit to schedule customers. You're scheduling them yourself, 30 minutes at a time.

For Competitive Research:

• An AI agent that can browse the web (Claude Code with web search)
• Your own competitive research dashboard (could be a simple Google Sheet)

You're not buying a competitive intelligence platform. You're training an AI agent to do the work for you.

For Prototyping:

• A code sandbox (Replit, CodeSandbox, or local environment)
• Claude Code for AI-assisted code generation
• Maybe Figma for one-off mockups

You're not waiting for a designer. You're building interactive prototypes yourself.

For Metrics:

• Your analytics platform (probably Mixpanel, Amplitude, or Segment)
• An AI agent that can query APIs
• A dashboard tool (Metabase is free and self-hosted)

You're not waiting for a data analyst to run queries. You're training an AI to do it.

The tools aren't expensive. Claude Code subscription. Maybe Metabase if your company isn't using it. That's it. Fewer than 100 a month in tools.

Why Guerrilla PMs Often Outperform PMs with Full Teams

This seems counterintuitive. A solo PM with 2 hours a week of investment should lose to a PM with a full research team, a designer, and an analyst.

They don't. Here's why:

Speed: A solo PM learns in days. A team with full support learns in weeks. By the time the research team finishes a qual study, the guerrilla PM has done five customer calls, spotted the pattern, prototyped a solution, and is testing it. Speed compounds.

Clarity: When you can't afford to be vague, you're forced to be clear. A guerrilla PM asks one core question and pursues it. A team with full support starts research projects that explore "customer needs in the onboarding journey" and end up with 200 pages of insights that don't inform decisions.

Bias toward testing: A solo PM doesn't have time for perfect research. So you prototype early, test quickly, learn by doing. You end up with empirical data instead of theoretical. Testing compounds.

Ownership: When you're doing the research and the prototyping and the analysis, you can't divorce yourself from the work. You stay curious. You care about the quality. A PM who outsources everything to specialists is one step removed from the reality.

I've seen this across companies. The scrappiest PMs make better decisions faster. Not always. But more often than not.

The 2-Hour Weekly Structure

Here's exactly how to run this:

Monday, 2 PM: Customer Discovery Call (30 min)

Block the time. Call a customer. No script. No survey. Just: "Tell me about what you're working on this week." Listen. Ask why. Take notes. Hang up.

If you don't have a list of customers to call, ask CS. Ask Sales. "Who's a power user I can talk to?" You'll find someone.

Tuesday, 10 AM: Competitive Scan (30 min)

Prompt an AI agent: "Check these five competitors. What have they launched in the past week? What are they saying about it?"

Read the results. 30 minutes. Done.

Wednesday, 3 PM: Prototype Session (30 min)

Pick one idea from this week. Spend 30 minutes building a prototype. Doesn't have to be pretty. Has to be interactive.

Use Claude Code to generate the structure. You fill in content. You test it.

Thursday, 11 AM: Metrics Review (30 min)

Prompt an AI agent: "Pull this week's metrics. Compare to last week and four weeks ago. What changed? What should I be concerned about?"

Read the dashboard. 30 minutes. You know what's happening with your product.

That's 2 hours. That's your week.

Common Concerns (And Why They Don't Matter)

"But I don't know how to code."

You don't need to. You're using Claude Code to generate code. You're editing it, not writing it from scratch. By week four, you'll be comfortable enough to iterate on simple prototypes.

"What if customers don't want to talk to me?"

Most do. Especially if you're building something for them. The hard part isn't getting them to talk. It's not interrupting to pitch. You're talking to understand, not to sell. Most customers appreciate that.

"Isn't this a lot of work for one person?"

2 hours a week is not a lot of work. It's what you do instead of status meetings. Instead of roadmap debates. Instead of Slack threads about whether to do feature X or feature Y. You spend that time on signal instead of noise.

"What if I can't find time?"

Then you're too busy with the wrong things. A PM who doesn't talk to customers, doesn't understand the market, doesn't validate ideas, and doesn't watch metrics is doing PM theater. Find the time. Block it. Protect it like it's an executive meeting.

Getting Started This Week

Don't wait. Don't plan a big launch of this system. Just start:

Monday: Call one customer. 30 minutes. Ask them what they're working on. Listen. Take notes.

Tuesday: Spend 30 minutes looking at what your top 3 competitors have shipped in the past month. Write it down.

Wednesday: Spend 30 minutes building a rough prototype of something you're curious about. Not perfect. Just interactive.

Thursday: Spend 30 minutes looking at your metrics from this week vs. last week. What changed?

That's it. One week. You've done all four activities. You'll be surprised by how much you learn.

Then do it again next week. And the week after.

By month three, this will feel normal. By month six, you'll be operating at a level that teams with full support never reach. By month twelve, you'll look back and realize that the constraint of working solo made you better at your job.

Not because you got better at product management. Because you got faster at learning.

And speed, in the end, is what matters.

Sources: Claude Code, Replit, Metabase, Loom.